Ambient and Indoor Particulate Matter 
Concentrations on the Mpumalanga 
Highveld 
 
 
B Wernecke 
orcid.org/0000-0001-5464-7158 
Previous qualification (not compulsory) 
 
 
Dissertation submitted in fulfilment of the requirements for the 
Masters degree in Geography and Environmental Management at 
the North-West University 
 
 
Supervisor:  Prof SJ Piketh 
Co-supervisor: Dr RP Burger 
 
 
Graduation May 2018 
24046868 
 
 
ii 
 
Abstract 
 
 
In order to better define tools for the protection of human health, the characterisation of particulate matter 
(PM) concentrations and the understanding of total exposure to these concentrations is critical. Many 
communities on the Mpumalanga Highveld in South Africa rely on coal for heating and cooking purposes. 
Consequently, individuals in such areas are often chronically and acutely exposed to elevated 
concentrations of PM, resulting in negative health impacts.  
 
An unprecedently rich data set was used to understand the magnitude as well as the spatial and temporal 
variability of PM concentrations in two low-income communities on the Mpumalanga Highveld, where 
solid fuel is the primary source of energy for heating and cooking purposes. Ambient, indoor and personal 
PM concentrations of different size fractions were simultaneously considered for analysis. Personal PM- 
and corresponding GPS measurements were used to define and contextualise personal exposure 
concentrations. Data was collected between 2013 and 2016 as part of Sasol and Eskom’s air quality offset 
pilot study sampling campaigns in KwaDela and KwaZamokuhle, respectively. 
 
Results showed that air quality in KwaDela and KwaZamokuhle is very poor, especially so in winter 
months. Particulate matter concentrations were often found to be higher indoors than in the ambient 
environment. Ambient, indoor and personal measured PM concentrations were highly variable in space 
and time, as influenced by larger- and local scale meteorological conditions as well as by socio-economic 
factors.  Peak personal exposure concentrations above the ambient air quality standards were not limited 
to a time of day nor to a specific micro-environment. However, high PM concentrations were most notable 
during peak burning times especially in and directly outside households. 
 
This study supports the notion that data logged at a centrally located ambient air quality monitoring station 
should be used with caution when making compliance related decisions and when conducting 
epidemiological studies in these areas. Whilst ambient air quality measurements are a useful guideline to 
use as a basis for air quality management interventions, it is important to include measurements taken in 
and around a household to ensure air quality management protocols are adequate to address the poor air 
iii 
 
quality that people breathe in low-income communities, and this particularly so on the Mpumalanga 
Highveld.  
 
 
Keywords: Ambient air quality, indoor air quality, low-income communities, variability of particulate 
matter concentrations, micro-environments, total personal exposure, inhalation dose 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
iv 
 
Table of Contents 
 
Abstract ............................................................................................................................... ii 
Table of Contents ............................................................................................................... iv 
List of Figures ................................................................................................................... vii 
List of Tables ...................................................................................................................... x 
List of Abbreviations ........................................................................................................ xii 
Acknowledgements .......................................................................................................... xiii 
Declaration ....................................................................................................................... xiv 
 Introduction ................................................................................................... 2 
1.1 Context and problem statement .................................................................................................... 2 
1.2 Aims and objectives ..................................................................................................................... 4 
1.3 Study Design ................................................................................................................................ 5 
1.4 Presentations and publications ..................................................................................................... 6 
 Literature Review .......................................................................................... 7 
2.1 Health effects of poor air quality .................................................................................................. 7 
2.2 Exposure to particulate matter concentrations ............................................................................. 9 
2.3 Fuel use and air pollution in low-income communities ............................................................. 13 
2.4 Particulate matter concentrations and variability in low-income communities ......................... 15 
2.4.1 Ambient and indoor particulate matter ............................................................................... 15 
2.4.2 Variability of particulate matter levels in a low-income context ........................................ 16 
2.5 Relationships between ambient, indoor and personal particulate matter concentrations ........... 18 
2.6 Relevance of this study .............................................................................................................. 19 
 Research Methodology ............................................................................... 21 
3.1 Study area ................................................................................................................................... 21 
3.1.1 Geographical location of the study sites ............................................................................. 21 
3.1.2 Meteorology in KwaDela and KwaZamokuhle .................................................................. 24 
3.2 Data collection ............................................................................................................................ 25 
3.2.1 Data collection as part of larger scale studies ..................................................................... 25 
3.2.2 Air quality monitoring in the ambient and the indoor environment ................................... 25 
3.2.3 Measuring personal exposure to particulate matter concentrations in various micro-
environments ..................................................................................................................................... 31 
3.3 Data analysis .............................................................................................................................. 32 
v 
 
3.3.1 Characterising particulate matter concentrations at a community level ............................. 32 
3.3.2 Characterising particulate matter concentrations at a household level ............................... 33 
3.3.3 Comparing ambient, indoor and personal particulate matter concentrations and 
understanding their relationships ...................................................................................................... 35 
3.3.4 Assessing personal exposure to particulate matter in different micro-environments ......... 37 
3.4 Ethical considerations ................................................................................................................ 38 
3.5 Limitations and assumptions ...................................................................................................... 38 
 Characterising Particulate Matter Concentrations ...................................... 42 
4.1 Spatial and temporal variability of particulate matter ................................................................ 42 
4.1.1 Ambient particulate matter concentrations ......................................................................... 42 
4.1.2 Spatial and seasonal trends of indoor and personal particulate matter concentrations within 
and between two communities .......................................................................................................... 49 
4.1.3 Indoor and personal hourly maximum particulate matter concentrations .......................... 53 
4.1.4 Diurnal patterns of ambient, indoor and personal particulate matter concentrations ......... 56 
4.1.5 Spatial variability of hourly maximum particulate matter concentrations per case per 
community ........................................................................................................................................ 61 
 Relationships between and personal exposure to particulate matter concentrations 65 
5.1 Household-level PM characteristics ........................................................................................... 65 
5.1.1 Characterising PM concentrations at a household level ..................................................... 67 
5.1.2 Temporal variation of particulate matter concentrations at a household level ................... 71 
5.2 Relationships between ambient, indoor and personal PM concentrations ................................. 75 
5.2.1 PM concentration relationships at a community level ........................................................ 76 
5.2.2 PM concentration relationships at a household level .......................................................... 78 
5.2.3 Ratios between indoor and personal PM measurements ..................................................... 80 
5.3 Personal total exposure to particulate matter concentrations in various micro-environments ... 82 
5.3.1 The identification of micro-environments in KwaDela ...................................................... 82 
5.3.2 Total personal exposure: A one-individual case study ....................................................... 83 
 Summary and conclusions .......................................................................... 88 
6.1 Characterising particulate matter concentrations In KwaDela and KwaZamokuhle ................. 89 
6.2 The relationships between ambient, indoor and personal PM concentrations in KwaDela and 
KwaZamokuhle ..................................................................................................................................... 90 
6.3 Personal exposure to particulate matter concentrations in KwaDela ......................................... 91 
6.4 Conclusion .................................................................................................................................. 92 
Project Funding ................................................................................................................. 94 
References ......................................................................................................................... 95 
vi 
 
 Annexures ................................................................................................. 109 
7.1 Annexure A –  Hourly maximum concentrations measured per case in KwaDela and 
KwaZamokuhle per campaign ............................................................................................................ 109 
7.2 Annexure B – Regression analyses between ambient, indoor and personal PM concentrations at 
a community level ............................................................................................................................... 117 
7.3 Annexure C – Overview of personal PM4 concentrations measured per micro-environment . 125 
7.4 Annexure D – Integrated exposure concentration and integrated potential dose calculation .. 127 
7.5 Annexure E – Daily average timeseries (after QA/QC) ........................................................... 131 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
vii 
 
List of Figures 
 
Figure 2.1.  Illustration of Inhalation Route: Exposure and Dose (U.S. EPA 1992) .......... 9 
Figure 3.1.  Map of study sites in relation to each other in the Mpumalanga Province, South Africa
 ............................................................................................................................... 22 
Figure 3.2.  Overview of study sites in relation to relevant suburbs and district municipalities 22 
Figure 3.3.  Overview of ambient and indoor monitoring sites located in KwaDela ....... 26 
Figure 3.4.  Overview of ambient and indoor monitoring sites located in KwaZamokuhle27 
Figure 3.5.  Examples of GPS coordinates plotted on a SPOT 6/7 image of KwaDela to help identify 
common micro-environments (The colours of the dots denote the movements of the various 
individuals wearing GPS monitors on different days. For the purpose of this study, the colours 
have no significance. This figure aims to illustrate movement patterns of individuals wearing the 
GPS monitors to assist in micro-environment identification and nothing more) . 32 
Figure 4.1.  Daily average ambient PM10 concentrations (µg/m³) in KwaDela and KwaZamokuhle across 
the seasons ............................................................................................................ 43 
Figure 4.2.  Daily average ambient PM2.5 concentrations (µg/m³) in KwaDela and KwaZamokuhle 
across the seasons ................................................................................................. 43 
Figure 4.3.  Daily average indoor PM4 concentrations (µg/m³) in KwaDela and KwaZamokuhle across 
the seasons ............................................................................................................ 49 
Figure 4.4.  Daily average personal PM4 concentrations (µg/m³) in KwaDela and KwaZamokuhle across 
the seasons ............................................................................................................ 50 
Figure 4.5.  Average hourly maximum indoor PM4 concentrations (µg/m³)  in KwaDela and 
KwaZamokuhle across the seasons ....................................................................... 55 
Figure 4.6.  Average hourly maximum personal PM4 concentrations (µg/m³) in KwaDela and 
KwaZamokuhle across the seasons ....................................................................... 55 
Figure 4.7.  Ambient particulate matter diurnal trends (µg/m³) per season in KwaDela . 58 
Figure 4.8.  Ambient particulate matter diurnal variation (µg/m³) per season in KwaZamokuhle 59 
Figure 4.9.  Indoor and personal articulate matter diurnal trends (µg/m³) per season in KwaDela 60 
Figure 4.10.  Indoor and personal particulate matter diurnal trends (µg/m³) per season in KwaZamokuhle
 ............................................................................................................................... 61 
Figure 5.1.  Comparing ambient, indoor and personal daily average PM concentrations (µg/m³) 68 
Figure 5.2.  Comparing ambient, indoor and personal daily average PM concentrations (µg/m³) 
measured per season for case 3KZ and case 9KZ in KwaZamokuhle ................. 69 
Figure 5.3 Diurnal variation of particulate matter (µg/m³) for houses 3KD and 11KD in KwaDela in 
Winter 2013 .......................................................................................................... 72 
Figure 5.4 Diurnal variation of particulate matter (µg/m³) for houses 3KD and 11KD in KwaDela in 
Summer 2014 ........................................................................................................ 73 
Figure 5.5 Diurnal variation of particulate matter (µg/m³) for houses 3KD and 11KD in KwaDela in 
Summer 2015 ........................................................................................................ 73 
Figure 5.6 Diurnal variation for houses 3KZ and 9KZ in KwaZamokuhle in Spring 201574 
Figure 5.7 Diurnal variation for houses 3KZ and 9KZ in KwaZamokuhle in Summer 2016 74 
Figure 5.8 Diurnal variation of particulate matter (µg/m³) for houses 3KZ and 9KZ in KwaZamokuhle in 
Winter 2016 .......................................................................................................... 75 
viii 
 
Figure 5.9.  Hourly average Personal PM4/ Indoor PM4 ratio in KwaDela (left) and KwaZamokuhle 
(right) .................................................................................................................... 81 
Figure 5.10.  Hourly average Personal PM4/ Ambient PM2.5 ratio in KwaDela (left) and KwaZamokuhle 
(right) .................................................................................................................... 82 
Figure 5.11.  Average percentage of time spent in each micro-environment per month for which GPS 
measurements were collected (July n = 8831, August n = 25 193, September n = 11 909) 83 
Figure 5.12.  GPS tracks showing average personal PM4 concentration exposure on the 25 August 2013. 
Micro-environments displayed are focused on “Inside a house” (delineated by an outlined 
rectangle) and “Directly outside a house” (the space outside the rectangles) ...... 84 
Figure 5.13.  Hourly average personal PM4 concentrations measured from 21-28 August 2013 85 
Figure 5.14.  Percentage of time spent in each micro-environment per day in question by the person 
wearing the GPS monitor between 21 and 28 August 2013 ................................. 85 
Figure 7.1.  Hourly maxima per PM type compared at household level KwaDela Winter 2013 110 
Figure 7.2.  Hourly maxima per PM type compared at household level KwaDela Summer 2014 111 
Figure 7.3.  Hourly maxima per PM type compared at household level KwaDela Winter 2014 112 
Figure 7.4.  Hourly maxima per PM type compared at household level KwaDela Summer 2015 113 
Figure 7.5.  Hourly maxima per PM type compared at household level KwaZamokuhle Spring 2015
 ............................................................................................................................. 114 
Figure 7.6.  Hourly maxima per PM type compared at household level KwaZamokuhle Summer 2016
 ............................................................................................................................. 115 
Figure 7.7.  Hourly maxima per PM type compared at household level KwaZamokuhle Winter 2016
 ............................................................................................................................. 116 
Figure 7.8.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of that 
same regression analysis when plotting overarching, community-level daily average particulate 
matter concentrations in KwaDela ...................................................................... 117 
Figure 7.9.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of that 
same regression analysis when plotting overarching, community-level daily maximum 
particulate matter concentrations in KwaDela .................................................... 118 
Figure 7.10.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of 
that same regression analysis when plotting overarching, community-level hourly average 
particulate matter concentrations in KwaDela .................................................... 119 
Figure 7.11.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of 
that same regression analysis when plotting overarching, community-level hourly maximum 
particulate matter concentrations in KwaDela .................................................... 120 
Figure 7.12.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of 
that same regression analysis when plotting overarching, community-level daily average 
particulate matter concentrations in KwaZamokuhle ......................................... 121 
Figure 7.13.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of 
that same regression analysis when plotting overarching, community-level daily maximum 
particulate matter concentrations in KwaZamokuhle ......................................... 122 
Figure 7.14.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of 
that same regression analysis when plotting overarching, community-level hourly average 
particulate matter concentrations in KwaZamokuhle ......................................... 123 
Figure 7.15.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of 
that same regression analysis when plotting overarching, community-level hourly maximum 
particulate matter concentrations in KwaZamokuhle ......................................... 124 
ix 
 
Figure 7.16 Timeseries of daily average PM concentrations (µg/m3) in KwaDela across the campaigns
 ............................................................................................................................. 131 
Figure 7.17 Timeseries of daily average PM concentrations (µg/m3) in KwaZamokuhle across the 
campaigns ........................................................................................................... 131 
Figure 7.18 Timeseries of ambient daily average PM concentrations (µg/m3) in KwaDela in winter 2013
 ............................................................................................................................. 132 
Figure 7.19  Timeseries of ambient daily average PM concentrations (µg/m3) in KwaDela in summer 
2014 ..................................................................................................................... 132 
Figure 7.20  Timeseries of ambient daily average PM concentrations (µg/m3) in KwaDela in winter 2014
 ............................................................................................................................. 133 
Figure 7.21  Timeseries of ambient daily average PM concentrations (µg/m3) in KwaDela in summer 
2015 ..................................................................................................................... 133 
Figure 7.22   Timeseries of ambient daily average PM concentrations (µg/m3) in KwaZamokuhle in 
spring 2015 .......................................................................................................... 134 
Figure 7.23   Timeseries of ambient daily average PM concentrations (µg/m3) in KwaZamokuhle in 
summer 2016 ....................................................................................................... 134 
Figure 7.24   Timeseries of ambient daily average PM concentrations (µg/m3) in KwaZamokuhle in 
winter 2016 ......................................................................................................... 135 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
x 
 
List of Tables 
 
Table 1.1.   South African National Ambient Air Quality Standards for Particulate Matter (South Africa, 
2009; South Africa, 2012b) ..................................................................................... 3 
Table 3.1.   Relevant community statistics for KwaDela and KwaZamokuhle taken from the Census 
2011 - an overview (Stats SA, 2012b) .................................................................. 23 
Table 3.2.  Descriptive statistics of 1-hour resolution meteorological data for each of the campaigns in 
KwaDela ............................................................................................................... 24 
Table 3.3.  Descriptive statistics of 1-hour resolution meteorological data for each of the campaigns in 
KwaZamohule ....................................................................................................... 25 
Table 3.4.  E-Sampler and E-Bam codes for in-text interpretation and orientation according to cardinal 
directions ............................................................................................................... 27 
Table 3.5.  Overview of ambient data collection in KwaDela and KwaZamokhule ........ 29 
Table 3.6.  Overview of indoor and personal data collection in KwaDela and KwaZamokhule 30 
Table 3.7.  Overview of cases used for comparative analyses in KwaDela ..................... 34 
Table 3.8.  Overview of cases used for comparative analyses in KwaZamokuhle ........... 34 
Table 3.9.  Overview of cases chosen for household-level analyses ................................ 35 
Table 4.1.  Seasonal characteristics (daily mean±S.D.) of ambient PM10 concentrations (µg/m³) in 
KwaDela and KwaZamokuhle .............................................................................. 44 
Table 4.2.  Seasonal characteristics (daily mean±S.D.) of ambient PM2.5 concentrations (µg/m³) in 
KwaDela and KwaZamokuhle .............................................................................. 44 
Table 4.3.  Number of NAAQS limit value exceedances per season per community for PM10 (total 
number of daily averages) ..................................................................................... 45 
Table 4.4.  Number of NAAQS limit value exceedances per season per community for PM2.5 (total 
number of daily averages) ..................................................................................... 45 
Table 4.5.  Seasonal characteristics (daily mean±S.D.) of indoor and personal PM concentrations 
(µg/m³) in KwaDela and KwaZamokuhle ............................................................ 50 
Table 4.6.  Seasonal characteristics (hourly max±S.D.) of indoor and personal PM concentrations 
(µg/m³) in KwaDela and KwaZamokuhle ............................................................ 56 
Table 5.1.  Overview of relevant survey results for chosen households in KwaDela ...... 66 
Table 5.2.  Overview of relevant survey results for chosen households in KwaZamokuhle66 
Table 5.3.  Overview of singled out highest coefficients of determination (R2) resulting from community 
level regression analyses in KwaDela ................................................................... 77 
Table 5.4.  Overview of singled out highest coefficients of determination (R2) resulting from community 
level regression analyses in KwaZamokuhle ........................................................ 77 
Table 5.5.  Correlation coefficients of determination for hourly average PM types per house, per season 
in KwaDela ........................................................................................................... 79 
Table 5.6.  Correlation coefficients of determination for hourly average PM types per house, per season 
(KwaZamokuhle) .................................................................................................. 79 
Table 5.7. Directly and indirectly derived potential doses (mg) over a specified time period 86 
Table 7.1.  Overview of personal PM4 concentrations measured per micro-environment (M-E) identified 
between 21 August 2013 and 28 August 2013 (I=Inside a house; DO=Directly outside a house; 
DR=On a dirt road; TR=On a tar road; OF=On an open field) .......................... 125 
xi 
 
Table 7.2. Derivation of time-weighted, integrated exposure concentrations and potential doses for an 
individual carrying a SidePak and GPS monitor between 21 and 28 August 2013 (U.S. EPA, 
2011) ................................................................................................................... 127 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
xii 
 
List of Abbreviations 
 
AMS Ambient Monitoring Station 
Case A case is a data set at an hourly average resolution that has been compiled to 
allow the direct comparison of ambient, indoor and personal PM concentrations 
as well as relevant meteorological data for a given time. A case typically pairs 
ambient and meteorological data that has been collected at the same time as 
indoor/-personal measurements were conducted for a specific house. As such 
for instance, one can compare the hourly average indoor PM4 concentrations 
measured in a given hour on a given day with ambient PM2.5 concentrations 
measured in that same hour on that same given day. This is useful for 
comparative purposes and forms the basis of analyses done in this thesis. 
 
 
 
 
 
 
DALY Disability adjusted life years 
NAAQS National Ambient Air Quality Standards 
PM Particulate Matter 
PM10 Particulate Matter; inhalable particles, with a diameter that is generally 10 
micrometres and smaller 
 
PM2.5 Particulate Matter; inhalable particles, with a diameter that is generally 2.5 
micrometres and smaller 
PM4 Particulate Matter; inhalable particles, with a diameter that is generally 4 
micrometres and smaller 
Sampling campaign Measurements taken in a specific community for a specific season. Four 
sampling campaigns took place in KwaDela (winter 2013, summer 2014, winter 
2014 and summer 2015). Three Sampling campaigns took place in 
KwaZamokuhle (spring 2015, summer 2016 and winter 2016). 
 
µg microgram 
 
 
 
xiii 
 
Acknowledgements 
 
 
The final destination of this journey would not have been reached without the guidance, assistance, 
encouragement and support provided by many individuals, who carried me through many a rough and 
tough patch. 
 
I would like to thank my supervisors for believing in me: Prof Stuart Piketh, thank you for being patient 
with this submission. Also, thank you for letting me come along to one of the unforgettable KwaDela 
sampling campaigns, which allowed me to experience first-hand how some of the data I worked with was 
collected. Dr Roelof Burger, thank you for your time:  You truly were available at all times, especially 
when I visited campus. You provided me with most valuable tips on how to conquer this mountain, and I 
will be forever grateful for the moral support. Thank you for helping me find parking on many occasions. 
Dr Kristy Langerman, thank you for helping me when I was unable to see the wood for the trees! Your 
mentorship and guidance were priceless in this process and I am grateful that I learn from you every single 
day. To a few other individuals, whose names cannot be omitted from my acknowledgements: 
 
- Alex Howard, Michael Perrie, Ryan Daniels and Christiaan Pauw for your relentless, but in the end, 
unfortunately futile efforts of trying to bring R closer to my heart. Alex, you were a rock of support 
- Riandi Venter and Michael Swaine for providing me with GIS shapefiles for my maps 
- Angelika Ronge, Margaux Giannaros for your moral support and understanding, having gone through this 
process yourselves  
- Chantal Taylor and Neil Wilson for your objective proof-reading, tips and time 
- Brigitte Language, Farina Lindeque, Ncobile Nkosi and Monray Belelie as fellow thesis submitters and 
NACA presenters 
 
Lastly, and most importantly, I would like to extend my gratitude to my family for being there for me 
through what can only be defined as thick and thin. What would I have done without you, Marianne, 
Andreas and Naomi Wernecke:  You carried me through what has been one of my most difficult, but most 
rewarding journeys. Zusammen schaffen wir alles! 
 
This research would not have been possible without generous funding from North West University and 
data provided by Sasol and Eskom as part of their respective air quality offset pilot studies. Thank you to 
NOVA and NWU for allowing me to use the fantastic datasets collected.  
 
 
 
 
xiv 
 
Declaration 
 
This thesis is the result of the author’s original work except where acknowledged or specifically stated in 
the text. It is being submitted for the degree of Master of Science at North West University, Potchefstroom. 
This thesis has not been submitted before for any degree or examination in any other degree.  
 
 
______________________ 
B Wernecke 
 
 
 
20 day of November 2017 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  
 
 
 
 
 
 
 
 
 
 
Everyone has the right— 
(a)  to an environment that is not harmful to their health or wellbeing; and 
(b)  to have the environment protected, for the benefit of present and future 
generations 
 
The Constitution of the Republic of South Africa 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2 
 
 Introduction 
 
The literary and research context, into which this study falls, is outlined and 
a clear problem statement is defined. The study design is touched upon and 
ultimate aims and objectives of this work are introduced.  
 
1.1 Context and problem statement  
 
 
It is stipulated in Section 24 of The Constitution of the Republic of South Africa (1996), that everyone 
has the right to clean air that is not harmful to people’s health and well-being. The air quality in many 
dense low-income communities in South Africa challenges this constitutional right. As has been 
emphasised in numerous health studies, as well as recently confirmed and published in the national “Draft 
Strategy to Address Air Pollution in Dense Low-Income Settlements”, measured ambient pollution 
concentrations have shown that, within low-income communities specifically, National Ambient Air 
Quality Standards (NAAQS) are regularly being exceeded (South Africa, 2016). This poses a threat to 
human health and to the environment in those areas (Engelbrecht et al., 2000; Norman et al., 2007; Lim 
et al., 2012).   
In order to design and support suitable implementation strategies to prevent and reduce health risks 
associated with air pollution, reliable estimates of human exposure to inhalable air pollutants are 
necessary.  Similarly, to be able to support health impact assessments and effective air quality 
management, it is also essential to develop a better understanding of individual exposure pathways in 
people's everyday lives. 
Domestic burning generates a range of hazardous outdoor and indoor air pollutants. Exposure to these 
emissions represents a prominent cause of morbidity and mortality in developing countries, attributing to 
more than four (4) million premature deaths globally (Lim et al., 2012; Fullerton et al., 2008; Rosenthal 
et al., 2017). People most affected, are those living in low-income communities who conduct solid fuel 
combustion activities to meet their primary domestic energy requirements. These communities are trapped 
3 
 
at the bottom of the energy ladder and make use of dirty fuels for cooking and heating, particularly in 
winter months (Kroon et al., 2011; Yangyang et al., 2015).  
 
In South Africa, the management of criteria pollutants, which are considered to be detrimental to the health 
of the public and to the environment, is based primarily on set NAAQS (Table 1.1)  (South Africa, 2009; 
South Africa, 2012b). Ambient air quality standards set thresholds for health-harmful pollution levels 
(WHO, 2016). Particulate matter (PM), specifically PM2.5, is an important criteria pollutant for which no 
low-concentration threshold exists that does not cause a discernible health effect (Jantunen et al., 1998). 
PM2.5 concentration levels have been found to exceed ambient air quality standards in low-income 
communities in South Africa (Hersey et al., 2015).  
 
Table 1.1.   South African National Ambient Air Quality Standards for Particulate Matter (South Africa, 
2009; South Africa, 2012b) 
PM Type Averaging Period Concentration (µg/m³) 
Allowed Frequency of 
Exceedance 
PM10 
24 hours 75 4 
1 year 40 0 
PM2.5 
24 hours 40 4 
1 year 20 0 
 
 
Human exposure to suspended particulates in  many South African low-income communities, stemming 
most notably from domestic burning practices, has been identified to be characterised by air quality which 
exceeds ambient standards by three (3) to four (4) times or more (Mdluli, 2007; Hersey et al., 2015; 
Garland et al., 2017). Exposure to such high concentrations of  air pollution has been shown to increase 
the risk of common and serious diseases such as respiratory and cardiovascular illnesses (Scorgie & 
Thomas, 2006; Norman  et al., 2007; Lim et al., 2012).  
 
Many will argue that a compliance status with the NAAQS represents a poor proxy upon which to draw 
conclusions for decision making purposes (Oglesby et al., 2000; Adgate et al., 2002; Wilson, 2006). 
Exposure studies have shown that it is immensely complex to ascertain and understand the air that people 
breathe and are exposed to, because total individual exposure varies greatly in space and time and is 
difficult to define (Adgate et al., 2002).  Compliance with the NAAQS does not take into account the air 
quality in different micro-environments, such as for instance the indoor environment, where many people 
4 
 
spend the majority of their time (Wang et al., 2015). Ambient air quality readings are widely used as a 
surrogate value to determine whether or not the air quality in a given area at a given time is safe to breathe 
(Oglesby et al., 2000; Payne-Sturges et al., 2004).  Though an important indicator of the general state of 
air, it is queried whether such centrally logged measurements are an adequate means to draw health-related 
conclusions (Oglesby et al., 2000).    
 
1.2 Aims and objectives 
 
 
This study aims to characterise PM concentrations in the ambient and the indoor environment within two 
low-income communities on the Mpumalanga Highveld, South Africa, namely, KwaDela and 
KwaZamokuhle.  Coal burning activities are particularly common here, as coal mines are an almost 
ubiquitous source of cheap fuel used for cooking and heating activities (Stats SA, 2012b). This, along with 
the fact that the creation of indoor fires is a culturally entrenched phenomenon, has led to the assumption 
that the air quality in these communities, much like other low-income communities in the province, is 
characterised by concentrations that regularly exceed ambient standards (Balmer, 2007; Mdluli, 2007; 
Garland et al., 2017). The specific research objectives of this study are to: 
1. Characterise overall ambient, indoor and personal PM concentrations in KwaDela and 
KwaZamokuhle, Mpumalanga; 
2. Determine the relationships between ambient, indoor and personal PM concentrations in KwaDela 
and KwaZamokuhle, Mpumalanga; and 
3. Explore personal exposure to PM in KwaDela. 
 
The results of this study ultimately help us to better understand the magnitude of ambient, indoor and 
personal PM concentrations that are prevalent in both settlements and how they vary in space and time. In 
particular, this study helps answer the question of whether what is measured in the ambient environment 
by a stationary monitoring site is an adequate proxy to use to represent the air that is breathed by the 
people in these communities.  
5 
 
1.3 Study Design 
 
 
The bulk of the findings of this study have been based on conclusions stemming from the analysis of 
empirical data gathered by means of physical measurements in various locations in two low-income 
communities on the Mpumalanga Highveld: KwaDela and KwaZamokuhle. These findings are, where 
possible, supported, re-enforced and contextualised by survey data results, as collected in these same two 
communities. A summary of ambient, indoor and personal PM concentrations is presented to introduce 
the reader to what has been measured in a low-income community setting throughout the seasons of the 
year. These measurements are compared to existing NAAQS to draw compliance-related conclusions.  
Factors governing the variability of PM concentrations are considered and personal exposure to these 
concentrations in various micro-environments is assessed by the use of zoomed-in case-studies. Finally, 
statements are made drawing inferences from the results which indicate what air pollution levels in these 
communities are like and whether or not ambient air quality measurements in KwaDela and 
KwaZamokuhle could be used as surrogate values for personal exposure assessments. To cover this 
subject matter, this thesis is divided into the following chapters: 
 
Chapter 1 – This introductory chapter sets the context of the research study. It also outlines the aims and 
objectives and the study design. 
Chapter 2 – A literature review is presented in this chapter of the thesis. It outlines where this study fits 
into the bigger picture of research conducted in this field.  
Chapter 3 – The study area is introduced in this chapter. Furthermore, the methodology used for data 
collection and data analysis is set out. 
Chapter 4 – This chapter characterises PM concentrations within and between the low-income 
communities considered in this study. Spatial and temporal variability of the data is defined. 
Chapter 5 – To better understand how societal parameters influence PM exposure at a household level, a 
two-household case-study is introduced. Traits characterising community-level PM concentrations are 
demonstrated at a household level.  The relationship between ambient, indoor and personal PM 
concentrations is discussed. Finally, personal PM measurements are corresponded with personal GPS 
coordinates and personal exposure to PM is assessed. 
Chapter 6 – Conclusions and main findings are outlined in the final chapter of this dissertation. 
 
6 
 
1.4 Presentations and publications  
 
Selected results presented in this study have been captured and presented at the National Association for 
Clean Air Conference in 2015 and 2016, respectively.  A paper titled “Indoor and outdoor particulate 
matter concentrations on the Mpumalanga Highveld - A case study” was published in the Clean Air 
Journal,  25(2):12-16 in 2015.  
 
 
********************* 
 
The aims and objectives of this study were outlined and placed into context. 
The next chapter represents a review of relevant literature.  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7 
 
 Literature Review 
 
The research conducted in this study is contextualised by placing it into 
current and existing national and international literature that covers the 
subject matter at hand: poor air quality in poverty-stricken areas and the 
resulting negative health implications.  Health impacts related directly to 
exposure to high PM concentrations are outlined. Thereafter, the topic of 
poor air quality in the South African context, particularly so in low-income 
communities on the Mpumalanga Highveld, is considered. The fact that PM 
concentrations in a low-income community setting are highly variable in 
space and time is touched upon to lead to an explanation of how high 
variability of PM concentrations makes it difficult to define total personal 
exposure. After variability of ambient, indoor and personal PM 
concentrations is discussed at length, the complex relationships between 
these concentrations is explored. Finally, the relevance of this study is 
explained.  
 
2.1 Health effects of poor air quality  
 
 
In 2015, over 90% of the world’s population lived in areas governed by “unhealthy air” (Health Effects 
Institute, 2017). There is no shortage of published work that delves into the negative health effects of air 
pollution, which can be described as a “multifaceted mix” made up of a combination of airborne particles 
and gases (Brunekreef & Holgate 2002; Cohen et al., 2015; Morakinyo et al., 2017). Exposure to 
suspended pollutants has been associated with increased disease, resulting shortened lives and death 
(Brunekreef & Holgate 2002; Lim et al., 2012).  In particular, a significant correlation has been found to 
exist between exposure to PM air pollution, morbidity and mortality (Iwai et al., 2005; Krall et al., 2013). 
Even short-term exposure to PM contributes to acute cardiovascular and pulmonary health problems, and 
exposure to elevated PM levels over the long-term can reduce life expectancy  (Brook et al., 2010; Krall 
et al., 2013).  Data on daily mortality has indicated that, globally, 4-8% of premature deaths may occur 
due to the exposure to total suspended particles (TSP), especially fine particles (PM2.5) in the ambient and 
indoor environment (Massey et al., 2009).  
 
8 
 
Exposure to air pollution from the combustion of solid fuels has been found to be a contributory cause of 
several diseases in developing countries. These include acute respiratory infections and otitis media 
(middle ear infection), chronic pulmonary disease, lung cancer (especially from coal smoke), asthma, 
nasopharyngeal and laryngeal cancer, tuberculosis, prenatal conditions and low birth weight (as a result 
of maternal exposure), and diseases of the eye such as cataract and blindness (Ezzati & Kammen, 2002; 
Lozano et al., 2007).  
 
A study conducted in the rural villages of the Himalayas for instance,  took into account short- and long-
term exposure events and estimated the burden of disease for acute lower respiratory infection, chronic 
obstructive pulmonary disease and lung cancer using World Health Organisation guidelines for rural 
households, using fuel-wood for cooking (Pandey, 2012). It was found that, households that used fuel-
wood for energy intensive activities, had disability adjusted life years (DALYs) lost and the number of 
deaths in these areas were found to be higher than the national average.  DALYs represent a measure of 
the “loss of healthy life expectancy and are calculated as the sum of the years of life lost from a premature 
death and the years lived with disability (for example, blindness, caused by the disease diabetes)” (Health 
Effects Institute, 2017). 
 
Different population groups are exposed to different PM concentration levels, due to their various 
lifestyles (Mehta & Shahpar, 2004). Marginalised socio-economic and demographic groups (women, 
young children and the elderly, for instance), are more at risk of being exposed to emissions stemming 
from domestic burning activities than working adults (mostly men), who spend most of their day outdoors 
and in their respective working environments (Ezzati & Kammen, 2002).  In Nigeria, blood tests were 
conducted on 59 mother-child pairs that were exposed to PM emissions stemming from solid-fuel burning 
practices (Oluwole et al., 2013).  Results indicated detectable adverse effects in mothers and children who 
were exposed to household air pollution (Oluwole et al., 2013).  
 
In a recent global burden of disease study, where burden of disease refers to the assessment of mortality, 
morbidity, injuries, disabilities and other risk factors specific to that country, indoor air pollution arising 
from the burning of coal and wood for domestic cooking and heating was found to be the seventh (7th) 
greatest disease risk in southern Africa, while ambient air pollution represented the 25th highest disease 
9 
 
risk (Lim et al., 2012). Estimates of the burden of disease are crucial for targeting health interventions that 
make a significant impact on the well-being of the population (Bradshaw et al., 2003). 
 
2.2 Exposure to particulate matter concentrations 
 
 
To understand the negative health effects of coming into contact with high PM concentrations, the notion 
of “exposure” to air pollution has been widely discussed in literature (Jantunen, 2007; Lozano et al., 2007; 
Fullerton et al., 2008; Chowdhury et al., 2013; Lin et al., 2013; Umoh et al., 2013; Thomas et al., 2015). 
The concept of exposure is defined differently across studies, but general consensus expresses it as “human 
contact with a chemical agent or a pollutant at a visible external boundary (e.g. the skin), where the 
chemical concentration at the point of contact is the exposure concentration” (U.S. EPA 1992). Total 
exposure, which takes into account the period of time an individual comes into contact with an exposure 
concentration, is derived by using the equation:   
 
  E = ∫ 𝑪(𝒕)𝒅𝒕
𝒕𝟐
𝒕𝟏
             (1) 
 
Where E is the magnitude of exposure, C(t) is the exposure concentration as a function of time and t is 
time, t2 – t1, representing the exposure duration (ED) (U.S. EPA 1992).  Once the magnitude of exposure 
has been determined, an exposure assessment can qualitatively or quantitively evaluate the contact with 
the exposure concentration. In the case of exposure to air pollution, this means that an exposure assessment 
can describe the amount of a pollutant that is inhaled into the body of an individual, i.e. the amount of a 
pollutant that actually crosses the external boundary, all of which is not absorbed by the body (potential 
dose) (Figure 2.1). 
 
Figure 2.1.  Illustration of Inhalation Route: Exposure and Dose (U.S. EPA 1992)  
10 
 
The general equation for potential dose for inhalation is the integration of the chemical intake rate 
(concentration of the pollutant in the air times the intake rate of the air, C times IR) over time (U.S. 
EPA1992): 
 
                   Dpot = ∫ 𝑪(𝒕) 𝑰𝑹(𝒕)𝒅𝒕
𝒕𝟐
𝒕𝟏
                            (2) 
 
Where Dpot is potential dose and IR(t) is the inhalation rate of an individual. The quantity t2 - t1, as before, 
represents the period of time over which exposure is being examined (U.S. EPA1992).  
 
The above methods describe how the intake of an exposure concentration is assessed. More detailed 
exposure assessments work towards defining the uptake of a contaminant or chemical and look at deriving 
the applied dose, the internal dose and finally the biologically effective dose of a contaminant (Figure 2.1). 
The determination of each of these dose types determines, with increasing level of complexity, how much 
an exposure concentration ultimately is taken up into the body and how it interacts with and affects organs 
and tissue within the body of an individual (U.S. EPA, 1992). For the purpose of this study, these dose 
uptake measures will not be elaborated upon, as they are not the focus of the research.   
 
Generally, it can be said that the amount of a pollutant that enters through the upper respiratory tract of an 
individual and then into the lungs, is in actual fact less than what is initially breathed into the body (the 
potential dose) (U.S. EPA, 1992). Individuals at different ages and of different genders have different 
breathing rates, which determine how much of a specific pollutant is inhaled into the respiratory system 
and how deeply (Smith, 1993; Wilson et al., 2000). Physiology, size, activity level and age are only a few 
of the factors that influence the breathing rates of an individual (Wilson et al., 2000). The U.S. EPA has 
published recommended long- and short term breathing rates for adults and for children, based on the 
findings of relevant studies (U.S. EPA, 2011).  Accordingly, long-term breathing rates for children for 
instance, range from 3.5 m3/day (for ages three (3) months – one (1) year) to 16.3 m3/day (for ages 16 - 
21 years). Average breathing rates for adults (men and women combined) range from 12.2 m3/day (for 
ages 81 years and older) to 16.0 m3/day (for ages 31 to 51 years) (U.S. EPA, 2011). Short-term breathing 
rates are useful when activity patterns are known and are typically expressed in volume per minute (U.S. 
EPA, 2011). The more active an individual, the faster his/her breathing rate and the larger the volume of 
air that is inhaled (Chowdhury et al., 2012). 
11 
 
A study conducted in Bangladesh to quantify indoor air pollution stemming from the use of domestic 
cookstoves, estimated the health effects of indoor air pollution on adult women (Chowdhury et al., 2012). 
The inhaled dose/ potential dose of PM2.5  exposure was ascertained by using the above equation (2). The 
inhalation rate used was 18 m3/day and the exposure concentrations used represented average indoor PM2.5 
concentrations. Potential inhalation doses ranged between 4.4mg and 5.8mg of PM2.5 per day (Chowdhury 
et al., 2012). 
 
It is possible to derive indirect and direct exposure concentration estimates to air pollution: Indirect 
estimation methods include the use of questionnaires/ time-activity diaries and measurements taken at 
stationary monitoring sites to ascertain lifestyle and household characteristics pertinent to relative 
exposure to ambient as well as indoor sources of PM. This includes the estimation of time spent in specific 
micro-environments and corresponding this with the average PM concentrations measured in those 
environments (Jones et al., 2000; Mdluli, 2007).  Direct methods, on the other hand, consider scientific 
measurements which include the concept of inhalation-based exposure assessments by using personal 
monitors (Abbey et al., 1999). Holistic total exposure estimates would take into account a combination of 
both indirect and direct measurements alongside detailed personal activity patterns.  
 
Although indirect exposure assessments may imply exposure levels by means of causation, direct personal 
concentration exposure measurements are considered the most accurate approximation method to use to 
understand total and true exposure for numerous air pollutants as indirect exposure assessments can often 
mask the complexities of total exposure to PM concentrations (Ezzati & Kammen, 2001; Wichmann, 
2006). Total exposure, regardless of whether it is defined directly or indirectly, is a function of time spent 
exposed to various indoor and ambient pollution sources, may these be natural or anthropogenic in nature. 
 
In a South African study conducted in KwaGuqa, Mpumalanga, concentration exposure to PM4 was 
estimated both directly and indirectly (Mdluli, 2007).  In the direct approach, exposure concentrations 
were determined on individuals by using a personal sampler, in the indirect approach, concentration 
exposure levels were measured by stationary DustTrak aerosol monitors.  Population exposure to a 
specific pollutant fraction type was ascertained: Daily concentrations of PM4 as well as the estimated 
amount of time that people spent inside their houses were used to assign population exposure to PM4. 
Time-budget and time-activity patterns received from surveys were used to support the monitored data. It 
12 
 
was found that, for the indirect exposure assessments, exposure to the highest PM4 concentrations was 
always found to be in the indoor environment. The direct exposure assessment however, showed that 
ambient pollution sources also contributed greatly to the total exposure of an individual. 
 
Limitations of most pollution exposure and epidemiology studies include (i) the short time frames of most 
of the research, which do not account for long-term exposure and its health effects; (ii) obtaining reliable 
data from personal monitors (individuals can easily disrupt the measurement); (iii) the high costs and 
labour intensive nature of the personal measurements and finally (iv) obtaining consent from individuals 
to participate in the study, as this is an invasion of privacy (Brunekreef & Holgate, 2002; Wichmann, 
2006).   
 
Ambient and indoor air are a potential source of exposure to toxic airborne substances (U.S. EPA, 2011). 
Ambient PM sources in the low-income community context specifically, include emissions from motor 
vehicles and untarred roads, wind-blown dust, industries burning dirty fossil fuels, waste burning activities 
and domestic use of highly polluting fuel types (Mdluli, 2007; Norman et al., 2007).  Indoor PM 
concentrations are defined by a similarly wide range of sources, such as sand, clay, soil, smoke residuals 
from heating, cooking or smoking, cleaning agents, residues from synthetic fibres, building materials and 
a multitude of other materials created in the home or transported in from the outside either manually or 
through infiltration/ penetration (Thatcher & Layton, 1995).  Studies have shown that coal burning 
activities related to cooking and heating activities represents one of the largest source contributors to 
ambient and indoor PM concentrations (Engelbrecht et al., 2002).  
 
Additionally, time of day and seasonal changes define the exposure of a person to PM in any micro-
environment, because temporal and seasonal factors influence the time-activity patterns of an individual 
and thus determine where a person spends time (Jones et al., 2000). Having noted this, it makes sense to 
say that total human exposure is characterised by multiple factors and that it is a concept that is not at all 
straight-forward to define.    
 
 
13 
 
2.3 Fuel use and air pollution in low-income communities  
 
At a global level, people living in poverty stricken and low-income areas are particularly affected by high 
ambient and indoor air pollution levels (Health Effects Institute, 2017). This is mainly because they rely 
primarily on dirty, solid fuels for their energy needs, which are substantially more polluting than liquid or 
gaseous fuels (Smith, 2000).    
 
In China, for example, poor air quality has increasingly become a concern, especially in the developing 
urban centres. It has been estimated that indoor air pollution from solid fuel use is responsible for 
approximately 420 000 premature deaths every year (Chen et al., 2011; Lin et al., 2013; Zhao et al., 2015; 
Xu et al., 2016). More than 60% of China’s population lives in a rural context, where solid fuels, like coal 
or wood, represent the main source of energy for cooking and heating activities, and so, health effects due 
to air pollution stemming from these activities in such areas is a major concern (Zhang & Smith, 2007). 
Similarly, in India, conservative estimates indicate that up to 550 000 premature deaths can be ascribed to 
the use of biomass fuels annually in population groups reliant on solid fuels for heating and cooking 
purposes (Smith, 2000).  
 
Domestic burning of solid fuels such as coal and wood for cooking and heating purposes, contributes 
significantly to the high levels of ambient and indoor PM levels in many South African communities, 
particularly in the winter months (Engelbrecht et al., 2000; Terblanche et al., 1992b). In the South African 
context specifically, no-income and low-income communities make up 45% of the population, indicating 
that exposure to high air pollution levels is a great concern, as people living in these communities make 
use of domestic burning practices (Stats SA, 2015). Here, no-income households are defined as households 
that survive off no income at all throughout the year, and low-income is defined as a household that has 
an annual income of between R1.00 and R 19 200.00 (Stats SA, 2015).   
 
Historically, in South Africa, working-class communities were built in close proximity to industrial hubs, 
mines and power stations (Dlamini, 2007; Norman et al., 2007; Kimemia & Annegarn, 2011). These areas 
have become characterised by high population densities. Increased levels of air pollution and associated 
health risks are also observed here (Friedl et al., 2008).  Research has shown that, even though such areas 
have access to electricity, electrified households will typically continue to use dirty fuels for cooking- and 
14 
 
space-heating activities, as these fuels are less expensive than electricity (WHO, 2006; Nkosi et al., 2017). 
This is particularly the case in many communities on the South African Highveld, Mpumalanga, which is 
home to most of South Africa’s coal mines, and where coal is readily available and cheap (Balmer, 2007; 
Friedl et al., 2008).  
 
“The survey of energy related behaviour and perception in South Africa” (South Africa, 2012a) indicates 
that households, especially those in no-income and low-income communities, tend to rely on a multitude 
of energy sources, irrespective of their electrification status. This contradicts prevailing energy transition 
theories and the ‘energy ladder’ model, which have typically assumed a straightforward, uni-directional 
shift from traditional to modern energies and appliances once households are provided with electricity 
(South Africa, 2012a). Dirty fuels have been found to be used primarily for cooking and space heating 
activities, whereas electricity has been found to be used mainly for lighting, entertainment and 
refrigeration purposes (Engelbrecht et al., 2002).  
 
Four (4) factors have proven especially significant in determining the energy consumption patterns of 
households in South Africa: namely, rural-urban location, climatic conditions and the associated space 
heating requirements in winter months, as well as proximity to the country’s coalfields (South Africa, 
2012a). In Mpumalanga, 26% of households use coal, this represents the highest of any of the nine 
provinces and stands at nearly four (4) times the national average (7%). This reflects the proximity of 
these communities to some of the country’s major coalfields (South Africa, 2012a). 
 
Among the many pollutants emitted during domestic combustion processes, PM species have emerged as 
the most critical pollutants in almost all urban areas of the world (Health Effects Institute, 2017).  These 
pollutants have been shown to have a significant impact on human health and the ecology of impacted 
environments (Yadav et al., 2014). Industries, waste burning, dirt roads and traffic, all represent sources 
of PM in low-income communities (Friedl et al., 2008).  Of all sources of air pollution that cause negative 
health effects in South Africa however, domestic sources have by far the largest impact (Engelbrecht et 
al., 2000; Mdluli, 2007; Friedl et al., 2008; Kimemia et al., 2016). This statement is supported by studies 
which estimate that ambient air pollution was responsible for 4637 deaths in the year 2000 and that indoor 
air pollution caused 2489 deaths for the same year in six (6) metropolitan areas within South Africa 
(Norman et al., 2007).  
15 
 
2.4 Particulate matter concentrations and variability in low-income communities 
 
2.4.1 Ambient and indoor particulate matter  
 
Concentrations of airborne pollutants, such as PM originating from domestic burning, are governed by a 
combination of complex large- and local-scale atmospheric phenomena. Overall, the temporal and spatial 
variability of airborne PM is influenced by prevailing meteorology including synoptic scale circulation, 
atmospheric stability and local-scale air flow patterns (Gatebe et al., 1999; Mkoma & Mjemah, 2011; 
Dagsson-Waldhauserova et al., 2016).  Times of air stagnation, characterised by high pressure, low winds, 
clear skies and inversions, for example, are responsible for the highest pollution concentrations (Zubkova, 
2003; Alvarado & Prinn, 2009). Conversely, good atmospheric mixing and dilution conditions lead to 
lower concentrations (Terblanche et al., 1992a; Briggs et al., 1997; Riekert, 2011).  Ultimately, PM 
concentrations in the atmosphere are a complex function of various sources of pollution and the capacity 
of the natural- and anthropogenic environment to disperse, dissipate and adsorb pollutants stemming from 
these sources (Cohen et al., 2004).  
 
Various literature sources have acknowledged that ambient pollution levels alone are not necessarily 
indicative of the concentrations of air pollution that humans in low-income communities are exposed to 
on a daily basis (Bruce et al., 2002; Ferro et al., 2004; Diapouli, 2011; Lim et al., 2012). Exposure in such 
communities is a function of time spent in proximity to various pollution sources present in numerous 
micro-environments (Vette et al., 2001). Even though emissions are more often than not dominated by 
outdoor sources, people spend the majority of their time in the indoor environment, where indoor sources 
exist. Significantly, it has been found that exposure, as a function of the degree of pollution in places 
where people spend their time, to a gram of pollution released indoors is likely to cause many times more 
harm than exposure to a gram released outdoors (Smith & Mehta, 2003; Norman et al., 2007). Activities 
which create and influence indoor sources of indoor particulate matter are discussed below.  
 
Significantly, though currently still in an unpublished draft format, the National Department of Health has 
developed a “Guideline for Monitoring Domestic Indoor Air Quality” (South Africa, 2017). Its intention 
is to present indoor air pollution limit values, which are to act as a “trigger to initiate action” to reduce 
16 
 
indoor air pollution, rather than to present a limit for regulatory purposes, as it is virtually impossible to 
limit what people breathe in their own homes.   Indoor PM10 limit values correspond with current ambient 
air quality standard values for PM10 (South Africa, 2017). These guidelines emphasise the importance of 
taking into account indoor concentrations and sources when designing necessary mitigation strategies to 
reduce air pollution in low income communities.  
 
2.4.2 Variability of particulate matter levels in a low-income context 
 
It has commonly been assumed that intra- and even inter-urban PM concentrations are largely homogenous 
in nature and that the use of a single monitoring site to measure ambient concentrations to draw 
compliance-related conclusions with relevant air quality standards is sufficient to make health-based 
decisions (Burton et al., 1996; Wilson, 2006; Morakinyo et al., 2017).  Whilst this assumption may hold 
true in first world country cities more so than not (although, many would argue that this assumption could 
also be refuted), this is an assumption  that cannot easily be made in a low-income community context in 
the developing world, where domestic burning practices are prevalent and poor air quality is a daily reality 
and where a high variability of air pollution concentrations has been demonstrated (Bruce et al., 2000; 
Ezzati & Kammen, 2002; Health Effects Institute, 2017).  
 
The high variability of particulate concentrations within low-income communities is complex and is 
influenced by a variety of factors (Shilton et al., 2002). Primarily, the variation of ambient PM 
concentrations in an urban environment is controlled by emissions, transport, transformation and loss 
processes in the atmosphere and how these processes change throughout the day and across the year 
(Yadav et al., 2014). This means for instance, that concentrations of suspended PM tend to vary with 
shifting seasons (Ali et al., 2015). In a low-income community context, because low-level burning 
activities most often correspond with cooking and heating needs, it makes sense that such activities would 
increase during winter months when the need to keep indoor environments warm, increases. Consequently, 
it has been shown that the PM concentrations in the ambient and indoor environments in the winter months 
are much higher than those measured in the summer months (Song et al., 2015). 
 
17 
 
The concentrations to which individuals are exposed in a low-income setting are also dependent on diurnal 
temperature- and social behaviour patterns. Typically, a bi-modal distribution is most representative of 
diurnal trends of PM concentrations: Two concentration peaks occur during the day, the first occurs during 
the early hours of the morning and the second during the later afternoon /early evening, which each 
coincide with the main burning periods (Yadav et al., 2014).  
 
In the above paragraphs, it has been alluded to that air pollution in an indoor environment is often just as 
bad or worse than in the ambient environment. This represents an example of how intra-urban spatial 
factors can result in PM concentration variability. In many cases in low-income areas, a cause of high 
indoor PM concentrations is the penetration of external emissions of ambient origin into the indoor 
environment (Wilson et al., 2000; Guo et al., 2010; Hoek et al., 2008; Orru et al., 2014; Yangyang et al., 
2015). Conversely, by cumulatively considering the emissions caused by domestic burning activities 
indoors, outdoor PM concentrations increase when household smoke is emitted into the atmosphere 
(Morawska et al., 2001).  
 
Studies on indoor PM concentrations associated with various household fuels from homes in a wide range 
of countries have identified that short-term, peak concentrations of PM10 for instance, can lie within the 
range of 300 to 3000 (or more) micrograms per cubic meter (μg/m3)  (Bruce et al., 2002).  To put this into 
context, the South African annual National Ambient Air Quality Standard for PM10 is 40 μg/m3. The 
equivalent daily PM10 standard is 75 μg/m3 which is one to two orders of magnitude lower than levels seen 
in many homes in developing countries (Bruce et al., 2002).  
 
Socio-economic factors play a large role in the determination of the variability of indoor PM 
concentrations. These include (i) insulation and structural integrity of a household; (ii) the type of house 
(size and design); (iii) the cooking and heating devices as well as the dominant solid fuel type used, and 
finally, (iv) the burning habits or practices of a household (Nkosi et al., 2017). Ultimately, all these 
features determine the amount of solid fuel burnt, which, in turn, determines the emissions to the 
atmosphere or the indoor environment (Scorgie et al., 2003).   
 
Insulation and structural integrity of a house as well as the house design influence the ventilation and 
volume of air that needs to be warmed and kept warm during the winter months.  The cooking and heating 
18 
 
device, as well as its maintenance determines which fuel and how much of it is utilized (Makonese et al., 
2017). Additionally, the fuel characteristics (calorific value, moisture content and density) and burning 
practices (lighting method, secondary airflow control, type of food cooked, or use of multiple stoves or 
fuels) have a significant influence on the emission rate (Ezzati & Kammen 2002). In a most recent study, 
it was demonstrated that using different ventilation rates and ignition methods alone can cause PM2.5 
emission factors to range between 0.2 and 3.3 g/MJ when using D-grade coal in a domestic coal burning 
brazier (Makonese et al., 2017).  
 
2.5 Relationships between ambient, indoor and personal particulate matter concentrations 
 
Understanding the relationships between indoor and ambient (outdoor) PM concentrations in different 
spatial and temporal contexts is important for air quality management in the domestic setting. In order to 
improve exposure understanding and to develop efficient regulatory guidelines for PM, sense needs to be 
made of how these relationships are influenced by different environmental conditions and also by societal 
habits (Morawska et al., 2001; Vette et al., 2001). It has been found that, when comparing ambient, indoor 
and personal PM concentrations with each other, ambient concentrations are typically lower than indoor 
concentrations, and indoor concentrations are typically lower than personal PM concentrations (Adgate et 
al., 2002). Simultaneous measurements of these three parameters have only shown weak to moderate 
correlations between them, highlighting their complex relationship, and at the very minimum, indicating 
that the sources in the ambient and the indoor environment are different (Mohammadyan et al., 2017). It 
follows then, that the more dissimilar the indoor and ambient PM loadings, the more likely it is that the 
sources are different.  A good relationship between ambient and personal or indoor and personal PM 
concentrations means that a person wearing a personal monitoring device spent a considerable proportion 
of time within the environment in question (Janssen et al., 2005).  
 
Measuring ambient, indoor and personal PM concentrations at the same time to investigate their 
relationships creates the best opportunity to understand what people in communities breathe and whether 
ambient or indoor concentrations alone can simply be used as a personal exposure surrogate (Nakai & 
Tamura, 2008).  If ambient, indoor and personal PM concentrations measured across various seasons are 
similar in trend, magnitude and variation, it can be carefully assumed that ambient measurements can 
19 
 
represent a surrogate for personal concentrations; if this is not the case however, then it can be argued that 
ambient measurements do not at all represent what people breathe and that straightforward compliance-
related decisions cannot be made using ambient measurements only (Huang et al., 2015).  
 
2.6 Relevance of this study  
 
The scientific community has for some time raised the concern that air quality in low-income communities 
in sub-Saharan Africa surpasses “safe-to-breathe” concentrations, guided by ambient standards (Chen & 
Smith 1991; Bruce et al., 2000; Scorgie et al., 2003; Josipovic et al., 2010; Garland et al., 2017; Health 
Effects Instutute,  2017).  Most studies have treated ambient, indoor and personal air pollution separately. 
Limited studies in South Africa have simultaneously considered air quality in both the ambient and the 
indoor environment for extended periods of time and concurrently considered personal exposure to air 
quality (Terblanche et al., 1992a; Terblanche et al., 1992b; Engelbrecht et al., 2000; Engelbrecht et al., 
2002; Mdluli, 2007). Previous findings have pointed to the fact that ambient air pollution levels, at times, 
exceed ambient standards, that indoor concentrations are higher than ambient concentrations (Petzer, 
2009) and that personal exposure to PM is unacceptably high (Terblanche et al., 1992b).  No South African 
studies have been conducted that have simultaneously measured and assessed the relationships between 
ambient, indoor and personal PM concentrations at such a large scale and this across various seasons. 
Similarly, this is the first study to include personal GPS monitor readings to help better define personal 
exposure in different micro-environments.  
 
This research seeks to confirm findings that have been made in previous studies which have postulated 
that ambient, indoor and personal PM concentrations measured in low-income communities in South 
Africa at times exceed air quality standards. Furthermore, this study brings the international debate of 
whether stationary ambient air quality monitoring stations are adequate proxies to use for compliance-
related decisions as well as for epidemiological enquiries into the South African arena.  For the first time 
in South Africa, a study corresponds simultaneously logged personal GPS readings and personal PM 
concentration measurements to better understand time-activity patterns of an individual in a low-income 
community context in various micro-environments and the PM concentrations breathed in those 
environments.  An unprecedently rich set of ambient, indoor and personal measurements, which were 
20 
 
collected during two separate pilot studies conducted for Sasol and Eskom, respectively in KwaDela and 
KwaZamokuhle, form the basis of this research.  
 
 
********************* 
 
This chapter presented the literature consulted to place this study into 
context. Main points to carry through the remainder of the thesis are that PM 
concentrations in communities that rely heavily on solid fuels for heating 
and cooking purposes have been shown to, at times, exceed ambient 
standards. The ambient, indoor and personal PM concentrations measured 
in these environments are highly variable in space and time.  High variability 
of PM concentrations in space and time make it difficult to use ambient 
measurements taken at one stationary ambient monitoring station as proxy 
data to represent air quality in the entire community. Assessing personal 
exposure and intake doses is a complex task and requires information on 
exposure concentrations and personal inhalation rates.  The following 
chapter discusses the study design, how data for the study were collected 
and then how these data were analysed. 
 
 
 
 
 
 
 
 
 
 
 
21 
 
 Research Methodology 
 
In this chapter, the methods used to collect and analyse the data used for this 
study are outlined in detail. As the data were collected at two different sites 
and as part of two different study campaigns, emphasis is placed on the 
detailed provision of information to ensure it is understood how, when, 
where and why the data were collected.  The study for this dissertation forms 
an amalgamation of two similar study designs created for Sasol and Eskom’s 
air quality offset pilot studies by NWU and The NOVA Institute. 
Limitations and assumptions taken into account in the study design and 
result interpretation are elaborated upon. 
 
3.1 Study area 
3.1.1 Geographical location of the study sites 
 
Two low-income communities, located within the Highveld Priority Area (declared in 2007), 
Mpumalanga, South Africa, have been chosen for the purposes of this comparative study, KwaDela and 
KwaZamokuhle (Figure 3.1 and Figure 3.2).  An airshed is declared as a designated Air Quality Priority 
Area because ambient air quality standards are being exceeded, or may be, in the future (South Africa, 
2007). Air quality on the Highveld is characterised by elevated concentrations of criteria pollutants 
stemming from industrial and non-industrial sources (South Africa, 2016; Garland et al., 2017). Notable 
non-industrial sources of criteria pollutants affecting air quality in many low-income communities on the 
Highveld are domestic burning activities (Mdluli, 2007; Kimemia & Annegarn, 2011). Basic demographic 
statistics illustrate that people living in both KwaDela and KwaZamokuhle are highly dependent on coal 
as a readily available energy source for domestic cooking and heating, and it has been shown that these 
practices have a negative impact on the air quality in KwaDela (Table 3.1) (Nkosi et al., 2017).  
 
 
 
22 
 
 
Figure 3.1.  Map of study sites in relation to each other in the Mpumalanga Province, South Africa 
 
 
Figure 3.2.  Overview of study sites in relation to relevant suburbs and district municipalities 
 
23 
 
KwaDela is a small community (population of 3 781) which lies in the Msulkaligwa Local Municipality 
and the Gert Sibande District. It is located next to the N17 between Bethal and Ermelo. Davel is the closest 
other town, situated just north of KwaDela (Figure 3.3). Apart from railroad traffic, road traffic on the 
N17 and local traffic within Davel and the community itself, solid fuel burning from industry and domestic 
households as well as waste burning can be considered the main contributors to ambient air pollution in 
KwaDela (Msukaligwa Local Municipality, 2010). 
 
KwaZamokuhle (population of 20 427) is situated about 40km north-east of KwaDela and lies within the 
Nkangala District Municipality just north of the town of Hendrina and close to Hendrina Power Station. 
Diurnal timeseries analyses comparing hourly average PM concentrations in KwaZamokuhle and nearby 
Hendrina have confirmed that KwaZamokhle’s air quality is characterised by early morning and late 
afternoon-/ evening peaks, indicating domestic burning activities associated with heating and cooking in 
the early mornings and late afternoons (Eskom, 2017).  
 
Coal resources on the Mpumalanga Highveld are an easy-to-access energy source for households in this 
region, making it the primary source of fuel for cooking and heating in many low-income communities. 
Numerous coal mines and twelve coal fired power stations are situated within the Highveld region, bearing 
testimony to the fact that coal is plentiful. 
 
Table 3.1.   Relevant community statistics for KwaDela and KwaZamokuhle taken from the Census 
2011 - an overview (Stats SA, 2012b) 
 KwaDela KwaZamokuhle 
Coordinates 26.4633 S, 29.6644 E 26.1346 S, 29.7317 E 
Municipality Msukaligwa local Steve Tshwete local 
Total population 3 781 20 427 
Number of households 982 5 874 
People per household 3.9 3.5 
People living with < R4 800.00/ 
month 
21.6% 16.9% 
Use of coal as a source for 
heating and cooking * 
46.7% and 42.2%, respectively 34% and 31.2%, respectively 
Use of electricity for lighting 
purposes 
96.7% 89.2% 
*primary energy carrier for these purposes 
24 
 
3.1.2 Meteorology in KwaDela and KwaZamokuhle 
 
 
Warm and temperate climatic conditions prevail in KwaDela and KwaZamokuhle throughout most of the 
year. The communities lie within a summer rainfall region and precipitation ranges from 83 - 132 
mm/month in summer to 5 - 29 mm/month in winter. Highest average maximum temperatures are 
experienced from December to March (~24 - 25°C) and lowest average minimum temperatures from May 
to August (~ 0 - 4°C) (Climate-Data, 2017). Winds are highly variable with more prevalent easterly and 
westerly components. 
 
During the study campaign the stark contrast between average morning/ night minimum and average 
daytime maximum temperatures during the winter months were noted (Table 3.2 and Table 3.3). 
Temperature, relative humidity and pressure patterns showed a diurnal cycle. 
 
Table 3.2.  Descriptive statistics of 1-hour resolution meteorological data for each of the campaigns in 
KwaDela 
 
Wind Speed 
(m/s) 
Temperature 
(°C) 
Relative 
Humidity 
(%) 
Solar 
Radiation 
(W/m2) 
Winter 2013 
Min 0.0 -4.5 5.3 0.0 
Max 12.9 28.2 100.0 804.1 
Ave 3.9 10.8 52.1 174.8 
Summer 2014 
Min 0.0 3.1 8.6 0.0 
Max 12.1 30.1 100 1439.8 
Ave 3.1 16.4 71.2 208.4 
Winter 2014 
Min 0.0 -3.8 3.2 0.0 
Max 15.7 34.1 100 1014.1 
Ave 3.4 11.9 44.2 203.6 
Summer 2015 
Min 0.0 6.7 14.8 0.0 
Max 13.2 32.6 99.2 1099.9 
Ave 2.9 17.7 68.8 213.5 
 
 
 
 
 
 
 
 
25 
 
Table 3.3.  Descriptive statistics of 1-hour resolution meteorological data for each of the campaigns in 
KwaZamohule 
 
Wind Speed 
(m/s) 
Temperature 
(°C) 
Relative 
Humidity 
(%) 
Solar 
Radiation 
(W/m2) 
Spring 2015 
Min 0.0 11.3 5.4 0.0 
Max 11.9 34.6 100 7999 
Ave 3.2 24.0 13.9 203.6 
Summer 2016 
Min 0.0 10.9 8.4 0.0 
Max 9.7 32.8 100 7999.0 
Ave 2.2 23.0 68.7 177.5 
Winter 2016 
Min 0 -6.4 0.93 0 
Max 10.3 31.1 100 7999 
Ave 2.3 11.5 47.8 177.9 
 
 
3.2 Data collection 
 
3.2.1 Data collection as part of larger scale studies 
 
Ambient, indoor and personal PM concentrations were measured as part of campaigns in KwaDela and 
KwaZamokuhle, respectively. Questionnaire based surveys were also conducted to better understand the 
energy use patterns of the respective areas and to contextualise the data stemming from ambient air quality 
assessments.  The campaigns in KwaDela and KwaZamokuhle were conducted from 2013 and 2014 
onwards for Sasol and Eskom’s air quality offset pilot studies and aimed to ultimately test the effectiveness 
of various air quality household offset interventions in a low-income community context to reduce ambient 
PM levels.  
 
3.2.2 Air quality monitoring in the ambient and the indoor environment  
 
A mobile ambient air quality monitoring station was set up to obtain continuous ambient concentrations 
of PM in both considered communities. PM10 and PM2.5 were measured in the ambient monitoring stations 
using Horiba BAM 1020 and MetOne E-Bam monitors in KwaDela and THERMO Beta Gauges in 
KwaZamokuhle. These devices are instruments typically used for compliance monitoring purposes (Table 
3.4 and Table 3.5) (Raja et al., 2017).  Basic meteorological data (temperature, humidity, pressure, wind 
26 
 
speed and direction, rainfall) were also monitored. Additionally, both campaigns made use of numerous 
individual stationary samplers, which were spread throughout the communities, to measure ambient PM10 
and PM2.5 concentrations in different locations across the settlements, to gain a better understanding of the 
spatial variability of PM concentrations (Figure 3.3 and Figure 3.4).  
 
 
 
Figure 3.3.  Overview of ambient and indoor monitoring sites located in KwaDela 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
27 
 
 
Figure 3.4.  Overview of ambient and indoor monitoring sites located in KwaZamokuhle 
 
Table 3.4.  E-Sampler and E-Bam codes for in-text interpretation and orientation according to cardinal 
directions 
E-Sampler /E-
Bam number on 
map 
Code per E-Sampler according to cardinal directions for 
interpretation and orientation in text 
 KwaDela KwaZamokuhle 
E1 KD-E1-NW KZ-E1-N 
E2 KD-E2-SW KZ-E2-NW 
E3 - KZ-E3-W 
E4 - KZ-E4-SE 
  
The sampling campaigns in KwaDela included winter (2013 and 2014) and summer (2014 and 2015) 
months, whilst the campaigns in KwaZamohule were undertaken in spring (2014), summer (2016) and 
winter (2016) months, to take into account seasonal variability of PM concentrations (Table 3.5).  
 
28 
 
Indoor air quality monitoring took place in 20 and nine (9) individual, semi-randomly selected and 
consenting households in KwaDela and Kwazamokuhle, respectively (Table 3.6). A multi-stage sampling 
method selected participating houses, where street blocks and one household from each street block were 
randomly selected. This method ensured a relatively even geographic distribution of monitored 
households in the two communities.  The indoor monitoring equipment was typically placed in the kitchen 
or the living room of the household. Indoor PM4 concentrations were measured using TSI DustTrak 
(Models 8520 and 8530) photometric monitors at 5-minute time intervals.  
 
Personal exposure of a small number of community members in households in which indoor air pollution 
was monitored was measured as PM4 at 5-minute time intervals using a TSI SidePak AM510 photometric 
monitor, which was deployed at a height equivalent to a person’s breathing zone. Informed consent was 
given by those who wore personal monitors.  To complement the PM measurements of the SidePak 
monitor, a few select individuals (middle-aged women) in KwaDela also carried along with them Global 
Positioning System (GPS) monitors which tracked their movement for a given period of time. It was a 
requirement of the study that only persons who spent most of their time in KwaDela were allowed to 
participate in this part of the study, so as to ensure readings were focused to concentrations experienced 
in KwaDela only.  
 
The indoor and personal PM4 measurements used have been corrected according to the specific 
photometric calibration factors obtained for both the DustTrak and SidePak instruments (Language et al., 
2015). As part of quality control and quality assurance procedures, the instruments within the mobile 
monitoring station were checked once a week during the sampling campaigns. The indoor instruments 
were zero calibrated and flow checked as per manufacturer’s directions. The personal monitoring 
instruments were flow checked once a week and zero calibrated each day before sampling proceeded. 
 
 
 
 
 
 
 
 
29 
 
Table 3.5.  Overview of ambient data collection in KwaDela and KwaZamokhule 
 KwaDela KwaZamokuhle 
Source of data Sasol Offset Pilot Study Eskom Offset Pilot Study 
   
Social data 
Number of Households at which 
surveys were conducted 
Total = 195 
Total = 919 
Good quality = 692 
Date of surveys August 2013 – July 2014 November 2014 – October 2015 
Content of surveys 
General household survey (energy 
use patterns) including a 
community health survey 
General household survey (energy 
use patterns) including a 
community health survey 
 
Ambient PM measurements 
Measured pollutants PM10/ PM2.5 PM10/ PM2.5 
Other available data Meteorological data Meteorological data 
Instrument type used 
Horiba BAM 1020 (PM10) – (beta 
ray attenuation), MetOne E-Bam 
(PM2.5) (beta ray attenuation), 
2 x MetOne E-Sampler (E1 = 
PM2.5, E2 = PM10) (Light-scattering 
aerosol monitors) 
THERMO – Beta gauge (PM10 and 
PM2.5) (non-step-wise beta ray 
attenuation) 
4 x E-Bam Beta gauges (E1 – E4 = 
PM2.5) (beta ray attenuation) 
Flow-rate and set-up 
Flow Rate 16.7 L.min-1 (actual or 
standard flow), automated span 
checks. 
Flow Rate 16.7 L.min-1, 
automated span checks 
Data resolution for this study 
5 minutes intervals averaged into 
60-minute interval data 
5 minutes intervals averaged into 
60-minute interval data 
Location of main ambient 
monitoring station 
Secondary School close to the 
centre of KwaDela 
On church grounds in the east of 
KwaZamokuhle 
Timeframes of campaigns 
Winter 2013 (4 July – 27 
September) 
Summer 2014 (10 February – 6 
May) 
Winter 2014 (11 July – 17 
September) 
Summer 2015 (11 February – 13 
April) 
Spring 2015 (2 September – 27 
November) 
Summer 2016 (24 February – 19 
April) 
Winter 2016 (12 July – 9 
September) 
 
 
 
 
 
 
 
30 
 
Table 3.6.  Overview of indoor and personal data collection in KwaDela and KwaZamokhule 
Indoor PM measurements 
 KwaDela KwaZamokuhle 
Measured pollutants  PM4 PM4 
Instrument type 
TSI DustTrak Model 8530 (light-
scattering laser photometers) 
TSI DustTrak Model 8530 (light-
scattering laser photometers) 
Data resolution 
5 minutes intervals averaged into 
60-minute interval data 
5 minutes intervals averaged into  
60-minute interval data 
Set-up 
A 10-mm Nylon Dorr Oliver 
Cyclone inlet was used with each 
instrument, at a flow rate of 1.7 
L.min-1, to acquire the required 
50% cut size at 4 μm. The output 
for the particulate mass 
concentration is given in milligram 
per cubic meter (mg.m-3). An 
averaging period of five minutes 
was used.  These instruments are 
factory calibrated using Arizona 
test dust. 
A 10-mm Nylon Dorr Oliver 
Cyclone inlet was used with each 
instrument, at a flow rate of 1.7 
L.min-1, to acquire the required 
50% cut size at 4 μm. The output 
for the particulate mass 
concentration is given in milligram 
per cubic meter (mg.m-3). An 
averaging period of five minutes 
was used.  These instruments are 
factory calibrated using Arizona 
test dust. 
Number of households in which 
indoor monitoring took place 
20 9  
Timeframe 
Winter 2013 (4 July – 27 
September) 
Summer 2014 (10 February – 6 
May) 
Winter 2014 (11 July – 17 
September) 
Summer 2015 (11 February – 13 
April) 
Spring 2015 (29 September – 18 
November) 
Summer 2016 (22 February – 20 
April) 
Winter 2016 (12 July – 23 August) 
 
Personal PM measurements 
Measured pollutants PM4 PM4 
Instrument type 
TSI SidePak AM510 (light-
scattering laser photometers) 
TSI SidePak AM510 (light-
scattering laser photometers) 
Data resolution 
5 minutes intervals averaged into 
60-minute interval data 
5 minutes intervals averaged into 
60-minute interval data 
Set-up 
Scattering nephelpometers (laser 
has λ = 680 nm).  The instruments 
are light weight and equipped with 
10-mm Nylon Dorr-Oliver 
Cyclones effecting a 50 % cut-off 
at 4 μm.  The battery life of the 
Scattering nephelpometers (laser 
has λ = 680 nm).  The instruments 
are light weight and equipped with 
10-mm Nylon Dorr-Oliver 
Cyclones effecting a 50 % cut-off 
at 4 μm.  The battery life of the 
31 
 
personal monitors was between 8 
and 12 hours.  Individuals were 
therefore requested to carry the 
monitors only during the daytime 
07H00 to 19H00.   
personal monitors was between 8 
and 12 hours.  Individuals were 
therefore requested to carry the 
monitors only during the daytime 
07H00 to 19H00.   
Number of people wearing 
SidePaks 
20 individuals 8 individuals 
Time period of GPS 
measurement availability for 
one-individual-case-study  
21-28 August 2013 (Individual was 
a middle-aged woman) 
NA 
Timeframe 
Winter 2013 (4 July – 27 
September) 
Summer 2014 (10 February – 6 
May) 
Winter 2014 (11 July – 17 
September) 
Summer 2015 (11 February – 13 
April) 
Spring 2015 (29 September – 18 
November) 
Summer 2016 (22 February – 20 
April) 
Winter 2016 (12 July – 23 August) 
 
 
3.2.3 Measuring personal exposure to particulate matter concentrations in various micro-
environments 
 
 
One unique component of the study in KwaDela was to track the daytime movement patterns of consenting 
individuals within the community using personal monitors and mobile GPS monitoring devices (Table 
3.6). The purpose of this exercise was to establish where individuals were exposed to the greatest PM 
concentrations. Micro-environments were identified by plotting all of the recorded GPS readings taken in 
KwaDela on a SPOT 6/7 (2015) satellite image of KwaDela using QGIS version 2.14.1 (Figure 3.5).  This 
is how it was ascertained where individuals spent most of their time whilst wearing personal monitoring 
devices (notably, this was during day-time hours). 
 
32 
 
 
Figure 3.5.  Examples of GPS coordinates plotted on a SPOT 6/7 image of KwaDela to help identify 
common micro-environments (The colours of the dots denote the movements of the various individuals 
wearing GPS monitors on different days. For the purpose of this study, the colours have no significance. 
This figure aims to illustrate movement patterns of individuals wearing the GPS monitors to assist in 
micro-environment identification and nothing more) 
 
3.3 Data analysis 
 
3.3.1 Characterising particulate matter concentrations at a community level 
 
 
Differently averaged continuous ambient PM measurements were used for the purposes of statistical 
analysis. Microsoft Excel 2016 was used to conduct statistical analyses for summary purposes. Five (5)-
minute average data were used to create hourly- and daily averaged data sets for PM10, PM4 and PM2.5. 
Statistica 13.2 was used to create summary box plots to compare different daily average PM concentrations 
at a community level. For the purposes of averaging, the 99th percentile was used to ensure results could 
be used to make compliance related conclusions. Where the daily average PM concentrations were 
derived, the geometric mean with its standard deviation, were considered. Data gaps were dealt with by 
means of mean-substitution and regression analyses considered pairwise deletion.  
33 
 
Statistica 13.2 was also used to create boxplots to compare averaged maximum values measured across 
the sampling points. Outliers and extreme observations were included in graphs and in analyses where 
median values are displayed but excluded where mean values are displayed as the mean is sensitive to 
outliers. Here outliers are defined as data points with a value greater/ smaller than the upper/ lower value 
of the box on the plot (i.e. the 25th and 75th percentile value) ± the coefficient of 1.5 multiplied by the 
difference between the upper and the lower value of the box. Extreme values plotted represent the same 
value, derived in the same manner, using two times the outlier coefficient (these are default settings in 
Statistica 13.2). Extreme values have been kept in these graphs to indicate events of extreme/ acute 
exposure.  
 
As PM concentrations are influenced by seasonal variation, spring, summer and winter concentrations 
were considered separately in every analysis conducted. Where possible, comparisons were made between 
KwaDela and KwaZamokuhle.  
 
3.3.2 Characterising particulate matter concentrations at a household level 
 
 
Two households per community were selected in order to gain a better understanding of household-level 
PM characteristics and relationships. The households were selected based on the fact that indoor and 
personal PM measurements were conducted for these selected houses and so cases could be formed, 
allowing for seasonal comparison of ambient, indoor and personal PM concentrations. A case represents 
a dataset which contains different parameters which were available simultaneously, i.e. data were available 
for ambient PM10 and PM2.5 as well as for indoor and personal PM4 measured during the same time periods, 
allowing direct comparisons between these PM types. The number of cases typically represents the 
number of indoor sampling campaigns that took place in a community. Across the seasonal campaigns, a 
total of 33 cases were identified in KwaDela and a total of 24 cases were identified in KwaZamokuhle 
(Table 3.7 and Table 3.8). For this study, house 3 and 11 were chosen in KwaDela and house 3 and 9 were 
chosen in KwaZamokuhle (Table 3.9). 
 
 
 
 
34 
 
Table 3.7.  Overview of cases used for comparative analyses in KwaDela 
Campaign Case name Campaign Case name 
Winter 2013 
(W13) 
H2 
Summer 2014 
(S14) 
H3* 
H3* H5 
H4 H11* 
H10 H13 
H11* H15 
H13 H18 
H14  
Campaign Case name Campaign Case name 
Winter 2014 
(W14) 
H5 
Summer 2015 
(S15) 
H1 H11* 
H6 H3* H13 
H14 H4 H18 
H15 H5 H21 
H18 H6 H22 
H19 H8 H23 
H22 H10  
Note: Going forward, a specific case is referred to by combining the season in which a sampling campaign took place with the number of the house in which   
indoor and personal monitoring took place, e.g. case W13H2 represents the ambient, indoor and personal measurements taken at the same time in winter 2013 
in house 2 and case S14H18 represents the ambient, indoor and personal measurements taken at the same time in summer 2014 in house 18.  
*Cases chosen for the in-depth household-level analyses 
 
 
Table 3.8.  Overview of cases used for comparative analyses in KwaZamokuhle 
Campaign Case name Campaign Case name Campaign Case name 
 
 
Spring 2015 
(Spg15) 
H3* 
Summer 2016 
(S16) 
H1 
Winter 2016 
(W16) 
H1 
H4 H2 H2 
H5 H3* H3* 
H7 H4 H4 
H8 H5 H5 
H9* H6 H6 
 H7 H7 
  H8 H8 
  H9* H9* 
*Cases chosen for the in-depth household-level analyses 
 
 
 
 
 
 
 
 
 
 
 
35 
 
Table 3.9.  Overview of cases chosen for household-level analyses 
KwaDela W13 S14 S15 
 Dates Ambient 
Temp 
Dates Ambient 
Temp 
Dates Ambient 
Temp 
H3 24/7-21/8 10.12±5.49 10/2-26/2 18.56±4.59 11/2-2/3 19.86±5.22 
H11 21/8-29/8 12.85±5.43 5/3-18/3 17.48±2.70 11/2-2/3 19.86±5.22 
KwaZamokuhle Spg15 S16 W16 
 Dates Ambient 
Temp 
Dates Ambient 
Temp 
Dates Ambient 
Temp 
H3 29/9-26/10 22.36±6.29 7/3-24/3 19.06±4.49 27/7-23/8 10.27±8.03 
H9 27/10-18/11 19.26±7.04 23/2-7/3 20.32±5.28 27/7-23/8 10.27±8.03 
 
 
Statistica 13.2 was used to plot daily average bar charts for every PM type measured per case for every 
season. Microsoft Excel 2016 was used to plot hourly average timeseries to ascertain the diurnal variations 
of PM concentrations at an ambient, indoor and personal level in direct comparison. Hourly indoor PM4: 
personal PM4 ratios and indoor PM4: ambient PM2.5 ratios were ascertained for the chosen houses and also 
plotted on an hourly average timeseries graph.  
 
3.3.3 Comparing ambient, indoor and personal particulate matter concentrations and understanding 
their relationships  
 
 
Simple box plots were plotted for maximum hourly concentrations for every case for every sampling 
campaign by using Statistica 13.2. Medians, 25th and 75th percentiles as well as minimum and maximum 
concentrations were plotted for each PM type for each case. Outliers and extreme observations were 
included in the graphs to show extreme/ acute exposure events. 
 
Two different sets of regressions were plotted of hourly, lognormally transformed datasets using Statistica 
13.2 and Microsoft Excel 2016 to ascertain whether there was a noteworthy relationship between ambient, 
indoor and personal PM concentrations. Data points above the 90th percentile were excluded from this 
analysis, as regressions are sensitive to outliers. The coefficient of determination resulting from the 
regression (R2) analysis was used to determine how the percent of variation in one PM type could be 
attributed to the variation in another PM type. The closer an R2 value is to 1, the greater the likelihood that 
the variation in one PM type could be attributed to the variation in another PM type: 
36 
 
1. To ascertain whether or not there were notable relationships between PM type concentrations 
measured at a household-level, regressions were plotted for each and every case, for each and 
every ambient, indoor and personal PM combination (i.e. indoor and personal PM4 vs ambient 
PM10, indoor and personal PM4 vs ambient PM2.5, indoor and personal PM4 vs ambient PM2.5 
measured at E-Sampler-1/ E-Bam 1, E-Sampler 2/ E-Bam 2 and so forth). Regressions were also 
plotted between indoor and personal- as well as between the various ambient PM types themselves 
(E.g. E-Sampler 1 vs E-Sampler-2 or E-Bam 1 vs E-Bam 2 etc.).   Results of these regression 
analyses have been illustrated by using the cases chosen for the two-household case study (Chapter 
5). 
 
2. Regressions that indicated more overarching relationships between indoor and ambient PM 
concentrations at a community-level were plotted by considering total daily- and hourly average 
concentrations as well as daily- and hourly maxima and plotting these against each other for each 
community. To illustrate any overarching community-level trends the coefficient of determination 
resulting from every regression analysis was then plotted against its corresponding slope in a 
scatter plot using R (version i386 3.3.1).  
 
Additional to the above, relationships between indoor and personal as well as ambient and indoor PM 
concentrations were explored through the derivation of ratios, where 
 
                                          Personal PM4/ Indoor PM4 ratio = Ci/ Cp                      (3) 
 
and where Ci represents indoor PM concentrations and Cp represents personal PM concentrations. Ratios 
>1 indicate higher indoor PM concentrations and ratios <1 indicate higher personal concentrations. If the 
ratio = 1, it means that indoor and personal PM concentrations are the same, and thus it can be extrapolated 
that the individual wearing the personal monitor is in the vicinity of the same source of PM that determined 
the indoor concentrations measured. Similarly, 
 
 Indoor PM4/Ambient PM2.5 ratio = Ci/Ca        (4) 
 
37 
 
where Ca represents ambient PM2.5 concentrations. On the same note, a ratio < 1 indicates higher PM2.5 
concentrations and a ratio of >1 indicates higher indoor PM4 concentrations. If the ratio = 1, it means that 
the same sources are influencing the magnitude of the indoor and ambient PM concentrations.  
 
3.3.4 Assessing personal exposure to particulate matter in different micro-environments 
 
Personal PM4 concentration measurements from 21 – 28 August 2013 were, where possible, paired with 
the GPS tracking data at ten-minute intervals, to ascertain the PM4 concentrations to which an individual 
was exposed in a given location and at a given time period (Table 7.1 in Annexure C). PM4 concentrations 
were paired to a specific micro-environment as explained in section 3.2.3 above.   
 
A map was created in QGIS to illustrate the above described for one day, 21 August 2013. For the sake of 
anonymity, polygons were drawn around the households within which the individual spent time. Dots 
representing the specific individual’s locations were coloured in a graduated manner, and so the colour of 
the dot represents the PM4 concentration measured by the SidePak monitor at a given time and in a given 
location. Indoor PM4 measurements for the house in which the individual spent the most time were 
included in the analysis. Hourly average diurnal timeseries were plotted in Excel for personal 
measurements taken from the 21-28 August 2013.  
 
Indirect, time-weighted, integrated exposure concentrations as well as time-weighted, integrated potential 
intake doses were ascertained by using the U.S. EPA equations (1) and (2) listed in Chapter 2 (Table 7.2 
in Annexure D).  Input data consisted of indoor PM4  measurements and ambient PM2.5 measurements as 
well as information gathered on the averaged time-activity patterns of the person who wore the GPS 
monitor. PM4  and PM2.5  were directly compared, based on rationale explained in section 3.5 below. Direct 
intake doses were derived using the personal measurements available for the case-study. Short-term 
inhalation rates (m3/min) for an average middle-aged woman at light intensity activity, as recommended 
by the U.S. EPA, were used for the purposes of this exercise. This was then converted to an equivalent 
daily breathing rate (0.012 m3/min *60 *24 =17.28 m3/day) (U.S. EPA, 2011). It must be noted that the 
GPS/ Personal PM4 data used do not represent a full day’s worth of data, and hence the short-term 
breathing rate was deemed appropriate. High resolution ambient PM2.5 and indoor PM4 data were also 
used to ascertain indirect dose inhalation rates for the exact time periods during which personal 
38 
 
measurement were available, to see whether or not the resulting doses would differ substantially from the 
directly derived dose rates. 
 
3.4 Ethical considerations 
 
 
This research was conducted in an ethically responsible manner and relevant norms and standards were 
complied with. Due to the health-related focus of the study, the Faculty of Natural Science and Agriculture 
of the North West University referred the ethical clearance of the project to the National Health Research 
Ethics Council (HREC). Two separate project proposals, one for KwaDela and one for KwaZamokuhle, 
were reviewed and monitored by the national HREC. Ethical clearance numbers for KwaDela (NWU-
00066-13-A3) and for KwaZamokuhle (NWU-0158-14-S3) were received. The identity of all individuals, 
whose movements, locations and personal information were tracked in this study remain anonymous and 
data was only gathered if explicit consent was given to do so.  
 
3.5 Limitations and assumptions 
 
 
Though main findings resulting from this study are clear, assumptions are being made that these findings 
represent a true reflection of PM concentrations to which individuals are exposed in the indoor and 
ambient environment in KwaDela and KwaZamokuhle, respectively. The large intra- and inter-urban 
variability of the concentrations measured, makes it difficult to quantify and summarise concentrations 
with certainty and it is impossible to draw statistically significant conclusions, unless adequately large 
sample sizes are considered. The size of a study design of this nature is limited however by financial- and 
human resources. It is a time-consuming exercise and tedious to ensure the equipment is maintained to 
manufacturer’s standards and that it is up and running to ensure data collected can be adequately used for 
research purposes. Though large variability in ambient, indoor and personal PM concentrations in this 
study is evident, an assumption is being made that the snapshot in space and time that is presented here, 
gives a realistic indication of what is measured on the ground in these communities on a day-to -day basis.  
 
39 
 
The results ascertained are highly variable and highly context specific. This means for instance that the 
PM concentrations measured in one location at a given time will not necessarily represent the PM 
concentrations measured at a different location at the same time. Even though temporal, spatial and socio-
economic factors are accounted for to make sense of the data presented, these conclusions are drawn with 
caution, and cognisant of the fact that other factors not mentioned in this study could well have played a 
role in determining PM concentrations measured at any given moment in time.  
 
A further assumption that is being made is that PM concentrations of different size fractions measured in 
the ambient and the indoor environment can easily be compared. This means that PM10 and PM2.5 
concentrations in the ambient environment are being compared with PM4 concentrations in the indoor 
environment to ascertain relationships. This is considered possible because the PM size fractions represent 
sub-sets of each other. PM4 concentrations measured in the indoor environment and with personal 
monitors have been directly compared to PM2.5 concentrations measured in the ambient environment. It is 
assumed that, as most indoor PM concentrations stem from domestic burning practices, indoor PM will 
more than likely fall under the PM2.5 size fraction, and so the PM4 concentrations measured will likely be 
represented to a large degree by particles < 4 microns. Additionally, as there are no ambient standards for 
PM4, it is being assumed that if daily PM4 concentrations exceed the daily average PM10 standard, non-
compliances with the NAAQS are being incurred. This is in line what has been drafted in the National 
Guidelines for Monitoring Domestic Air Quality, which suggest indoor PM10 limit values equal to those 
of the ambient PM10 limit values of the NAAQS (South Africa, 2017). 
 
Another constraint is represented by the fact that data is collected in two separate settlements at two 
different times. This means that, though average winter and summer concentrations for specific PM types 
are ascertained and ultimately used for comparative purposes, the years in which the data were collected 
do not correspond entirely. In fact, only Summer/ Spring 2015 represents a time during which the data 
corresponds to a degree (i.e. data were collected in both KwaDela and KwaZamokuhle during this time). 
This means that any larger scale influences that might have determined ambient or indoor concentrations 
during the sampling campaign in KwaDela in a given year will not have been present in KwaZamokuhle 
and vice versa.  
 
40 
 
Any studies that deal with people are governed by large complexities. Large assumptions are being made 
that survey results are a true reflection of reality, though they might be subjective at times. Additionally, 
when individuals handle personal monitoring equipment (SidePak or GPS monitors), it is assumed that 
these are being worn at all times and by the same individual who consented to wearing them. An immediate 
limitation is that the personal PM measurements taken in KwaDela represent only measurements taken 
during the day hours and not at night, and so “total” exposure is a relative term. Paired GPS and personal 
PM data are not continuous. This study could be meaningfully built upon by:  
 
- measuring simultaneous ambient, indoor and personal PM concentrations of the same size 
fractions to allow for a more direct relationship comparison  
- calculating and measuring infiltration rates to better understand how ambient air quality affects 
indoor air quality and vice versa 
- adding detailed time-activity journals at an individual level to even better understand and dissect 
what determines total personal exposure. This would help determine to what degree even personal 
habits such as cleaning/ sweeping or smoking contribute to indoor PM exposure as compared to 
indoor burning activities 
-  taking GPS tracking measurements in KwaZamokuhle as well, to see if the micro-environments 
identified in this study in KwaDela correspond with those that would be identified in 
KwaZamokuhle 
-  taking Personal PM- and GPS measurements continuously to ensure a full picture of personal 
exposure is painted 
 
 
********************* 
 
This chapter has outlined the methods used to achieve the objectives of this 
study. Main data collection methods that were used were 1) continuous 
ambient, indoor and personal PM monitoring and 2) collection of survey 
data.  Real-time personal GPS data was also collected. Data were then 
analysed by means of appropriate time-averaging exercises (hourly averages 
were ascertained for diurnal time-series graphs and daily average 
concentrations were ascertained to compare concentrations with existing 
NAAQS). Box plots were created to compare mean and median daily 
41 
 
average and maximum concentrations at a community level. Cases were 
identified where data were simultaneously available for ambient, indoor and 
personal PM measurements, which enabled the direct comparison between 
these measurements at a household and at a community level. Finally, real-
time PM and GPS measurements were paired to ascertain personal exposure 
and potential inhalation doses of an individual in KwaDela. The next section 
represents the first results and discussion chapter, where results are 
presented and interpreted.  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
42 
 
 Characterising Particulate Matter Concentrations  
 
In this first results and discussion chapter, the first objective of the study is 
addressed in full as PM concentrations in KwaDela and KwaZamokuhle are 
characterised. Daily average ambient, indoor and personal PM 
concentrations measured in KwaDela and KwaZamokuhle are compared 
and contrasted with each other across various seasons. Reasons for 
concentration variability within and between the communities are outlined 
and reference is made to existing literature where considered appropriate. 
Hourly average data are shown by means of diurnal concentration patterns. 
Objective two of the study, which seeks to better understand relationships 
between ambient, indoor and personal PM concentration is touched upon 
here, but further elaborated upon in a subsequent chapter.  
 
4.1 Spatial and temporal variability of particulate matter 
 
 
Ambient, indoor and personal air quality in KwaDela and KwaZamokuhle was confirmed to be 
unacceptably poor, particularly in the colder months of the year. NAAQS exceedances were found in both 
communities, but overall air pollution levels in KwaZamokuhle were higher than in KwaDela, and this in 
summer and in winter months.   
 
4.1.1 Ambient particulate matter concentrations  
 
Ambient daily average winter PM10 concentrations were found to be twice as high as summer 
concentrations in KwaDela and KwaZamokuhle, respectively and 1.7 times higher in winter than in spring 
in KwaZamokuhle (Table 4.1 and Figure 4.1). Similarly, ambient daily average winter PM2.5 
concentrations were found to be on average 2.1 and 1.7 times higher than summer concentrations in 
KwaDela and KwaZamokuhle, respectively and 2.1 times higher than spring concentrations in 
KwaZamokuhle (Table 4.2 and Figure 4.2). The variation of daily averages around the respective PM type 
mean values was generally high, however, it was higher in spring than in summer, and highest in winter, 
this trend was apparent in both KwaDela and in KwaZamokuhle. 
43 
 
 
Figure 4.1.  Daily average ambient PM10 concentrations (µg/m³) in KwaDela and KwaZamokuhle across 
the seasons 
 
 
Figure 4.2.  Daily average ambient PM2.5 concentrations (µg/m³) in KwaDela and KwaZamokuhle 
across the seasons 
Mean; Box: Mean±0.95 Conf. Interval; Whisker: Min-Max
 KwaDela
 KwaZamokuhle
Spring Summer Winter 
Season
0
100
200
300
400
500
P
M
1
0
 c
o
n
ce
n
tr
a
tio
n
 (
µ
g
/m
³)
Mean; Box: Mean±0.95 Conf. Interval; Whisker: Min-Max
 KwaDela
 KwaZamokuhle
Spring Summer Winter
Season
0
100
200
300
400
500
P
M
2
.5
 c
o
n
ce
n
tr
a
tio
n
 (
µ
g
/m
³)
44 
 
When comparing PM concentrations between the two communities, daily average PM10 concentrations in 
KwaZamokuhle were found to be on average 1.2 times higher than the concentrations measured in 
KwaDela in both summer and winter (Figure 4.1 and Table 4.1). Summer daily average PM2.5 
concentrations were found to be on average 2.4 times higher in KwaZamokuhle than in KwaDela. In 
winter, PM concentrations with a diameter of less than 2.5 micrometres were found to be on average 1.9 
times higher in KwaZamokuhle than in KwaDela (Figure 4.2).  
 
Table 4.1.  Seasonal characteristics (daily mean±S.D.) of ambient PM10 concentrations (µg/m³) in 
KwaDela and KwaZamokuhle 
KwaDela   Winter 
2013 
Summer 
2014 
Winter 
2014 
Summer 
2015 
Average 
Winter  
Average 
Summer  
AMS 96.7±34.77 51.2±16.8 101.1±28.3 55.1±11.5 99.2±33.2 52.1±15.3 
KD-E2-SW 95.7±79.2 22.4±15.9 47.0±19.1 21.5±11.3 65.9±60.5 22.7±14.4 
Average 99.9±46.3 39.9±20.3 74.7±38.1 44.3±19.9 87.5±47.3 42.9±20.3 
   Spring 
2015 
Summer 
2016 
Winter 
2016 
KwaZamokuhle 61.2±31.0 52.6±23.6 103.6±31.7 
 
Table 4.2.  Seasonal characteristics (daily mean±S.D.) of ambient PM2.5 concentrations (µg/m³) in 
KwaDela and KwaZamokuhle 
KwaDela 
Winter 
2013 
Summer 
2014 
Winter 
2014 
Summer 
2015 
Average 
Winter 
Average 
Summer 
AMS 34.0±11.2 19.1±11.9 29.3±8.7 17.9±3.8 31.4±9.5 18.3±7.8 
KD-E1-NW 40.3±16.5 15.5±8.5 53.7±36.2 18.0±25.7 44.8±32.8 16.7±16.5 
Average 36.9±13.8 18.1±14.6 37.2±19.2 17.1±10.2 36.3±22.2 17.6±12.4 
KwaZamokuhle 
Spring 
2015 
Summer 
2016 
Winter 
2016 
AMS 24.9±12.7 24.4±13.5 55.4±19.4 
KZ-E1-N 33.2±7.5 37.4±13.6 57.9±13.2 
KZ-E2-NW 30.3±8.4 33.7±8.7 NA 
KZ-E3-W 42.3±16.8 38.2±7.9 67.2±19.7 
KZ-E4-SE 38.1±10.7 69.3±28.7 108.7±44.3 
Average 32.6±13.5 42.2±24.8 69.6±32.5 
 
 
When considering compliance with the NAAQS, the daily average ambient measurements for PM10 and 
PM2.5 surpassed the allowed frequency of exceedances of the respective limit value (4) at every monitoring 
site during all winter campaigns in both communities, except for at one site (KD-E2-SW) in KwaDela in 
winter 2014 (Table 4.3 and Table 4.4). In summer, the allowed frequency of exceedances was surpassed 
for PM10 and PM2.5 in KwaDela in summer 2014 at the ambient monitoring station only, and for every 
45 
 
PM type measurement in Summer 2016 in KwaZamokuhle, except for the PM2.5 at KZ-E3-W. In spring 
2015 in KwaZamokuhle, the allowed frequency of exceedances was surpassed at every ambient 
monitoring site except for one (KZ-E4-SE) monitoring site.   
 
On average, 2% and 4% of the PM10 and PM2.5 daily averages in summer and 59% and 34% of the PM10 
and PM2.5 daily averages in winter, respectively were found to be above the NAAQS limit in KwaDela. 
In KwaZamokule, 32%, 13% and 86% of the PM10 daily averages measured in spring, summer and winter, 
respectively were found to lie over the NAAQS limit value. PM2.5 daily averages were found to be above 
the NAAQS limit value 23%, 40% and 91% of the time in spring, summer and winter, respectively.  
 
Ambient air quality in the winter months, particularly in KwaZamokuhle, was characterised by daily 
average PM concentrations that exceeded the NAAQS most of the time. Daily average PM2.5 
concentrations were on average higher in KwaZamokuhle than they were in KwaDela, and this particularly 
in winter. This could be attributed to the prevalence of more source contributors of the smaller size fraction 
particles into the ambient air in KwaZamokuhle, but this is unpacked further into the dissertation.  
 
Table 4.3.  Number of NAAQS limit value exceedances per season per community for PM10 (total 
number of daily averages) 
KwaDela Winter 2013 Summer 2014 Winter 2014 Summer 2015 
AMS  41(50) 6(78) 22(29) 0(29) 
KD-E2-SW  8(12) 0(23) 3(31) 0(25) 
KwaZamokuhle Spring 2015 Summer 2016 Winter 2016 
AMS  23(73) 7(52) 51(59) 
*Daily limit value = 75 µg/m³; Allowed frequency of exceedances = 4 per annum 
 
 
Table 4.4.  Number of NAAQS limit value exceedances per season per community for PM2.5 (total 
number of daily averages) 
KwaDela Winter 2013 Summer 2014 Winter 2014 Summer 2015 
AMS 17(62) 4(46) 8(68) 0(61) 
KD-E2-SW 10(27) 0(55) 25(42) 2(28) 
KwaZamokuhle Spring 2015 Summer 2016 Winter 2016 
AMS 8(84) 8(46) 45(59) 
KZ-E1-N 0(53) 19(55) 46(49) 
KZ-E2-NW 10(63) 6(37) NA 
KZ-E3-W 30(60) 3(7) 41(43) 
KZ-E4-SE 13(35) 45(52) 39(39) 
*Daily limit value = 40 µg/m³; Allowed frequency of exceedances = 4 per annum 
46 
 
The clearly defined seasonal variability of lower ambient PM concentrations in summer and higher PM 
concentrations in winter, as is apparent in both KwaDela and KwaZamokuhle, is influenced by numerous 
factors: 
 
In winter, the Mpumalanga Highveld is characterised by high atmospheric stability, clear skies and low 
wind speeds associated with high pressure systems and circulation is generally anticyclonic (Tyson, 1988).  
In summer, less stable conditions, dominated by easterly circulation patterns with high rainfall and high 
wind speeds, prevail (Fourie et al., 2008).  Considering such macro-scale weather conditions, the dry 
winter climate prevailing on the Highveld leads to increased particle load in the atmosphere and decreased 
atmospheric mixing. A dry winter climate in the Highveld contributes to the load of suspended dust 
particles in the atmosphere in winter in particular (Jury, 2017).  
 
Source apportionment studies that were conducted as part of KwaDela and KwaZamokuhle’s air quality 
offset pilot studies, indicated that soil dust contributed a small fraction of the aerosol loading when 
considering the coarse and fine fraction (6.2% and 5.5%, respectively) in KwaDela (Sasol, 2015). Road 
dust contributed 14% of the coarse fraction aerosol loading in KwaZamokuhle, and road and wind-blown 
dust contributed 12% of the fine fraction aerosol loading in winter (Eskom, 2017). Significantly, in 
KwaZamokuhle, source apportionment results showed that aeolian dust, which represents dust from 
surface sediments and soils, contributed 28% to the coarse fraction of PM in winter. Conversely, in 
climatic regions where summer-rainfall patterns prevail, such as the Mpumalanga Highveld, the 
contribution of such dust particles is expected to be correspondingly lower, as increased humidity and 
rainfall suppresses the entrainment of dust into the atmosphere. Source apportionment results indicated 
that road and wind-blown dust contributed up to 43 % to the course fraction and 16% to the fine fraction 
in summer in KwaZamokuhle. Though this might seem contradictory, it makes sense that wind-blown 
dust contributes greatly to aerosol loading in summer relative to other source contributors, as coal 
combustion, which contributes the most to the aerosol loading in winter, contributes a substantially lower 
fraction.  Observed low wind speeds in winter in KwaZamokuhle in particular, represent a further large-
scale climatic condition that exacerbates high concentrations of locally emitted PM. Winds contribute 
significantly to the dispersion and dilution of emissions, thus when low wind speeds prevail, high pollutant 
concentrations are measured. 
 
47 
 
The South African Highveld is well known for the formation of well-developed inversion layers that 
typically form during the night and persist until later morning (Garstang et al., 1996; Freiman & Tyson, 
2000; Hirsikko et al., 2012). Most of the emissions are released from sources located near the earth’s 
surface up to a few hundred meters. The existence of frequent inversion layers forces the diffusion and 
transport of pollutants to take place in the lower parts of the atmosphere for prolonged periods of time 
causing high PM concentrations (Fourie et al., 2008). In winter in particular, this effect is intensified 
because higher amounts of particles are trapped under the inversion layer and so emissions from domestic 
burning practices remain stuck close to the earth’s surface (Jury, 2017). 
 
Finally, local activities on the ground contribute greatly to the seasonal variability of ambient PM 
concentrations. In KwaDela and KwaZamokuhle the high PM concentrations in winter and the reduced 
PM concentrations in summer are directly proportional to the amount of domestic burning and heating 
activities that take place during those seasons. In KwaDela for instance, it was found that households used 
on average 216 kg of coal for heating and cooking activities during an average winter month compared to 
117 kg of coal in an average summer month (Sasol, 2015).  Consequently, burning increased amounts of 
coal leads to increased suspended particles in the atmosphere in winter months.  
 
Source apportionment data supports the above statements as it was found that the biggest contribution to 
the aerosol loading in both communities in winter was from coal combustion: In KwaZamokuhle, in 
winter, 38% of the course fraction and 52% of the fine fraction could be attributed to coal combustion. 
Similarly, in KwaDela, 56% and 61% of the fine and course fraction, respectively could be attributed to 
coal combustion related sources (Sasol, 2015). This can be strongly contrasted to 15% and 3% contribution 
to the course and fine fraction in summer in KwaZamokule (Eskom, 2017). 
 
The prevalence of spring data in KwaZamokuhle allows for interesting seasonal transition observations, 
where the changeover from winter to summer becomes clear and predominantly cold days slowly give 
way to, on average, warmer temperatures. Warmer days presumably lead to less frequent and prolonged 
heating and cooking activities at a household-level, and with that, lower PM concentrations are evident.  
 
KwaDela and KwaZamokuhle lie in relative close proximity to each other geographically and are thus 
governed by the same macro-scale meteorological cycles described above.  Therefore, starkly contrasting 
48 
 
differences in air pollution levels between the communities can more than likely be attributed to local-
scale phenomena, more specifically to anthropogenic sources.  It should not be discounted however, 
because data collected for the different communities were not collected in the same years, that though one 
would expect similar large-scale seasonal meteorological trends to remain consistent, it could be the case 
that winter 2013/14 (KwaDela) and winter 2016 (KwaZamokuhle) were in fact characterised by different 
local meteorological conditions. Consequently, it was noted that stronger westerly waves and more severe 
cold front conditions, contributed to the observed increased PM concentrations in KwaZamokuhle. 
Considering the average meteorological data (Table 3.2 and Table 3.3), average minimum temperatures 
in KwaZamokuhle in winter 2016 were lower than those measured in winter 2013 and 2014 in KwaDela, 
which could point to an overall perceived colder winter in 2016 and corresponding increased solid fuel 
burning activities for heating purposes. 
 
KwaZamokuhle is a larger community when compared to KwaDela, with a population about five (5) times 
higher. A logical conclusion to draw from this, would be that a larger number of households results in a 
proportionally larger number of coal burning households. However, taking into account the socio-
economic complexities that persist in different communities, it is inappropriate to draw over-simplistic 
conclusions. 
 
According to Statistics South Africa (Stats SA, 2012b), proportionately, a higher percentage of households 
make use of coal for cooking and heating activities in KwaDela than in KwaZamokuhle (Table 3.1). When 
delving further into demographic specifics, one can see that a higher proportion of households is electrified 
in KwaDela (97%) than in KwaZamokuhle (89%).  It is often assumed that access to electricity and the 
consequent use of electrical appliances on the one hand and the use of coal for heating and cooking 
purposes on the other hand are mutually exclusive (Stats SA, 2012a). However, high unemployment rates 
and low incomes force households to rely on cheaper, dirtier fuels, and so, though houses may be 
electrified, coal is still relied on as a main domestic energy source (Stats SA, 2012a).  There are more 
people who live below the poverty line in KwaZamokuhle than in KwaDela, pointing to a higher 
proportion of people who would have to rely on coal for cooking and heating as it is the cheaper fuel 
option, even if electricity is available (Stats SA, 2012b).  
 
49 
 
4.1.2 Spatial and seasonal trends of indoor and personal particulate matter concentrations within and 
between two communities 
 
When comparing the overall daily mean indoor PM concentrations across the seasons within each 
community, winter PM4 concentrations in the indoor environment were shown to be on average 1.7 and 
3.0 times higher in winter than in summer in KwaDela and KwaZamokuhle, respectively (Figure 4.3 and 
Table 4.5). Personal PM4 concentrations were on average 1.9 and 3.1 times higher in winter than they 
were in summer in KwaDela and KwaZamokuhle (Figure 4.4 and Table 4.5). The variation of daily 
averages around the mean values was higher in spring than in summer, and highest in winter; it is notable 
that the standard deviation around the arithmetic mean was on average higher in the indoor environment 
than it was in the ambient environment. Speculatively. this high variability could be attributed to (i) the 
number of burning activities that take place in a specific household on a given day; (ii) the number of 
times and the way in which the stove is refuelled; (iii) the different types of stoves that are used, and even 
(iv) how these factors change across the seasons.  
 
 
Figure 4.3.  Daily average indoor PM4 concentrations (µg/m³) in KwaDela and KwaZamokuhle across 
the seasons 
Mean; Box: Mean±0.95 Conf. Interval; Whisker: Min-Max
 KwaDela
 KwaZamokuhle
Spring Summer Winter 
Season
0
200
400
600
800
1000
1200
1400
1600
1800
2000
P
M
4
 c
o
n
ce
n
tr
a
tio
n
 (
µ
g
/m
³)
50 
 
 
 
Figure 4.4.  Daily average personal PM4 concentrations (µg/m³) in KwaDela and KwaZamokuhle across 
the seasons 
 
On average, indoor PM4 concentrations in KwaDela were 1.5 times the concentrations measured in 
KwaZamokuhle in summer and the concentrations in KwaZamokuhle were 1.2 times the concentrations 
measured in KwaDela in winter. Personal PM4 concentrations in KwaZamokuhle, were on average 0.7 
and 2.3 times the concentrations measured in KwaDela in summer and in winter (Table 4.5).  
 
Table 4.5.  Seasonal characteristics (daily mean±S.D.) of indoor and personal PM concentrations 
(µg/m³) in KwaDela and KwaZamokuhle 
KwaDela 
PM 
Type 
Winter 
2013 
Summer 
2014 
Winter 
2014 
Summer 
2015 
Average 
Winter 
Average 
Summer 
Indoor PM4 183.6±144.9 62.4±65.2 149.1±96.0 138.2±89.4 166.0±137.4 99.3±99.7 
Personal PM4 92.0±97.1 27.4±17.5 47.2±23.1 47.7±21.9 69.6±106.1 37.7±178.4 
KwaZa-
mokuhle 
PM 
Type 
Spring 2015 Summer 2016 Winter 2016 
Indoor PM4 78.5±84.4 65.1±50.7 192.9±181.2 
Personal PM4 58.3±68.3 52.9±44.6 161.1±103.7 
 
Mean; Box: Mean±0.95 Conf. Interval; Whisker: Min-Max
 KwaDela
 KwaZamokuhle
Spring Summer Winter
Season
0
200
400
600
800
1000
1200
1400
1600
1800
2000
P
M
4
 c
o
n
ce
n
tr
a
tio
n
 (
µ
g
/m
³)
51 
 
Though indoor PM guideline limits are soon to be published, no regulatory indoor or personal air quality 
standards exist in the Republic of South Africa. There is also no standard which limits particles with a 
diameter that is generally four (4) micrometres and smaller. For this reason, daily average indoor and 
personal PM4 concentrations, as measured in the sampling campaigns used for this study, cannot be 
directly compared to a limit value to ascertain whether or not PM concentrations in the indoor environment 
are compliant with prescriptive legislation.  Nonetheless, PM with a diameter of <2.5 µm represents a 
subset of PM with a diameter of four (4) µm, which in turn, represents a subset of PM with a diameter of 
ten (10) µm. Measured daily averages will thus be roughly contextualised by considering the daily PM10 
NAAQS, as this would represent a conservative assessment of the exceedances of the PM standards. 
 
In KwaZamokuhle, indoor and personal daily average PM4 concentrations were found to lie above the 
PM10 NAAQS of 75µg/m
3 39%, 28% and 85% of the time in spring, summer and winter, respectively. 
Personal PM4 daily average concentrations were above the PM10 limit 27%, 22% and 79% of the time in 
spring, summer and winter. In KwaDela, personal and indoor daily average PM4 concentrations were 
above the PM10 limit value 35% and 75% of the time in winter and 11% and 45% of the time in summer.  
 
Indoor and personal PM concentrations measured during the KwaDela and KwaZamokuhle sampling 
campaigns were highly variable in space and time and most notably, measured concentrations even 
reached into the mg/m3 range. Complex larger and smaller scale seasonal factors that trigger the variability 
of these concentrations would mostly be related to what negatively influences the thermal comfort within 
a household, such as low ambient temperatures and high precipitation- and wind speeds. Other large-scale 
temporal factors influencing this variability can be attributed to the time of day, as this greatly governs 
burning times in a home (discussed in more detail in Chapter 5 below). More local-scale, socio-economic 
factors that can contribute to the inhomogeneous temporal and spatial distribution of anthropogenic 
emissions of PM concentrations in different households are:  income levels, energy-/ fuel type and 
availability, how this fuel is burnt and what stove it is used, cultural preferences, social class, the number 
of the household inhabitants, location of the household itself  and the structure and the integrity of the 
house (Scorgie et al., 2003; Kimemia & Annegarn, 2011; Kimemia et al., 2016; Masekameni, 2017). This 
list is not exhaustive but does give an indication of the complexity of the problem.  
 
52 
 
A study conducted to better understand what contributing factors govern the variability of domestic 
burning habits on the South African Highveld, with a particular focus on KwaDela, showed that about 
98% of the individuals who participated in the fuel use observation in KwaDela use coal, wood, dung and 
paper in a single burning event, and not all did this in the same manner (Nkosi et al., 2017). It was also 
shown that the appliances used to burn the coal varied tremendously in shape, form, quality and age, some 
stoves being as old as 25 years. The majority of the stoves used were broken and leaks were also apparent, 
allowing smoke to escape from the combustion chamber and to enter the household. Another way in which 
smoke enters the house is when the chimney attached to the stove is either not fully extended through the 
roof, or is not structurally intact (Nkosi et al., 2017). The area within which coal burning takes place 
differs from household to household: many stoves are located within the kitchen of the household; 
however, some are also located in the living room, or in a different room entirely (sometimes this room 
would be an external extension to an existing house structure) (Nkosi et al., 2017). Burning practices in 
low-income communities in South Africa are mostly focused on cooking activities in the warmer months 
and include heating activities in the colder months. Coal is preferred, even in households that have access 
to an electricity supply due to its availability and its affordability, and due to its multi-functional nature 
(supports cooking, heating and lighting functions) (Scorgie et al., 2003).   
 
High indoor PM4 concentrations related to burning activities (not taking into account other sources of PM 
in the indoor environment, such as smoking) are mostly a function of combustion activities that take place 
directly inside the household but can also be attributed to infiltration of external emissions stemming from 
the ambient environment. How much air pollution infiltrates into a household from the ambient 
environment (stemming from burning and other non-burning related sources) depends on the proximity 
of the external source to the household, the prevailing weather conditions, the structural intactness of the 
household in question (i.e. how airtight it is), and whether or not windows and doors are open to let in the 
external emissions (Meng et al., 2005). Outdoor particles can enter indoor environments by convective 
flow (e.g. through an open window) or by diffusional flow (i.e. infiltration) and through cracks and fissures 
in the barrier of the building envelope (Meng et al., 2005; Cao et al., 2012). Contributions of outdoor 
sources to indoor PM2.5 concentrations have been estimated to range between 23–67%  in low-income 
communities (Meng et al., 2005; Cao et al., 2012). This has not been estimated in an average South African 
low-income household.  
 
53 
 
Though it is easy to say that the overall variability in indoor PM concentrations measured in KwaDela and 
KwaZamokuhle is influenced by a combination of a wide range of large-scale and local-scale factors as 
described above, it is only possible to really understand what influences indoor and personal exposure to 
PM concentrations if one delves into specific households individually and tries to better understand the 
specific parameters that could influence the air quality that is breathed in that specific household itself. 
This approach is touched upon in Chapter 5 below.  
 
The variability of personal PM concentrations, i.e. the concentrations measured by the SidePak device 
worn by a select number of individuals within KwaDela and KwaZamokuhle for short periods of time is 
even more complex to unravel as these concentrations are measured in various locations as the individual 
wearing the monitor moves through space and time and sources to which an individual is exposed are 
present in the indoor and the ambient environment.  Generally spoken, personal PM4 concentrations 
measured across these study campaigns will have been influenced by the same macro- and micro- scale 
factors that have been outlined to influence ambient and indoor PM concentrations in the paragraphs 
above, as they essentially represent an amalgamation of ambient and non-ambient components (Wilson et 
al., 2000). This means that, a person exposed to particulates stemming from indoor and ambient sources 
in winter months, is bound to be exposed to higher average PM concentrations regardless of which micro-
environment the individual finds himself/herself in, than the same individual would be exposed to in 
summer months in the same micro-environments.  
 
A one-individual case study outlined in Chapter 5 of this thesis expands on the complexities of personal 
PM exposure by considering a one-individual case study in KwaDela. 
 
4.1.3 Indoor and personal hourly maximum particulate matter concentrations  
 
Average hourly maximum winter PM4 concentrations in the indoor environment were on average 1.1 and 
3.2 times higher in winter than in summer in KwaDela and KwaZamokuhle, respectively. Personal average 
hourly maximum PM4 concentrations were on average 2.2 and 2.1 times higher in winter than they were 
in summer in KwaDela and KwaZamokuhle (Figure 4.5 and Figure 4.6).  
 
54 
 
In KwaZamokuhle, average winter indoor hourly maximum PM4 concentrations were 2.3 times higher 
than in spring and average winter personal maximum PM4 concentrations were 5.2 times higher than in 
spring. For the sake of this comparison, the highest hourly average value for each day was ascertained and 
then averaged for the days for which measurements were conducted. 
 
Indoor and personal hourly maximum PM4 concentrations are important indicators of extreme exposure 
events (Table 4.6). People spend most of their time in the indoor environment (particularly in the colder 
months), and consequently the health of the individuals in the indoor environment is likely to be affected 
by such acute exposure events during fire ignition activities for instance (Smith & Mehta, 2003).   People 
living in KwaDela and KwaZamokuhle are exposed to high levels of PM concentrations at both longer 
and shorter durations resulting in increased risk of both acute (short-term) and chronic (long-term) health 
effects (Bundi, 2004).   
 
While it is incredibly useful to compare averaged PM concentrations for potential compliance against 
longer-term air pollution standards, or to assess chronic exposures, this does not offer a reliable means of 
assessing short-term, extreme concentrations and acute exposures, and hence maximum values need to be 
considered (Briggs et al., 2000). Highest indoor and personal exposure events are most likely to be 
experienced by those individuals who use coal stoves in poorly ventilated environments, or who are in 
close proximity to them (typically women and children) (Chen & Smith, 1991; Fullerton et al., 2008).  
Personal exposure events however are not limited to the indoor environment and can also be experienced 
in the ambient environment, though these maximum exposure events are less likely to be as extreme as 
those experienced indoors, as particles are more likely to diffuse. This becomes clear when considering 
the differences between the average maximum concentrations measured in the indoor environment and 
the average maximum concentrations measured on the personal monitors (Figure 4.5 and Figure 4.6). The 
general trend of higher average concentrations in both KwaDela and KwaZamokuhle (even of maximum 
nature) in the winter months is recorded here.  
 
55 
 
 
Figure 4.5.  Average hourly maximum indoor PM4 concentrations (µg/m³)  in KwaDela and 
KwaZamokuhle across the seasons 
 
 
Figure 4.6.  Average hourly maximum personal PM4 concentrations (µg/m³) in KwaDela and 
KwaZamokuhle across the seasons 
Mean; Whisker: Min-Max
 KwaDela
 KwaZamokuhle
Spring Summer Winter
Season
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
5500
6000
6500
7000
P
M
4
 c
o
n
ce
n
tr
a
tio
n
 (
µ
g
/m
³)
 
Mean; Whisker: Min-Max
 KwaDela
 KwaZamokuhle
Spring Summer Winter
Season
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
5500
6000
6500
7000
P
M
4
 c
o
n
ce
n
tr
a
tio
n
 (
µ
g
/m
³)
 
56 
 
Indoor and personal measured hourly maximum PM4 concentrations were found to be on average 1.9 and 
3.9 times higher in KwaZamokuhle than in KwaDela in the winter season. Indoor summer average 
maximum PM4 concentrations in KwaDela were 1.5 times the concentrations measured in 
KwaZamokuhle. Maximum personal concentrations measured in the summer campaigns were found to be 
on average 3.2 times higher in KwaZamokuhle than in KwaDela.  
 
Table 4.6.  Seasonal characteristics (hourly max±S.D.) of indoor and personal PM concentrations 
(µg/m³) in KwaDela and KwaZamokuhle 
KwaDela 
PM 
Type 
Winter 
2013 
Summer 
2014 
Winter 
2014 
Summer 
2015 
Average 
Winter 
Average 
Summer 
Indoor PM4 622.8±731.1 432.7±378.5 675.9±619.2 703.9±902.8 649.4±37.5 568.0±191.3 
Personal PM4 224.3±286.4 63.1±61.1 76.0±71.7 72.6±94.3 150.2±104.9 67.8±6.7 
KwaZa-
mokuhle 
PM 
Type 
Spring 2015 Winter 2016 Summer 2016 
Indoor PM4 551.2±956.6 1261.1±1773.1 391.2±369.0 
Personal PM4 113.6±96.7 595.3±317.8 220.8±202.2 
      
 
4.1.4 Diurnal patterns of ambient, indoor and personal particulate matter concentrations  
 
 
A pronounced bimodal diurnal pattern was identified when plotting overall hourly average time-series 
graphs for ambient PM measurements in both KwaDela and KwaZamokuhle (Figure 4.7 and Figure 4.8). 
Obvious concentration peaks were evident between 5-9 am in the morning and again between 5-9 pm in 
the late afternoon/ early evening. In some instances, random hourly average peak concentrations were 
measured during the day (for instance for PM10 at KD-E2-SW in KwaDela in winter 2013). Midday peaks 
were also measured at some of the ambient monitoring sites in winter 2013, winter 2014 and summer 2015 
in KwaDela and in summer and winter 2016 in KwaZamokuhle.  
Pronounced seasonal differences of ambient hourly average concentrations were observed in both 
communities with the highest average concentrations typically measured in winter (average concentrations 
ranged between 30 – 300 µg/m³) and the lowest concentrations generally measured in the summer (ranging 
between 15 – 100 µg/m³). Winter hourly average concentrations were observed to be higher than summer 
concentrations in both considered communities, showing correspondence once again between increased 
suspended particle loads in the atmosphere and larger scale climatic conditions such as low temperatures 
57 
 
and low wind speeds and inversion layers as well as local scale conditions such as increased low-level 
burning practices at a household level contributing to this difference. Notably, the time of day affects the 
ambient concentrations that are measured; as such, peak concentrations correspond directly with peak 
burning times (Oglesby et al., 2000; Huang et al., 2015; Wang et al., 2015). Additionally, PM 
concentrations are higher when local atmospheric conditions are most stable (i.e. the mornings and the 
evenings) as concentrations remain largely undiluted. Concentrations are lower when the local atmosphere 
is unstable and atmospheric mixing takes place and concentrations are diluted (during the day) (Ali et al., 
2015).  
Surveys that considered burning times during the campaigns in KwaZamokuhle support previous findings 
that the first peak burning times in the day typically occurred in the early mornings, when individuals 
commenced their morning routines to start the day, which include cooking and heating water for bathing 
purposes and space heating in winter (Eskom, 2017; Nkosi, 2017). Daytime burning activities take place 
for cleaning purposes and occasionally for cooking.  Late afternoon and early evening fires are primarily 
made for cooking-, but also for space heating-, bathing- and socialising purposes and can be made indoors 
and outside. Late evening fires are made again (including rekindling of existing fires) mostly for space 
heating, particularly in winter (Ali et al., 2015; Bartington et al., 2017). The highest and most well-defined 
peaks of PM are typically experienced in the mornings and the evenings.  Sporadic peaks do occur during 
the day but are comparatively low. Hourly average PM2.5 concentrations, though following a similar 
diurnal trend to the hourly average PM10 concentrations, typically plotted at lower average concentration 
values (with the exception of the occasional concentration peak); this makes sense as PM2.5 represents a 
subset of PM10 concentrations. The different PM2.5 concentrations measured at the various E-Samplers 
and E-Bams across the communities are a testimony to the spatial distribution of the sampling sites and 
the high variability of PM concentrations in the ambient environment across the communities. The fact 
that PM2.5 and PM10 concentrations do not always follow the exact same diurnal trend can be attributed to 
different source contributions to that specific size fraction during that time.  
 
 
 
58 
 
       
        
 
 
Figure 4.7.  Ambient particulate matter diurnal trends (µg/m³) per season in KwaDela 
 
 
      
 
59 
 
 
 
 
Figure 4.8.  Ambient particulate matter diurnal variation (µg/m³) per season in KwaZamokuhle 
 
Similar to the ambient diurnal variation patterns, the indoor PM4 concentrations showed two distinct 
peaks, one corresponding to the early hours of the morning (between 4 - 10am) and one corresponding to 
the early evening hours (typically starting from 5pm, but sometimes starting from as early as 3pm – 9pm), 
this temporal variation is most attributable to the different burning times in different households (Nkosi et 
al., 2017) (Figure 4.9 and Figure 4.10). It should be pointed out here that there is a large variability in the 
burning times at a household level: it is likely that no two houses display the exact same burning patterns 
in terms of time, duration and even fuel type used. Notably, indoor hourly average concentrations were on 
average higher than those measured in the ambient environment with hourly average values reaching up 
to 700 µg/m3 in winter.   
 
Diurnal variations for indoor and personal PM4 concentrations were plotted for each sampling campaign 
for each community (Figure 4.9 and Figure 4.10).  Personal measurements taken during the KwaDela 
sampling campaigns were typically only available during the day hours, leading to gaps in the data during 
peak hours, whereas personal PM4 measurements were available for all hours of the day in 
KwaZamokuhle. Highest hourly personal concentrations were up to twice as high in winter than in summer 
in both communities, reaching concentrations of up to 800µg/m³.  
 
Congruence between indoor and personal hourly average concentrations was evident across all seasons in 
both communities and more so in KwaZamokuhle than in KwaDela in the winter months.  
 
60 
 
    
 
     
Figure 4.9.  Indoor and personal articulate matter diurnal trends (µg/m³) per season in KwaDela 
 
 
 
 
61 
 
 
Figure 4.10.  Indoor and personal particulate matter diurnal trends (µg/m³) per season in KwaZamokuhle 
 
It should be pointed out that evening concentrations are observed to be the highest peaks measured for 
ambient and indoor concentrations. This same trend was mostly apparent when considering the personal 
measurements, though personal PM measurements in winter 2014 in KwaDela and spring 2015 in 
KwaZamokuhle did not always show the same peaks. This could be attributed to individuals being in close 
proximity to heat sources besides the ones within their homes. The fact that evenings represent times 
during which people return from work and burn coal to cook, heat, bath and socialise makes this evening 
peak plausible. Increased traffic activity on dusty roads should not be discounted as a meaningful 
contributor to peak PM concentrations in the mornings, when people go to work, and in the afternoon/ 
evenings when they return (Lozano et al., 2007).  
 
4.1.5 Spatial variability of hourly maximum particulate matter concentrations per case per 
community 
 
In KwaDela and KwaZamokuhle, maximum hourly concentrations measured at a case level were 
generally higher in winter than they were in summer; this applies to all PM types measured (Figure 7.1 - 
Figure 7.7 in Annexure A). Indoor PM4 concentrations fairly consistently showed a larger range of hourly 
maximum concentrations than did ambient concentrations, with a number of extreme values.  
 
At a glance, a high variability of hourly maximum values between the cases is evident, pointing to the 
contribution that spatial factors play in determining the variability of PM concentrations across a 
community. Highest hourly maximum concentrations identified for the different PM types measured 
62 
 
during a specific campaign did not necessarily all correspond to one and the same case, meaning that 
maximum concentrations were not always measured in the same location across the seasons.  As such, in 
winter 2013 and winter 2014, highest maximum hourly indoor concentrations were measured in different 
houses (houses two (2) and 14 and in houses 19 and 14, respectively, for example).  
 
Hourly maximum indoor and personal PM4 concentrations measured across the campaigns are more often 
than not measured in the mg/m3 range. This is not the case for most ambient PM10 and PM2.5 
concentrations. Some ambient PM2.5 measurements taken at KZ-E4-SE in KwaZamokuhle show 
concentrations also reaching up into the mg/m3 ranges (For case 4 in spring 2015, for cases 12,4,5,6 and 
9 in summer 2016 and cases 1,2,4,6,8 and 9 in winter 2016). This might mean that there is a burning 
source located next to the E-Bam monitor, contributing to direct local suspended particle concentrations.  
 
Most notably, it was observed from this exercise that, in some houses in KwaDela (house 3 in winter 2013 
and summer 2015, house 15 in winter 2014 and summer 2014 as well as house 18 in summer 2014, winter 
2014 and summer 2015) and in some houses in KwaZamokuhle (house 2 in summer 2016 and winter 
2016, house 5 in spring 2015, summer 2016 and winter 2016 as well as house 6 in summer 2016 and 
winter 2016 and house 7 in winter 2016), though little to no coal burning activities were taking place 
internally, indoor PM concentrations showed similarly high concentrations as those houses in which 
burning did take place.  
 
Additionally, it was noted that these houses showed higher concentrations than those concentrations 
measured in the ambient environment. This is most likely a sign that air pollution from a near-by, 
independent ambient source has filtrated into the indoor environment by entering through either open 
doors or windows or through cracks and fissures in the housing structure itself, raising air pollution levels 
in the household (Meng et al., 2005; Hoek et al., 2008). The magnitude of the high concentration events 
indicates that it is likely that these households must have been located in close proximity to other 
independent PM sources, speculatively, such sources could be external waste burning events, nearby social 
fires, veld fires or in the mornings, traffic related pollution.  This means that, regardless of the source of 
energy that is utilised inside a household, the likelihood of the air quality in the indoor environment 
exceeding the NAAQS is high, and this in the summer as well as the winter months.  
 
63 
 
Interestingly, cases even exist where maximum values are higher in the ambient environment than in the 
indoor environment (case 13 in KwaDela in winter 2013 for example). This points to the fact that it is 
possible to have indoor PM concentrations that are lower than those that govern the ambient environment.  
 
The results highlighted in this section are in line with previous studies that have examined extreme peak 
exposure events in the indoor environment in households which rely heavily on solid fuels. Rajagopalan 
and Brook (2012) reported peak exposures of  > 30 000 µg/m3 during periods of cooking. Hourly 
maximum concentrations measured in House 8 in KwaZamokuhle in winter 2016 went up to 32 000 µg/m3 
(Figure 7.7). Significantly, findings of this research also support findings made in other studies, where 
contributions of outdoor sources to indoor PM concentrations can be very high (Meng et al., 2005).  
 
Another important finding that is made here is that maximum PM concentration of all size fractions 
measured (PM10, PM2.5 and PM4) are highly variable in space and time in both KwaDela and 
KwaZamokuhle. Though variability is visible for ambient maximum PM concentrations, it is more evident 
for personal measurements and particularly so for indoor measurements. Whilst considering each case on 
an individual basis, it is clear that indoor and personal maxima are on average higher than ambient 
maxima, the difference between the average maximum concentrations measured across the houses varies 
significantly, with some houses showing maximum hourly values within the range of ambient maximum 
concentrations, and others showing maximum concentrations that reach over 10 000 µg/m3 in the winter 
season. This bears testimony to the fact that there is stark intra-urban contrast between measured maximum 
PM concentrations at an ambient, indoor and personal PM nature. 
 
A detailed study conducted by a PhD student at the University of Canterbury sought to challenge the 
widely-held assumption that PM concentrations within an urban setting are spatially uniform, particularly 
so in a city which is affected by high seasonal variations of PM concentrations (Wilson, 2006). Wilson 
(2006), along with many other researchers, states that most epidemiological studies that statistically 
associate PM air pollution exposures with health outcomes make this assumption, and by inference, 
assume that, within a given city, exposures to PM concentrations are also without significant variability, 
which allows for the use of compliance status with ambient standards as a proxy for overall community 
health status (Oglesby et al., 2000; Jantunen, 2007; Diapouli et al., 2011). It is further stated that it is of 
critical importance to any environmental epidemiological work to determine an accurate exposure estimate 
64 
 
for a considered population or indeed, a considered individual (Chow et al., 2006). It is suggested that 
there may be greater variation within urban areas than previously reported (Briggs et al., 1997).  
 
What this chapter emphasises is that, were this study based solely on PM measurements taken at a central 
ambient monitoring in both KwaDela and KwaZamokuhle, the true magnitude and the true variability of 
PM concentrations that is present in these communities in the various micro-environments would not come 
to the fore. It further emphasises that, even though KwaDela and KwaZamokuhle both represent low-
income communities which lie in relative close geographical proximity to each other, and though they are 
governed by similar socio-economic demographics and similar macro-scale meteorological patterns, it is 
not possible to assume that these communities would have the same air quality, though air quality in both 
areas can be considered extremely poor.  
 
A necessary and bold assumption of using a single monitoring site as a proxy for community-wide 
exposure, is that PM2.5 concentrations are homogeneously distributed and that the concentrations between 
sites are well correlated in space, magnitude and time (Pinto et al., 2004; Wilson, 2006). The next section 
of this thesis aims to better understand correlations between PM measurements recorded at different 
locations in both KwaDela and KwaZamokuhle. 
 
 
******************* 
 
This chapter outlined results and discussions unpacking the first objective 
of the study which sought to “characterise ambient, indoor and personal 
PM concentrations in KwaDela and KwaZamokuhle”. This chapter also 
attempted to address, in part, the second objective of this study “Determine 
the relationships between ambient, indoor and personal PM concentrations 
on KwaDela and KwaZamokuhle, Mpumalanga”. The next chapter aims 
to unpack objective two in more detail. Additionally, chapter 5 addresses 
objective three: “Assess personal total exposure to PM in KwaDela”. 
 
65 
 
 Relationships between and personal exposure to particulate 
matter concentrations  
 
This chapter presents results from two households selected in both KwaDela 
and KwaZamokuhle to explore the relationship between ambient, indoor and 
personal PM concentrations in individual households and the possible 
determining factors of these relationships. Additionally, results of case-level 
and community-level regression analyses are presented and discussed. 
Finally, a case study of a single household is discussed to assess personal 
exposure to PM concentrations.  
 
5.1 Household-level PM characteristics 
 
Though socio-economic factors have been taken into account in interpreting the overarching PM 
concentration variability at a community level in the sections above, it is useful to consider individual 
households to better understand how people’s day-to-day habits can influence PM concentrations 
measured directly in and around the house and how these relate to total personal exposure levels.  
 
Two houses for each of the sampling campaigns have been selected to illustrate the contribution of 
household-level factors to PM concentrations. Houses 3 and 11 in KwaDela and houses 3 and 9 in 
KwaZamokuhle were selected, as these were the houses for which case-level data (i.e. measurements for 
each of the PM types (ambient, indoor and personal PM)) were available simultaneously for three 
campaigns. This allowed for a good comparison of data across the seasons. Survey data, as collected 
during these respective campaigns has been used to better understand contributing factors to air pollution 
levels in and around these households (Table 5.1 and Table 5.2).  
 
 
 
 
 
 
 
66 
 
Table 5.1.  Overview of relevant survey results for chosen households in KwaDela 
  House 3KD House 11KD 
House type Formal (brick) Formal (brick and corrugated iron, 
but not extended) 
Stove type No coal stove Welded (Union) 
Energy sources used Electricity Electricity, coal, wood and dung 
Lighting Electricity Electricity 
Use of coal for heating No Yes 
Use of coal for cooking No Yes 
Other uses for coal  NA Heating water for bathing and 
washing dishes in winter, ironing 
Ignition times in winter NA 5pm – 9 pm 
Ignition times in summer NA 5pm 
Purchase coal NA As needed 
Smoker   No  Yes. One individual smokes, but it 
is not known whether he/she 
smokes indoors 
 
Table 5.2.  Overview of relevant survey results for chosen households in KwaZamokuhle 
  House 3KZ House 9KZ 
House type Formal (brick) with informal house 
extension 
Formal (brick) 
Stove type Cast iron stove Welded stove 
Energy sources used Coal and electricity Coal and electricity 
Lighting Electricity Electricity 
Use of coal for heating In winter only In winter only 
Use of coal for cooking In winter and summer In winter and summer 
Other uses for coal  Bathing in winter, washing dishes in winter Bathing in summer and winter and 
socialising 
Ignition times in winter 6 am and 4 pm 6 am and 4 pm 
Ignition times in summer 7 am 5 am 
Purchase coal As needed @ small bag of 25kg Monthly @ small bag of 25kg 
Smoker Yes. Two individuals smoke, but it is not 
known whether they smoke indoors 
No 
 
 
 
 
 
 
 
67 
 
5.1.1 Characterising PM concentrations at a household level 
 
 
Daily average PM concentrations of all PM type measurements in KwaDela for cases 3KD and 11KD 
were on average higher in winter than in summer (Figure 5.1). Though indoor PM4 daily average 
concentrations were higher in the winter time than in the summer time, the 75th percentile value of the 
indoor PM4 daily average values was higher in the summer months than in the winter months and 
maximum daily average concentrations measured in summer were comparable to those measured in 
winter. Overall, indoor PM4 concentrations measured for case 11KD were higher than those measured for 
case 3KD. This applies to all seasons. Median personal daily average values did not correspond with 
median indoor daily average values at a case level.  Ambient daily average values measured were similar 
when comparing both cases for each individual season, and highest, overall, in winter. 
 
Lowest daily average concentrations were observed for both cases in KwaZamokuhle in summer (Figure 
5.2). Highest ambient median concentrations were evident for PM2.5 measurements taken at the KZ-E4-
SE monitoring site and the PM10 measurements taken at the main AMS in summer. In spring, daily average 
indoor and personal PM4 concentrations were on average, higher than those measured in summer, 
however, highest median concentrations were measured for both cases in winter. Indoor and personal 
median daily average concentrations corresponded well across the sampling campaigns, but most 
obviously so in winter. Winter concentrations were more variable than spring concentrations and spring 
concentrations were more variable than summer concentrations.  
 
68 
 
 
Figure 5.1.  Comparing ambient, indoor and personal daily average PM concentrations (µg/m³) 
measured per season for case 3KD and case 11KD in KwaDela 
69 
 
 
Figure 5.2.  Comparing ambient, indoor and personal daily average PM concentrations 
(µg/m³) measured per season for case 3KZ and case 9KZ in KwaZamokuhle 
 
 
In KwaDela, collected survey data indicated that House 3KD does not use coal as its main energy source, 
not even for cooking and heating activities (Table 5.1). In stark contrast, house 11KD relies on a range of 
sources to cover its energy needs (electricity is used for lighting, wood, coal and dung are used for heating, 
cooking, ironing, heating water for bathing and washing dishes). This background information can help 
contextualise the measured data, and thus it makes sense that indoor and personal PM measurements taken 
in house 11KD would be higher than those measured in house 3KD, as more indoor air pollution is being 
created through the solid fuel combustion processes in the indoor environment. The fact that house 11KD 
also makes mention of corrugated iron as a structural element of the house, could also mean that infiltration 
of ambient air into the indoor environment could play a large role in defining indoor air quality due to 
70 
 
unsealed sections of the house structure. Significantly, though house 3KD does not make use of dirty fuels 
inside the house itself, high PM concentrations are nonetheless measured in the inside environment during 
peak burning times, so much so, that indoor measurements in house 3KD are sometimes measured to be 
higher than those measured in house 11KD.  High indoor PM concentrations in house 3KD could be 
attributed to smoke, from a neighbouring house or a fire close by, infiltrating directly into the house. An 
unaccounted for indoor source, such as an individual’s smoking habits could also contribute to higher PM 
concentrations indoors.  When considering the social survey data however, it is noted that there is no 
smoker in house 3KD, supporting the case that infiltration of ambient PM into the indoor environment is 
more than likely the main contributory cause for high PM emissions in house 3KD. The fact that there is 
a smoker in house 11KD is another reason for high indoor PM concentrations.  
 
Survey data for the two houses showcased for KwaZamokuhle (Table 5.2), implies that the indoor PM 
monitor in house 9KZ would read, on average, higher PM concentrations than the monitor in house 3KZ. 
This assumption is based on the fact that the household member in house 9KZ expressed a preference for 
using coal over electricity for heating and cooking purposes, and additionally, the household burns coal 
to heat water for bathing purposes in winter months and summer months as well as for social gatherings 
(whereas this is not the case in house 3KZ). Additionally, the coal stove used in house 9KZ is welded, as 
opposed to the cast iron stove used in house 3KZ. A welded stove could presumably be more prone to 
leaks and therefore would possibly allow more smoke to escape from the burning chamber directly into 
the household, instead of from the chimney into the ambient environment, as would be the case for a cast 
iron stove.  
 
The fact that measured PM concentrations in house 3KZ however, are in fact higher than those measured 
in house 9KZ, is a perfect illustration of how complex it is to understand indoor air pollution.    There are 
many other factors that could influence the indoor PM concentrations in these households, over and above 
those listed in the table above. These could explain the, on average, higher concentrations in house 3KZ; 
open doors or windows in house 9KZ during peak burning times could lead to lower indoor PM 
concentrations, for instance, through dispersion of pollutants into the ambient environment.  It should be 
factored into this interpretation that house 3KZ has an informal extension. It is possible that the extension 
is where burning activities take place. Smoke from these burning activities could easily infiltrate into the 
main brick building, through cracks or open windows/ doors increasing indoor PM concentrations there. 
71 
 
Considering the ambient PM daily average concentrations measured during the same time, it seems 
unlikely that general ambient PM concentrations would have caused a large contribution to higher indoor 
PM concentrations in house 3KZ in the winter time. Higher summer indoor PM concentrations in house 
9KZ could be attributed to the burning of coal in summer months for bathing and cooking purposes. When 
social survey data is brought into the picture, it becomes clear, as in the KwaDela case study, that smoking 
in the indoor environment could contribute greatly to indoor air pollution: The social survey data from 
house 3KZ indicated that an individual living in the household was an active smoker. Whether or not this 
person smoked inside the house is not known, but it cannot be denied that this could be the reason the 
indoor PM measurements were high in this home. There was no smoker in house 9KZ. 
 
A very clear observation is that the indoor and personal daily average readings measured for both 
households in KwaZamokuhle correspond highly in winter, most likely indicating that the person wearing 
the PM monitor spent more time inside the house in question during colder months, possibly even igniting 
the fire, supporting that the individual was more exposed to the indoor air pollution.   
 
5.1.2 Temporal variation of particulate matter concentrations at a household level  
 
 
Diurnal hourly average trends, as ascertained at the community level in chapter 4 are reiterated here as 
measured for the four (4) cases. On average, morning peak concentrations were measured between 5 - 10 
am and evening peaks were measured between 5 – 9 pm in the four (4) households considered in this 
section of the study (Figure 5.3 - Figure 5.8). For every case, indoor and personal PM4 concentrations 
were not only consistently the highest measured concentrations when compared to ambient concentrations, 
but also the most variable.  PM concentrations of all types were consistently found to be highest in winter 
and higher in spring than in summer. Personal PM4 measurements were typically only available for the 
day hours in KwaDela, but for day and night time hours in KwaZamokuhle. This represents one of the 
limitations of the study, as conclusions are being drawn on an incomplete personal PM4 data set in 
KwaDela. 
 
Other overarching recognisable trends are burning times which correspond inversely with ambient 
temperatures, where lowest ambient temperatures typically represent times of peak burning activity and 
72 
 
highest PM concentrations in the indoor and the ambient environment. Indoor PM peak concentrations do 
not always correspond directly with ambient PM peak concentrations, regardless of the size fraction 
measured (PM10 or PM2.5), though a general trend towards early morning and late afternoon elevated 
average concentrations is measured for both ambient and indoor PM. Non-corresponding peaks could be 
attributed to lag effects, where it takes a while for ambient air to infiltrate into the indoor environment or 
vice versa. This could very likely be the case in house 3KD in KwaDela in particular, where indoor PM 
concentrations are high, though no coal burning activities take place. This remains mere speculation 
though, and this would have to be confirmed with infiltration rate measurements. Additionally, times in 
which indoor and personal PM concentrations do not follow the same diurnal trend as ambient 
concentrations, could point to different aerosol source contributions (House 11KD in winter 2014 in 
KwaDela and house 3KZ in summer 2015 in KwaZamokuhle). Where ambient concentrations are likely 
to be influenced by all domestic burning activities taking place in the community during the course of the 
day and traffic during the early hours of the morning and the later hours of the afternoon, indoor 
concentrations are, to a large degree, likely to be determined by indoor sources stemming from the 
particular house in question, and to a smaller degree by ambient sources. 
 
 
Figure 5.3 Diurnal variation of particulate matter (µg/m³) for houses 3KD and 11KD in KwaDela in 
Winter 2013 
 
 
 
73 
 
 
Figure 5.4 Diurnal variation of particulate matter (µg/m³) for houses 3KD and 11KD in KwaDela in 
Summer 2014 
 
Figure 5.5 Diurnal variation of particulate matter (µg/m³) for houses 3KD and 11KD in KwaDela in 
Summer 2015 
 
Hourly averge PM concentrations measured for the selected cases in KwaDela are well explained by 
considering survey data collected.  Diurnal bimodal peaks re-emphasise seasonal trends illustrated by 
daily average measurements where higher emissions are measured in winter than in summer. At a first 
glance, it is immediately apparent that hourly indoor and personal PM4 concentrations measured in house 
11KD are higher than those measured in house 3KD (except in summer 2015). Erratic hourly indoor 
measurements measured in house 11KD in summer 2014 could be attibutable to numerous burning times 
during the day adopted for cooking in the morning, midday and then again in the evening (Figure 5.3 - 
Figure 5.5). Again it is evident that indoor concentrations in house 3KD are high during peak burning 
times, even though no burning takes place in that specific house, according to survey data.  
74 
 
 
Figure 5.6 Diurnal variation for houses 3KZ and 9KZ in KwaZamokuhle in Spring 2015 
 
 
Figure 5.7 Diurnal variation for houses 3KZ and 9KZ in KwaZamokuhle in Summer 2016 
 
 
75 
 
 
Figure 5.8 Diurnal variation of particulate matter (µg/m³) for houses 3KZ and 9KZ in KwaZamokuhle in 
Winter 2016 
 
Peaks of hourly average concentrations of all PM types were evidently bimodal in all seasons in 
KwaZamokuhle (Figure 5.6 - Figure 5.8). Uncharacteristically, intermittent daytime peaks were measured 
at the KZ-E4-SE monitoring site in summer 2016. Indoor peak concentrations correspond well with survey 
results received where burning times in the mornings in House 9KZ start earlier in winter months than 
burning times in house 3KZ. The late afternoon burning evidently starts from 4pm, as also indicated in 
the surveys. Higher hourly average indoor and personal concentrations were also measured in house 3KZ 
in winter, which talks to the higher daily average concentrations that were measured for the same house 
for the same season. 
5.2 Relationships between ambient, indoor and personal PM concentrations 
 
Trends of closely matching indoor and personal PM concentrations, as suggested in the sections above, 
point to the fact that total exposure to PM (as measured by the SidePaks) can be heavily defined by PM 
concentrations stemming from the indoor micro-environment within which an individual is situated (Long 
& Sarnat, 2004).   
 
PM concentrations stemming from sources in the ambient environment can influence concentrations 
measured indoors, if infiltration occurs. Conversely, indoor air pollution can escape from chimneys or 
diffuse through gaps in a household structure to influence the ambient air (Meng et al., 2005). Bearing 
76 
 
this in mind, it could be expected that, were one to juxtapose different PM measurements taken at the 
same time, but in different locations through means of regression analysis, that distinct relationship trends 
could be defined (Wilson et al., 2000). 
 
5.2.1 PM concentration relationships at a community level 
 
 
When comparing PM types in KwaDela and KwaZamokuhle at a community level, no real clear 
discernible relationship trends could be found between PM concentrations measured at different 
monitoring sites (Figure 7.8 - Figure 7.15 in Annexure B). Low coefficients of determination were found 
for most regression analyses conducted at a community-level between indoor and ambient, as well as 
personal and ambient concentrations for all ambient PM types considered. Plotting different ambient PM 
concentrations from different E-Samplers/E-Bams against each other rendered similarly low coefficients 
of determination. This pertains to both KwaDela and KwaZamokuhle. On average, highest R2 values were 
found when plotting indoor and personal PM4 concentrations against each other.  Two household-level 
case studies have been extracted in Section 5.2.2 of this chapter to better showcase these case-level 
relationships between hourly average ambient, indoor and personal concentrations (Table 5.5 and Table 
5.6).  
 
At a community-level, it was difficult to ascertain whether there were any meaningful overarching 
relationship trends that governed whether or not the variation in one PM type could explain the variation 
in another PM type. This is true for all seasonal campaigns when plotting ambient, indoor and personal 
PM parameters against each other at every possible combination.  Single ad hoc high R2 values were found, 
but it is difficult to unpack these cases, as explanations are more based on speculation, rather than facts 
(Figure 7.8 - Figure 7.15 in Appendix B): 
 
In KwaDela, in winter 2013 for instance, of the 14 regressions that were plotted for hourly maximum 
concentrations, eight (8) of the regressions showed high R2 values (R2 > 0.6). In winter 2014 however, 
only three (3) of the same 14 regressions that were plotted for hourly maximum concentrations showed 
high R2 values. Across all sampling campaigns, only three (3) instances existed in KwaDela and four (4) 
instances existed in KwaZamokuhle, where the regressions for daily- and hourly mean as well as daily- 
and hourly maximum concentration regressions all showed high R2 values consistently across the 
77 
 
campaigns. This leads to the assumption that there could well be a meaningful overarching relationship 
observable. These instances have been singled out for interpretation (Table 5.3 and Table 5.4). 
 
Table 5.3.  Overview of singled out highest coefficients of determination (R2) resulting from community 
level regression analyses in KwaDela 
 Campaign Regression Daily 
average 
Daily 
maximum 
Hourly  
average 
Hourly 
maximum 
1) Winter 2013 PM10 vs PM2.5 0.8770 0.9416 0.9221 0.8292 
2) Winter 2013 PM2.5 vs PPM4 0.7063 0.9111 0.8640 0.8606 
3) Summer 2014 PM2.5 vs IPM4 0.6000 0.7963 0.7531 0.7546 
 
 
Table 5.4.  Overview of singled out highest coefficients of determination (R2) resulting from community 
level regression analyses in KwaZamokuhle 
 Campaign Regression Daily 
average 
Daily 
maximum 
Hourly 
average 
Hourly 
maximum 
1) Spring 2015 KZ-E1-N vs KZ-E2-NW 0.7803 0.7864 0.9424 0.8959 
2) Spring 2015 KZ-E2-NW vs KZ-E3-W 0.6750 0.8996 0.7622 0.8242 
3) Spring 2015 IPM4 vs PPM4 0.7320 0.5400 0.7571 0.9575 
4) Summer 2016 PM10 vs PPM4 0.9327 0.7623 0.7941 0.8364 
A higher coefficient of determination, as listed in the tables above, would indicate that the concentrations 
of the various PM types that have been plotted against each other in the regression analysis, have common 
sources or are affected by the same processes, as they follow the same trends, whereas a lower coefficient 
of determination would indicate the source contribution of varying sources to the various PM size fractions 
(Alves et al., 2014). This means that, for instance, in summer 2016 in KwaZamokuhle (Ambient PM10 vs 
Personal PM4 daily average concentrations, R
2 = 0.9327), 93% of the variation in personal PM4 
concentrations could be explained by the variation in ambient PM10 concentrations. In an attempt to 
interpret this strong correlation, one could say that, as these figures represent community-wide trends, the 
majority of people wearing personal monitors during summer 2016 spent a large proportion of their time 
in the ambient environment and were thus subjected to the same or similar sources or processes that 
influenced the concentrations measured at the AMS during that period. Of course, this is a hypothetical 
and simplified interpretation based on assumptions, and the cause for the high correlation could only truly 
be understood taking into account a range of complexities, such as cumulative time-activity patterns of 
individuals as well as larger- and smaller scale factors that influenced the ambient particulate 
concentrations measured during that time.  
78 
 
5.2.2 PM concentration relationships at a household level 
When considering relationships between PM types at a case level, no consistent relationship trends could 
be found i.e. at a household level between the various PM measurements. As such, there was no 
combination of PM types that rendered consistently high coefficients of determination across the board. 
High R2 values are highlighted in red. 
Out of all PM type regression combinations, the variability in indoor hourly average concentrations was, 
however, most likely to explain the variability in the personal hourly average concentrations.  For purposes 
of demonstration, cases 3KD and 11KD in KwaDela and 3KZ and 9KZ in KwaZamokuhle were 
considered (Table 5.5 and Table 5.6).  Three (3) out of six (6) R2 values in KwaDela and four (4) out of 
six (6) R2 values in KwaZamokuhle resulting from plotting indoor against personal PM concentrations 
showed R2 values >0.5. The high correspondence between indoor and personal PM values measured has 
been repeatedly noted in results discussed in section 4.1.5 in Chapter 4 and in results discussed in this 
chapter.   
In cases in which high R2 values were found for PM type combinations other than indoor and personal 
PM concentrations, it becomes difficult to understand, with certainty, what could have contributed to the 
good correlation. For example, in KwaDela, three (3) out of six (6) correlation coefficients of 
determination showed high R2 values when plotting hourly average PM2.5 concentrations measured at KD-
E1-NW against indoor PM4 concentrations, two of which were noted for winter 2013 when using indoor 
measurements from house 3KD and house 11KD. The location of KD-E1-NW in respect to house 3KD 
and house 11KD is very different, indicating that whatever source contributions defined ambient PM2.5 
concentrations at that location could have simultaneously defined indoor PM concentrations in houses 
3KD and 11KD. PM concentrations are generally high in winter, regardless of the micro-environment or 
PM type considered. Hence, it is more likely that overarching increases in burning practices taking place 
in different places in winter at specific times of the day caused PM concentrations in the ambient and 
indoor environment to rise or fall concurrently.  
In KwaZamokuhle, coefficients of determination were generally below 0.5 indicating that the variability 
of one PM type was less likely to explain the variation of another PM type most of the time.   
79 
 
Table 5.5.  Correlation coefficients of determination for hourly average PM types per house, per season 
in KwaDela  
 Winter 2013 Summer 2014 Summer 2015 
 House 3 House 11 House 3 House 11 House 3 House 11 
PM10 vs PM2.5 0.2797(*) 0.2797(*) 0.1877(*) 0.0680 0.2119(*) 0.2496(*) 
PM10 vs E1PM2.5 0.3283(*) 0.3283(*) 0.2206(*) 0.8547(*) 0.3084(*) 0.3030(*) 
PM10 vs E2PM2.5 0.0336 0.0336 0.0044 - 0.3909(*) 0.3061(*) 
PM10 vs IPM4 0.3814(*) 0.3814(*) 0.0617(*) 0.0293 0.3434(*) 0.2582(*) 
PM10 vs PPM4 0.2863 0.2863 0.4455(**) 0.9851(*) 0.7967(*) 0.2348 
PM2.5 vs E1PM2.5 0.6464(*) 0.6463(*) 0.2589(*) - 0.3075(*) 0.3362(*) 
PM2.5 vs E2PM2.5 0.0345 0.0345 0.0086 - 0.2866(*) 0.2907(*) 
PM2.5 vs IPM4 0.4326(*) 0.4326(*) 0.0546(*) 0.0007 0.1494(*) 0.1486(*) 
PM2.5 vs PPM4 0.3574 0.1147(**) 0.1965 0.9888(*) 0.0180 0.6399(*) 
E1PM2.5 vs E2PM2.5 - - 0.0010 - 0.4260(*) 0.4221(*) 
E1PM2.5 vs IPM4 0.5250(*) 0.5257(*) 0.1715(*) 0.7580(**) 0.2824**) 0.2188(*) 
E1PM2.5 vs PPM4 - - 0.7216(*) - 0.4703 0.3961(**) 
E2PM2.5 vs IPM4 0.0208 0.0208 0.0008 - 0.3464(*) 0.2318(*) 
E2PM2.5 vs PPM4 - - 0.4287 - 0.0656 0.5126(*) 
IPM4 vs PPM4 0.6207(*) 0.1589(**) - 0.8921 0.0811 0.9714(*) 
*correlation is significant at the 0.01 level 
**correlation is significant at the 0.05 level 
 
 
It can be assumed that the R2 values could improve between indoor and ambient concentrations if a lag 
effect were factored into the metaphorical equation. Studies have shown that there is a well-known lag 
between indoor and outdoor concentrations caused by the penetration time needed for ambient 
concentrations to come in through the building envelope (Guo et al., 2010). The duration of this lag effect 
depends on many factors.  If windows and doors are open for example, ambient air pollution can quickly 
infiltrate the indoor environment. Air entering the indoor environment through smaller cracks in the walls 
of the roof of the house, on the other hand, would infiltrate at a slower rate (Guo et al., 2010).  
 
 
Table 5.6.  Correlation coefficients of determination for hourly average PM types per house, per season 
(KwaZamokuhle) 
 Spring 2015 Summer 2016 Winter 2016 
 House 3 House 9 House 3 House 9 House 3 House 9 
PM10 vs PM2.5 0.5388(*) 0.4290(*) 0.6394(*) 0.3932(*) 0.3560(*) 0.3003(*) 
PM10 vs E1PM2.5 0.0685 0.1565(*) 0.2714(*) 0.1364(*) 0.4441(*) 0.2444(*) 
PM10 vs E2PM2.5 0.0376 0.1243(*) 0.3502(*) 0.2231(*) - - 
PM10 vs E3PM2.5 0.0075 0.1153(*) - 0.2448(*) 0.2373(*) 0.2699(*) 
PM10 vs E4PM2.5 - 0.2071(*) 0.2796(*) 0.0000 0.0031 0.2637(*) 
PM10 vs IPM4 0.2731(**) 0.0130(**) 0.4175(*) 0.1625(*) 0.4971(*) 0.2102(*) 
PM10 vs PPM4 - 0.0103 0.2532(*) 0.0007 0.3547(*) 0.2061(*) 
PM2.5 vs E1PM2.5 0.4277(*) 0.2311(*) 0.2621(*) 0.1777(*) 0.3191(*) 0.2240(*) 
PM2.5 vs E2PM2.5 0.3359(*) 0.1947(*) 0.3717(*) 0.2572(*) - - 
PM2.5 vs E3PM2.5 0.2496(*) 0.2098(*) - 0.2943(*) 0.1879(*) 0.2751(*) 
80 
 
PM2.5 vs E4PM2.5 0.3092(*) 0.2536(*) 0.2040(*) 0.0027 0.1467 0.2038(*) 
PM2.5 vs IPM4 0.3022(*) 0.0261(*) 0.4640(*) 0.2115(*) 0.2847(*) 0.2021(*) 
PM2.5 vs PPM4 - 0.0426(*) 0.3292(*) 0.0251 0.2940(*) 0.2380(*) 
E1PM2.5 vs E2PM2.5 0.4549(*) 0.1708(*) 0.3975(*) 0.3353(*) - - 
E1PM2.5 vs E3PM2.5 0.1430(*) 0.1467(*) - 0.2638(*) 0.2314(*) 0.4177(*) 
E1PM2.5 vs E4PM2.5 0.1953(*) 0.2449(*) 0.1958(*) 0.0025 0.1389 0.1124(*) 
E1PM2.5 vs IPM4 0.1067(*) 0.0264(*) 0.2082(*) 0.1337(*) 0.5761(*) 0.2751(*) 
E1PM2.5 vs PPM4 0.0579 0.0502(*) 0.0941(*) 0.1888(*) 0.4441(*) 0.2588(*) 
E2PM2.5 vs E3PM2.5 0.3168(*) 0.1840(*) - 0.4577(*) - - 
E2PM2.5 vs E4PM2.5 0.2770(*) 0.1205(*) 0.1750(*) 0.0007 - - 
E2PM2.5 vs IPM4 0.0828(**) 0.1195(*) 0.2609(*) 0.3167 - - 
E2PM2.5 vs PPM4 0.0198 0.0652(*) 0.1172(*) 0.2790(*) - - 
E3PM2.5 vs E4PM2.5 0.2265(*) 0.0758(*) - 0.0018 0.0111 0.0556 
E3PM2.5 vs IPM4 0.0880(**) 0.0880(*) - 0.1599(*) 0.3969(*) 0.4670(*) 
E3PM2.5 vs PPM4 0.0610 0.0569(*) - 0.1340(*) 0.5044(*) 0.3938(*) 
E4PM2.5 vs IPM4 0.2052(*) 0.0057 0.2385(*) 0.0141 0.0032 0.1134(*) 
E4PM2.5 vs PPM4 - 0.0464(*) 0.1920(*) 0.0589(**) 0.0722 0.1388(*) 
IPM4 vs PPM4 0.8959(*) 0.5351(*) 0.4905(*) 0.1086(*) 0.8796(*) 0.8434(*) 
*correlation is significant at the 0.01 level 
**correlation is significant at the 0.05 level 
 
 
Considering the above outlined, it becomes clear once again that PM concentrations in the low-income 
context are highly variable in space and time and so it is very difficult to understand and ascertain 
meaningful relationships.  
 
5.2.3 Ratios between indoor and personal PM measurements 
 
Regression analyses are one way to determine relationships between varying PM types. Ratios represent 
another means to gain an understanding of whether or not a relationship exists between PM measurements 
(Chen & Zhao, 2011). The hourly average Personal PM4/ Indoor PM4 ratios for the case study houses 
differed in all houses and across the seasons (Figure 5.9). In KwaDela, the Personal PM4/ Indoor PM4 
ratio for cases 3KD and 11KD were > 1 most of the time, except for case 11KD in winter 2013 during the 
late afternoon and case 3KD between 10:00 and 12:00 in the day. Notably, the highest ratios were evident 
during the middle of the day in all seasons and also in the early hours of the evening for case 3 in winter 
2013. This could mean that the person wearing the SidePak monitor was outside, and not exposed to high 
indoor concentrations during the day, particularly so in summer and during the evenings in winter 2013. 
It is reiterated that personal measurements were not available for the earliest hours of the morning and the 
latest hours of the afternoon in KwaDela.  
 
81 
 
Personal PM4/ Indoor PM4 ratios in KwaZamokuhle were closer to one (1) across all seasons and across 
most hours of the day. The Personal PM4/ Indoor PM4 ratio was found to be >1 in the early hours of the 
morning for case 9KZ in spring and in summer and in particular, in the late hours of the night for case 
3KZ in spring.  The individual in case 3KZ seems to spend more time indoors than the individual in case 
9KZ, as indoor and personal concentrations correspond. 
 
The hourly average Personal PM4/ Ambient PM2.5 ratios were more often than not >1, indicating that 
indoor PM4 concentrations were higher than ambient PM2.5 concentrations most of the time (Figure 5.10). 
This was particularly the case in the early mornings and the evenings, as per the typical diurnal peaks 
illustrated in all hourly average timeseries graphs plotted for this study.  
 
The non-congruence between ambient and indoor PM measurements is most likely to be attributed to the 
various source contributors: indoor PM concentrations will more than likely have been governed by 
localised indoor burning practices, whereas ambient concentrations will have been governed by a 
combination of various external source contributors. These findings echo what has been found in other 
research studies, namely that personal exposure to PM is on average, poorly correlated with ambient 
concentrations and more highly correlated with indoor concentrations (Janssen, 2005). 
    
Figure 5.9.  Hourly average Personal PM4/ Indoor PM4 ratio in KwaDela (left) and KwaZamokuhle 
(right) 
 
82 
 
 
Figure 5.10.  Hourly average Personal PM4/ Ambient PM2.5 ratio in KwaDela (left) and KwaZamokuhle 
(right) 
 
 
5.3 Personal total exposure to particulate matter concentrations in various micro-
environments 
 
5.3.1 The identification of micro-environments in KwaDela 
 
 
Total exposure takes into account the period of time an individual comes into contact with an exposure 
concentration throughout the day and in various micro-environments (U.S. EPA 1992).  In KwaDela, total 
exposure was explored by using GPS tracking devices on consenting individuals who also carried personal 
PM4 monitors.   
 
A total of five (5) different micro-environments were identified by considering where all individuals 
carrying GPS monitors spent most of their time in July, August and September 2013, respectively. These 
included, in order of importance: 1) Inside a house, 2) Directly outside a house, 3) On a dirt road, 4) On a 
tar road and 5) On an open field (Figure 5.11).  Most of the personal movements (on average 87%) were 
concentrated inside and directly outside homes. Individuals spent on average 11% of their time on roads, 
moving between destinations within the settlement and a mere 2% of their time on open fields in transit 
from one destination to another. The GPS coordinate readings represented day time movement patterns 
ranging from 9am in the morning to 6pm in the late afternoon. No position readings were taken during the 
night, but it can be assumed that the individuals spent a low fraction of time outside of the household 
83 
 
during the evening, and spent most of their time indoors at night. People were found to spend more time 
indoors in colder months and conversely spent more time outdoors as months became progressively 
warmer (Figure 5.11).  
 
 
Figure 5.11.  Average percentage of time spent in each micro-environment per month for which GPS 
measurements were collected (July n = 8831, August n = 25 193, September n = 11 909) 
 
 
5.3.2 Total personal exposure: A one-individual case study 
 
 
When plotting location points and corresponding PM concentrations on a SPOT 6/7 satellite image of 
KwaDela, it was observed that the person carrying the GPS and personal PM4 monitors was exposed to a 
wide range of concentrations in every micro-environment (Figure 5.12).  Here, the individual was exposed 
to PM4 concentrations ranging from 36 – 2781 µg/m³ on 25 August 2013. When plotting similar figures 
for every other day between 21 and 28 August 2013, it was found that personal exposure concentrations 
in this case study were highly variable in space and time and that they did not necessarily follow a 
predictable trend (Table 7.1 in Appendix C). It is evident that the individual is exposed to high PM4 
concentrations throughout the day in both the indoor environment as well as in close proximity to a house, 
i.e. in the ambient environment.  
 
53%
35%
2%
6% 5%
48%
39%
3%
7%
4%
38%
49%
2%
8%
3%
46%
41%
2%
7%
4%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
In house Directly outside
house
Open field Dirt road Tar road
A
v
er
ag
e 
ti
m
e 
sp
en
t 
in
 m
ic
ro
-
en
v
ir
o
n
m
en
t 
(%
)
Micro-environment
July August September Total
84 
 
During the time in question, the individual (a middle-aged woman), spent the majority of her time in two 
micro-environments whilst wearing the monitors, namely inside a house and directly outside a house. 
Because measurements were only taken during the day hours, it makes sense that the individual spent a 
large proportion of time outside (Figure 5.14). It is assumed that the night hours were mostly spent in the 
indoor environment, however, no data is available to substantiate this statement. It is assumed that, 
because the measurements taken only represent exposure concentrations between 9 am and 6 pm, that 
peak PM4 concentrations, resulting during main burning periods have not been accounted for (Figure 
5.13). 
 
 
 
Figure 5.12.  GPS tracks showing average personal PM4 concentration exposure on the 25 August 2013. 
Micro-environments displayed are focused on “Inside a house” (delineated by an outlined rectangle) and 
“Directly outside a house” (the space outside the rectangles) 
 
 
 
85 
 
 
Figure 5.13.  Hourly average personal PM4 concentrations measured from 21-28 August 2013 
 
 
 
Figure 5.14.  Percentage of time spent in each micro-environment per day in question by the person 
wearing the GPS monitor between 21 and 28 August 2013 
 
 
Potential PM4  intake doses ascertained using indirect and direct methods for the days concerned were 
variable (Table 5.7). Indirect, daily time-weighted integrated potential doses ranged from 0.92 mg on 26 
August 2013 to 2.96 mg on 28 August 2013 (Table 7.2 in Annexure D).  When considering direct potential 
dose derivation methods, intake doses were found to range between 0.02 mg and 0.76 mg for the time for 
0
500
1000
1500
2000
2500
3000
3500
4000
4500
0
100
200
300
400
500
600
700
800
900
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
P
M
4
 c
o
n
ce
n
tr
at
io
n
 f
o
r 
2
5
 A
u
g
u
st
 2
0
1
3
 o
n
ly
 (
µ
g
.m
³)
P
M
4
 c
o
n
ce
n
tr
at
io
n
 (
µ
g
.m
³)
Time (Hours of the day)
21-Aug-13 22-Aug-13 23-Aug-13 24-Aug-13
26-Aug-13 27-Aug-13 28-Aug-13 25-Aug-13
86 
 
which measurements were available (only a few hours in the day).   Highest doses corresponded to the 
days on which the most readings were available. It is not possible to directly compare direct and indirect 
results for an entire day, as there is no full 24-hour dataset for personal measurements. However, results 
indicate that, when comparing indirectly derived inhalation doses with directly derived inhalation doses 
only for the time during which personal measurements were available, it is at times underestimated what 
is breathed in by the individual concerned (i.e. indirectly derived doses are lower for five (5) out of eight 
(8) days when compared to directly derived doses).  
 
Table 5.7. Directly and indirectly derived potential doses (mg) over a specified time period 
 
Indirect time-weighted 
potential dose for the entire 
day 
Indirect time-weighted 
potential dose for the hours 
in which personal 
measurements were 
available 
Direct potential dose 
Date Dose (mg) Time (hours) Dose (mg) Time (hours) Dose (mg) Time (hours) 
21-Aug-13 1.50 24.00 0.02 1.00 0.02 1.00 
22-Aug-13 2.89 24.00 0.08 4.83 0.05 4.83 
23-Aug-13 2.10 24.00 0.03 7.17 0.52 7.17 
24-Aug-13 1.49 24.00 0.14 3.50 0.21 3.50 
25-Aug-13 1.98 24.00 0.35 5.83 0.76 5.83 
26-Aug-13 0.92 24.00 0.25 4.33 0.19 4.33 
27-Aug-13 1.30 24.00 0.14 2.67 0.17 2.67 
28-Aug-13 2.96 24.00 0.23 3.50 0.34 3.50 
 
 
Above it was ascertained how much PM is breathed in by the concerned individual during specified time 
periods. The resulting figures are representative of doses of PM inhaled during winter as a result of time 
spent exposed to PM4 in the ambient and the indoor environment.   Though indirectly derived daily doses 
listed here fall below what has been demonstrated to be breathed in in other low-income communities (e.g. 
when compared to potential inhalation doses in Bangladesh where doses ranged between 4.4mg and 5.8mg 
of PM2.5 per day (Chowdhury et al., 2012)), it should be kept in mind that, particularly for smaller PM 
size fractions, no threshold exists that does not cause a discernible health effect. Additionally, it is very 
likely that the directly ascertained total daily dose intake fraction of the individual would in fact be much 
higher than what has been demonstrated by means of indirect derivation methods as presented here, as 
peak exposure concentrations during burning times have not been included, and most notably because the 
87 
 
direct doses have been ascertained with an incomplete dataset. It remains to be said that this exercise 
showed how an individual in a low-income community context is directly exposed to PM concentrations 
that by far exceed ambient standards in numerous micro-environments, and this leads to unacceptably 
high potential doses of PM being inhaled.  
 
 
 
******************** 
 
 
This chapter discussed whether larger scale PM characterisation trends are 
also evident at a household scale. As part of this, regression analyses were 
conducted to ascertain whether there are any clear relationships to be found 
when considering ambient, indoor and personal PM concentrations at a case 
and at a community level. A one-individual case study was also presented, 
illustrating the concentrations to which an individual was exposed in 
specific micro-environments and what dose of PM was inhaled over 
different timeframes.  
 
The next chapter represents the final chapter of this study, which outlines 
the main conclusions that can be drawn from the results presented herein. 
 
 
 
 
 
 
 
 
 
 
 
88 
 
 Summary and conclusions 
It was the intention of this study to characterise ambient, indoor and personal 
PM concentrations of different size fractions in two low-income 
communities on the Mpumalanga Highveld: KwaDela and KwaZamokuhle. 
Data used was collected between 2013 and 2016 as part of Sasol and 
Eskom’s air quality offset pilot studies. Relationships between ambient, 
indoor and personal PM concentrations were investigated. Personal 
exposure to PM concentrations was assessed by considering a one-
individual case study.  
Main findings are summarised in this chapter and they are relevant to the 
assessment of whether centrally localised ambient stations which are used 
to draw compliance and epidemiological conclusions in such communities, 
represent what people actually breathe. This is relevant for air quality 
management strategies in low-income communities on the Highveld.  
 
Air pollution has been identified as one of the leading causes of global disease burden, and this most 
particularly in low-income communities around the world (Cohen et al., 2017; Rosenthal et al., 2017). 
Studies conducted in low-income communities in South Africa support these findings, and, significantly 
relevant for this study, it has been found that residential combustion activities have been singled out to 
represent the greatest source of air pollution in low-income communities on the Mpumalanga Highveld 
(Engelbrecht et al., 2000; Engelbrecht et al., 2002). PM concentrations measured in such communities 
have been shown to exceed NAAQS limit values (Mdluli, 2007), and observed relationships between air 
pollution and morbidity and mortality have been demonstrated (Terblanche et al. 1992a; Wichmann, 
2006).   
 
As research may be termed one of the most important tools that can be used to provide a foundation for 
action geared towards public health improvement, it is the purpose of this study to help understand how 
PM concentrations in two low-income communities on the Mpumalanga Highveld are characterised by 
larger- and local-scale factors and how this influences variability and relationships between ambient, 
indoor and personal measured concentrations. This is of great importance when understanding personal 
exposure to PM concentrations in such communities, in particular where compliance and epidemiological 
related statements are made, using data collected at a sole stationary ambient monitoring station.  
89 
 
Ambient, indoor and personal PM concentrations were measured in KwaDela and KwaZamokuhle 
between 2013 and 2016 as part of Sasol and Eskom’s air quality offset pilot studies. The data collected in 
these campaigns was used to characterise PM concentrations within and between these two low-income 
communities by considering seasonal, temporal and socio-economic factors. Empirical relationships 
between ambient, indoor and personal PM concentrations were identified by means of regression analyses 
and ratios and total personal exposure was assessed by dissecting a one-individual case study, in which 
simultaneously collected personal PM4 concentrations and GPS location data were interpreted. The below 
sections list the main findings per research objective.  
 
6.1 Characterising particulate matter concentrations In KwaDela and KwaZamokuhle 
 
Ambient, indoor and personal PM concentrations were measured in KwaDela and KwaZamokuhle 
communities between 2013 and 2016. Comparative statements were made to explain what PM 
concentrations in KwaDela and KwaZamokuhle look like in different seasons of the year and different 
times of the day. Daily, hourly as well as maximum concentrations were considered for the purposes of 
this exercise.  Main findings are: 
 
1. Air quality in KwaDela and KwaZamokuhle is unacceptably poor, especially so in winter months, 
when NAAQS are exceeded most of the time.  
 
2. Ambient daily average winter PM10 and PM2.5 concentrations in both KwaDela and 
KwaZamokuhle were found to be on average, twice as high as summer PM concentrations. 
 
3. NAAQS were exceeded in spring, summer and winter, with most exceedances occurring in the 
winter time. 
 
4. Indoor and personal daily PM4 concentrations were up to three (3) times higher in winter than in 
summer in both communities. 
 
5. Indoor PM concentrations are typically higher than ambient PM concentrations, particularly so 
during peak burning periods and this effect is exacerbated in colder months. 
90 
 
6. Indoor and personal daily average and hourly maximum PM4 concentrations were on average 
higher in KwaZamokuhle than in KwaDela, except in summer. 
 
7. Hourly maximum indoor and personal PM concentrations were higher in winter than in summer 
in both communities.  
 
8. A pronounced bimodal diurnal pattern was identified when plotting overall hourly average time-
series graphs for ambient PM measurements in both KwaDela and KwaZamokuhle pointing to 
peak concentrations during main burning times in the morning and the late afternoon/ evening. 
Winter peaks were more pronounced than summer peaks. 
 
9. The variation of daily and hourly maximum concentrations around the respective PM type mean 
values was generally high within and across the communities, however, it was higher in spring 
than in summer, and highest in winter.  
 
10. Results found at a community level were evident at a household level where: 1) Winter average 
PM concentrations were higher than summer concentrations. 2) PM concentrations were highly 
variable in space and time. 3) Indoor PM measurements were higher than ambient PM 
measurements. 
 
6.2 The relationships between ambient, indoor and personal PM concentrations in 
KwaDela and KwaZamokuhle 
 
Simultaneously collected hourly average ambient, indoor and personal PM measurements were log 
normally transformed and compared by means of regression analysis. It was identified whether or not 
discernible relationships between ambient, indoor and personal PM concentrations were evident. Relevant 
ratios were determined. Main findings that unpack this objective are: 
 
11. Overall, no clear discernible relationship trends could be found between PM concentrations 
measured at different monitoring sites, be this at a community or a household level. 
91 
 
12. However, out of all PM type regression combinations, the variability in indoor hourly average 
measurements was most likely to explain the variability of the personal hourly average 
concentrations, and this in particular in the winter time, when most people spend more of their 
time in the indoor environment. This is supported by the finding that indoor and personal PM 
concentrations are most likely to correspond in the winter time (i.e. indoor: personal PM4 ratio was 
most likely to be 1 during winter and less so during summer). 
13. High variability of PM concentrations in space and time made it difficult to identify clear 
relationships between the different PM types.  
 
14. Indoor PM4 concentrations are generally higher than ambient concentrations, this even in some 
households that do not conduct domestic burning activities, pointing to the fact that pollution 
infiltration into the indoor environment is a major contributor to indoor air pollution.  
 
6.3 Personal exposure to particulate matter concentrations in KwaDela 
 
GPS tracking data and personal PM4 measurements taken between 21 and 28 August 2013 were paired at 
ten-minute average intervals. By plotting these datasets onto an aerial photograph of KwaDela using QGIS 
2.14.1, total time-activity patterns helped understand personal PM exposure per micro-environment.  Here 
main findings could be summarised as: 
 
15. Five micro-environments within which individuals spend their time were identified in KwaDela: 
1) Inside a house, 2) Directly outside a house, 3) On a dirt road, 4) On a tar road and 5) In an open 
field. 
 
16. Most individuals who carried GPS monitors spent on average 90% of their time in and around a 
household. Less time was spent in roads and open fields. 
 
17. Exposure to highest concentrations could not be pinpointed to the indoor environment only, as 
extreme personal PM measurements were also recorded when the individual carrying the GPS 
92 
 
monitor was in the ambient environment, however, most extreme exposure concentrations were 
recorded in the indoor environment and directly outside of houses.  
 
18. Dose intake rates calculated in an indirect manner using stationary monitoring data seemed to 
underestimate actual intake rates, as illustrated when considering dose rates calculated form direct 
personal exposure measurements.   
 
6.4 Conclusion  
 
An unprecedentedly rich dataset was used for the purposes of this research study. In no other South African 
study have there been as many simultaneously measured ambient, indoor and personal PM readings that 
span across numerous seasons and in two different communities. It was found that air quality in KwaDela 
and KwaZamokuhle is extremely poor throughout the year, and that air pollution levels are particularly 
poor in winter months. This confirms assumptions made, that KwaDela and KwaZamokuhle, just like 
many other low-income communities situated on the Mpumalanga Highveld, are plagued by air pollution 
levels that have a significant impact on the health and well-being of the concerned population. Findings 
further indicate that, regardless of the location of such a low-income community, the socio-economic 
factors in the respective location have a determining role to play in defining the air that people breathe.  
 
PM concentrations in the indoor environment were more often than not found to be higher than ambient 
PM concentrations, often reaching into the mg/m3 range during peak burning times. This points to acute 
exposure events that have been shown to represent a major contributor to respiratory and cardiovascular 
diseases (South Africa, 2017). The results of this study emphasise the importance of considering indoor 
air pollution levels alongside ambient levels when conducting epidemiological studies concerning the 
health impact of air pollution in low-income communities and when developing air quality management 
plans. In support of this notion, soon to be published indoor air quality guidelines will include indoor PM 
limit values, which aim to ensure that when a specific threshold of PM levels is exceeded indoors, this 
acts as a catalyst to call for action targeted towards the implementation of household-level interventions 
to reduce indoor air pollution (South Africa, 2017). The fact that the PM levels in some of the houses in 
this study were at times lower than ambient PM concentrations shows that it is indeed possible for indoor 
93 
 
concentrations to be lower than ambient concentrations and that it is possible to reduce indoor PM 
concentrations to safe breathing levels with the correct mitigation strategy.  
 
It follows on from the above stated that, because ambient, indoor and personal PM levels measured during 
this research study have been shown to be highly variable in space and time, epidemiologists should heed 
that indirect personal exposure assessments need to be carefully conducted. Optimally, when conducting 
exposure assessments in a low-income community setting, personal measurements should be used to 
calculate personal exposure concentrations and resulting dose intake fractions, as averaged readings 
stemming from stationary monitors alone will not render adequate results as they do not take into account 
this variability. Time-activity pattern data collected in this study showed that individuals spent most of 
their time inside or directly outside of a household. This means that the air quality in and directly outside 
of the household mostly determined the personal exposure concentrations of an individual. A centrally 
located ambient monitoring station situated at a school or a church, as was the case in KwaDela and 
KwaZamokuhle, is then evidently not a true representation of what people are breathing, as it doesn’t take 
into account the “household” air pollution, which represents the air directly inside and in close proximity 
to a household. 
 
In conclusion, this research recommends that compliance and epidemiologically inclined statements 
should express caution when using PM concentrations measured at stationary ambient monitoring stations 
as a proxy for personal exposure as PM concentrations in a low-income community context are non-
homogenous in space and time. Though using data measured at a single monitoring site may be useful 
when trying to understand chronic health effects associated with long-term exposure to suspended 
particles, this study shows that these measurements will most certainly only be of limited use when trying 
to understand acute exposure and this could lead to “exposure misclassification” (Wilson, 2006). Total 
acute and chronic exposure can best be defined by considering ambient, indoor and personal 
measurements simultaneously, and this information should be heavily supported by socio-economic data 
for contextualisation. If this holistic approach is taken into account, air quality management strategies will 
go far in making the air quality in low-income communities safer to breathe.  
 
 
 
94 
 
Project Funding 
 
This research study is funded by North West University. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
95 
 
References 
 
Abbey, D. E., Nishino, N., McDonnell, W. F., Burchette, R. J., Knutsen, S. F., Beeson, W. L. & Yang, J. 
X. 1999. Long-term inhalable particles and other air pollutants related to mortality in nonsmokers. 
American Journal of Respiratory and Critical Care Medicine,  159(2):373–382.  
Adgate, J. L., Ramachandran, G., Pratt, G. C., Waller, L. A. & Sexton, K. 2002.  Spatial and temporal 
variability in outdoor, indoor, and personal PM 2.5 exposure. Atmospheric Environment,  
36(2002):3255–3265. 
Ali, Z., Shahzadi, K., Sidra, S., Zona, Z., Zainab, I., Aziz, K., Ahmad, M., Raza, S. T., Nasir, Z. A. & 
Colbeck, I. 2015.  Seasonal variation of particulate matter in the ambient conditions of Khanspur, 
Pakistan.  Journal of Animal and Plant Sciences,  25(3):700–705. 
Alvarado, M. J. & Prinn, R. G. 2009.  Formation of ozone and growth of aerosols in young smoke 
plumes from biomass burning: 1. Lagrangian parcel studies.  Journal of Geophysical Research 
Atmospheres,  114(9):1–27.  
Alves, C.A., Urban, R.C., Pegas, P.N. & Nunes T. 2014. Indoor/Outdoor Relationships between PM10 
and Associated Organic Compounds in a Primary School.  Aerosol and Air Quality Research,  14:86-
98.  
Balmer, M. 2007.  Household coal use in an urban township in South Africa.   Journal of Energy In 
Southern Africa,  18(3):27–32.  
Bartington, S. E., Bakolis, I., Devakumar, D., Kurmi, O. P., Gulliver, J., Chaube, G., Manandhar, D. S., 
Saville, N. M., Costello, A., Osrin, D., Hansell, A. L. & Ayres, J. G. 2017.  Patterns of domestic 
exposure to carbon monoxide and particulate matter in households using biomass fuel in Janakpur, 
Nepal.  Environmental Pollution,  220(2017):38–45.  
Bradshaw, D., Groenewald, P., Laubscher, R., Nannan, N., Nojilana, B., Norman, R., Pieterse, D., 
Schneider, M., Bourne, D. E., Timæus, I. M., Dorrington, R. & Johnson, L. 2003. Initial burden of 
disease estimates for South Africa, 2000.  Cape Town, South African Medical Research Council. 
96 
 
Briggs, D. J., Collins, S., Elliott, P., Fischer, P., Kingham, S., Lebret, E., Pryl, K., VanReeuwijk, H., 
Smallbone, K. & VanderVeen, A. 1997.  Mapping urban air pollution using GIS: a regression-based 
approach.  International Journal of Geographical Information Science,  11(7):699–718.  
Briggs, D. J., De Hoogh, C., Gulliver, J., Wills, J., Elliott, P., Kingham, S. & Smallbone, K. 2000.  A 
regression-based method for mapping traffic-related air pollution: Application and testing in four 
contrasting urban environments.  Science of the Total Environment,  253(1–3):151–167.  
Brook, R. D., Rajagopalan, S., Pope, C. A., Brook, J. R., Bhatnagar, A., Diez-Roux, A. V., Holguin, F., 
Hong, Y., Luepker, R. V., Mittleman, M. A., Peters, A., Siscovick, D., Smith, S. C., Whitsel, L. & 
Kaufman, J. D. 2010.  Particulate matter air pollution and cardiovascular disease: An update to the 
scientific statement from the american heart association.  Circulation,  121(21):2331–2378.  
Bruce, N., Perez-Padilla, R. & Albalak, R. 2000.  Indoor air pollution in developing countries: a major 
environmental and public health challenge.  Bulletin of the World Health Organization,  78(9):1078–
1092.  
Bruce, N., Rehfuess, E., Mehta, S., Hutton, G. & Smith, K.  2002.   Indoor Air Pollution.  (In  Jamison, 
D.T.,  Breman, J.G., Measham, A.R., Alleyne, G., Claeson, M., Evans, D.B., Jha, P.,  Mills, A. & 
Musgrove, P., eds. Disease Control Priorities in Developing Countries.  New York: Oxford 
University Press.  p. 793-815). 
Brunekreef, B. & Holgate, S. T. 2002.  Air pollution and health.  The Lancet, 360:1233–1242.  
Bundi, P. M. 2004.  Spatial and Temporal Distribution of Tropospheric Ozone Over Southern Africa.  
Johannesburg: University of the Witswatersrand.  (Thesis - MSc).  
Burton, R.M., Suh, H.H. & Koutrakis, P.  1996.  Spatial Variation in Particulate Concentrations within 
Metropolitan Philadelphia.  Environmental Science & Technology,   30(2):400-407. 
Cao, J. J., Huang, H., Lee, S. C., Chow, J. C., Zou, C. W., Ho, K. F. & Watson, J. G. 2012.  
Indoor/outdoor relationships for organic and elemental carbon in PM2.5 at residential homes in 
Guangzhou, China.   Aerosol and Air Quality Research,  12(5):902–910.  
Chen, B. H. & Smith, K. R. 1991.  Indoor air pollution in developing countries.  The Lancet,  
337(8737):358–359. 
97 
 
Chen, C. & Zhao, B. 2011.  Review of relationship between indoor and outdoor particles: I/O ratio, 
infiltration factor and penetration factor.  Atmospheric Environment,  45(2):275–288.  
Chow, J.C., Watson, J.G., Mauderly, J.L., Costa, D.L., Wyzga, R.E., Vedal, S., Hidy, G.M., Altshuler, 
S.L., Marrakc, D., Heuss, J.M., Wolff, G.T., Pope III, C.A. & Dockery, D.W. 2006. Health Effects of 
Fine Particulate Air Pollution: Lines that connect.  Journal of the Air and Waste Management 
Association, 56:1368-1380.  
Chowdhury, Z., Thi Le, L., Masud, A.A., Chang, K.C., Alauddin, M., Hossain, M., Zakaria, A.B.M. & 
Hopke, P.K. 2012.  Quantification of Indoor Air Pollution from Using Cookstoves and Estimation of 
its Health Effects on Adult Women in Northwest Bangladesh. Aerosol and Air Quality Research,  
12:463-475. 
Chowdhury, Z., Campanella, L., Gray, C., Al Masud, A., Marter-Kenyon, J., Pennise, D., Charron, D. & 
Zuzhang, X. 2013.  Measurement and modeling of indoor air pollution in rural households with 
multiple stove interventions in Yunnan, China.  Atmospheric Environment,  67(2013):161–169.  
Climate-Data.  2017.  https://en.climate-data.org/location/189567/  Date of Access 23 Oct. 2017. 
Cohen, A. J., Anderson, H. R., Ostro, B., Pandey, K. D., Krzyzanowski, M., Künzli, N., Gutschmidt, K., 
Iii, C. A. P., Romieu, I., Samet, J. M. & Smith, K. R. 2004.  Urban Air Pollution.  Comparative 
Quantification of Health Risks,  77:1353–1434. 
Cohen, A. J., Brauer, M., Burnett, R., Anderson, H. R., Frostad, J., Estep, K., Balakrishnan, K., 
Brunekreef, B., Dandona, L., Dandona, R., Feigin, V., Freedman, G., Hubbell, B., Jobling, A., Kan, 
H., Knibbs, L., Liu, Y., Martin, R., Morawska, L., Pope, C. A., Shin, H., Straif, K., Shaddick, G., 
Thomas, M., van Dingenen, R., van Donkelaar, A., Vos, T., Murray, C. J. L. & Forouzanfar, M. H. 
2017.  Estimates and 25-year trends of the global burden of disease attributable to ambient air 
pollution: An analysis of data from the Global Burden of Diseases Study 2015.  The Lancet,  
389(10082):1907–1918.  
Dagsson-Waldhauserova, P., Magnusdottir, A. O., Olafsson, H. & Arnalds, O. 2016.  The spatial 
variation of dust particulate matter concentrations during two icelandic dust storms in 2015.  
Atmosphere,  7(6):1–12.  
98 
 
Diapouli, E. 2011.  Indoor and Outdoor Particle Number and Mass Concentrations in Athens. Sources, 
Sinks and Variability of Aerosol Parameters.   Aerosol and Air Quality Research,  11:632–642.  
Dlamini, T. S. 2007.  Analysis of trace gas emissions from spontaneous coal combustion at a South 
African colliery.  Johannesburg: University of the Witswatersrand.  (Thesis - MSc). 
Elgethun, K. A. I., Yost, M. G., Fitzpatrick, C. T. E. & Nyerges, T. L. 2007.  Comparison of global 
positioning system (GPS) tracking and parent-report diaries to characterize children's time – location 
patterns.  Journal of Exposure Science and Environmental Epidemiology,  17:196–206.  
Eskom.  2017.  The Execution of a Household Emissions Offset Study in the Highveld Priority Area, 
Mpumalanga: Pilot Project Year 2.  Final Report.  Johannesburg: Eskom. (Unpublished). 
Engelbrecht, J. P., Swanepoel, L., Zunckel, M., Chow, J. C., Watson, J. G. & Egami, R. T. 2000.  
Modelling PM 10 aerosol data from the Qalabotjha low-smoke fuels macro-scale experiment in South 
Africa.  Ecological Modelling,  127:235–244 . 
Engelbrecht, J. P., Swanepoel, L., Chow, J. C., Watson, J. G. & Egami, R. T. 2002.  The comparison of 
source contributions from residential coal and low-smoke fuels, using CMB modeling, in South 
Africa.  Environmental Science and Policy,  5(2):157–167.  
Ezzati, M. & Kammen, D. M. 2001.  Quantifying the effects of exposure to indoor air pollution from 
biomass combustion on acute respiratory infections in developing countries.  Environmental health 
perspectives,  109(5):481–488.  
Ezzati, M. & Kammen, D. M. 2002.  Household Energy, Indoor Air Pollution, and Health in Developing 
Countries: Knowledge Base for Effective Interventions.   Annual Review of Energy and the 
Environment,  27(1):233–270.  
Ferro, A. R., Kopperud, R. J. & Hildemann, L. M. 2004.  Elevated personal exposure to particulate 
matter from human activities in a residence.  Journal of exposure analysis and environmental 
epidemiology, 14(1):34–40.  
Fourie, G. D., Djolov, G.D., & Pienaar, J.J.  2008.   Modelling long-range transport and chemical 
transformation of pollutants in the southern Africa region.  Journal of Basic and Applied Research 
International,  4(2): 2395-3446.  
99 
 
Freiman, M. T. & Tyson, P. D. 2000.  The thermodynamic structure of the atmosphere over South 
Africa: Implications for water vapour transport.  Water SA,  26(2):153–158. 
Friedl, A., Holm, D., John, J., Kornelius, G., Pauw, C.J., Oosthuizen, R. & van Niekerk, A.S.   2008.  
Air pollution in dense, low-income settlements in South Africa.  Paper presented at the National 
Association for Clean Air Conference. Nelspruit, Mpumalanga, 1-3 October.  
Fullerton, D., Bruce, N. & Gordon, S.  2008.   Indoor Air Pollution from Biomass Fuel Smoke is a 
Major Health concern in the Developing World.   Transactions of the Royal Society of Tropical 
Medicine and Hygiene,  102 (9):843–851. 
Garland, R. M., Naidoo, M. & Sibiya, B. 2017.   Air quality indicators from the Environmental 
Performance Index : potential use and limitations in South Africa.  The Clean Air Journal,  27(1):1–9. 
Garstang, M., Tyson, P. D., Swap, R. J., Edwards, R., Kallberg, P. & Lindesay, J. A. 1996.  Horizontal 
and vertical transport of air over southern Africa.  Journal of Geophysical Research,  10(19):23721–
23736. 
Gatebe, C. K., Tyson, P. D., Annegarn, H., Piketh, S. & Helas, G. 1999.  A seasonal air transport 
climatology for Kenya.  Journal of Geophysical Research,   104(D12):14237–14244.  
Guo, H., Morawska, L., He, C., Zhang, Y.L., Ayoko, G. & Cao, M. 2010.  Characterization of particle 
number concentrations and PM2.5 in a school: influence of outdoor air pollution on indoor air.    
Environmental Science and Pollution Research,  17(6):1268-1278.  
Health Effects Institute.  2017.  State of Global Air 2017.  Special Report. Boston, MA: Health Effects 
Institute. 
Hersey, S.P., Garland, R.M., Crosbie, E., Shingler, T., Sorooshian, A., Piketh, S. and Burger R.  2015.  
An overview of regional and local characteristics of aerosols in South Africa using satellite, ground, 
and modelling data.  Atmospheric Chemistry and Physics.  15:4259-4279. 
Hirsikko, A., Vakkari, V., Tiitta, P., Manninen, H. E., Gagné, S., Gagné, S., Laakso, H., Kulmala, M., 
Mirme, A., Mirme, S., Mabaso, D., Beukes, J. P. & Laakso, L. 2012.  Characterisation of sub-micron 
particle number concentrations and formation events in the western Bushveld Igneous Complex, 
South Africa.   Atmospheric Chemistry and Physics,  12(9):3951–3967.  
100 
 
Hoek, G., Beelen, R., de Hoogh, K., Vienneau, D., Gulliver, J., Fischer, P. & Briggs, D. 2008.   A 
review of land-use regression models to assess spatial variation of outdoor air pollution.  Atmospheric 
Environment,  42(33):7561–7578.  
Huang, L., Pu, Z., Li, M. & Sundell, J. 2015.  Characterizing the indoor-outdoor relationship of fine 
particulate matter in non-heating season for urban residences in Beijing.  PLoS One,  10(9):1–17.  
Iwai, K., Mizuno, S., Miyasaka, Y. & Mori, T. 2005.  Correlation between suspended particles in the 
environmental air and causes of disease among inhabitants: Cross-sectional studies using the vital 
statistics and air pollution data in Japan.  Environmental Research,  99(1):106–117.  
Janssen, N. a H., Lanki, T., Hoek, G., Vallius, M., de Hartog, J. J., Van Grieken, R., Pekkanen, J.& 
Brunekreef, B. 2005.  Associations between ambient, personal, and indoor exposure to fine 
particulate matter constituents in Dutch and Finnish panels of cardiovascular patients.  Occupational 
and environmental medicine,  62(12):868–877.  
Jantunen, M. J. 2007. Effect of outdoor generated pollutants on indoor air quality and health.   Paper 
presented at Clima 2006 Well Being, Key-note lectures, Kuopio, Finland. 
http://www.irbnet.de/daten/iconda/CIB6901.pdf  Date of access: 28 Oct. 2017. 
Jantunen, M.J., Hanninen, O., Katsouyanni, K., Knoppel, H., Kuenzli, N., Lebret, E., Maroni, M., 
Saarela, K., Sram, R. & Zmirou, D.  1998.  Air Pollution exposure in Eurpoean cities: the 
“EXPOLIS” study.  Journal of Exposure Analysis and Environmental Epidemiology,  8:495-518.  
Jones, N. C., Thornton, C. a., Mark, D. & Harrison, R. M. 2000.   Indoor/outdoor relationships of 
particulate matter in domestic homes with roadside, urban and rural locations.  Atmospheric 
Environment,  34(16):2603–2612.  
Josipovic, M., Annegarn, H. J., Kneen, M. A., Pienaar, J. J. & Piketh, S. J. 2010. Concentrations, 
distributions and critical level exceedance assessment of SO2, NO2 and O3 in South Africa.  
Environmental Monitoring and Assessment,  171(1–4):181–196.  
Jury, M. R. 2017.  Statistics and Meteorology of Air Pollution Episodes over the South African 
Highveld Based on Satellite–Model Datasets.  Journal of Applied Meteorology and Climatology,  
56(6):1583–1594. 
101 
 
Kimemia, D. & Annegarn, H. 2011.  An urban biomass energy economy in Johannesburg, South Africa.  
Energy for Sustainable Development,  15(4):382–387. 
Kimemia, D., Annegarn, H., Makonese, T., Molapo, V. & Robinson, J. 2016.  Characterisation of 
Domestic Biomass Combustion Technologies Used in Setswetla, Alexandra Township, Gauteng.  
Retrieved from 
https://www.researchgate.net/publication/228535714_CHARACTERISATION_OF_DOMESTIC_BI
OMASS_COMBUSTION_TECHNOLOGIES_USED_IN_SETSWETLA_ALEXANDRA_TOWNS
HIP_GAUTENG  
Krall, J.R., Anderson, B., Dominici, F., Bell, M.L. & Peng, R.D.  2013.  Short-term Exposure to 
Particulate Matter Constituents and Mortality in a National Study of U.S. Urban Communities.  
Environmental Health Perspecitves,  121(1):1148-1153.  
Kroon, B. Van Der, Brouwer, R. & Beukering, P. Van 2011.  The energy ladder : Theoretical myth or 
emperical truth ? Results from a meta-analysis.  Amsterdam: IVM Institute for Environmental 
Studies.  
Language, B., Piketh, S. J. & Burger, R. P. 2015.   Correcting respirable photometric particulate 
measurements using a gravimetric sampling method.  Clean Air Journal,  26(1):10–14.  
Lim, S. et al. 2012.  A comparative risk assessment of burden of disease and injury attributable to 67 risk 
factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden 
of Disease Study 2010.  The Lancet,  380(9859):2224–2260.  
Lin, G., Fu, J., Jiang, D., Hu, W., Dong, D., Huang, Y. & Zhao, M. 2013.  Spatio-temporal variation of 
PM2.5 concentrations and their relationship with geographic and socioeconomic factors in China.  
International Journal of Environmental Research and Public Health,  11(1):173–186.  
Long, C. M. & Sarnat, J. A. 2004.  Indoor-Outdoor Relationships and Infiltration Behavior of Elemental 
Components of Outdoor PM 2.5 for Boston-Area Homes.  Aerosol Science and Technology, 38:91–
104. 
Lozano, N., Urós, A. & Carceller, X. 2007.  Variability of atmospheric particulate matter at an urban and 
a regional background in NE Iberian Peninsula.  Barcelona: University of Barcelona.   
102 
 
Makonese, T., Masekameni, D.M., Annegarn, H.J. & Forbes P.B.C.  2017.  Emission factors of domestic 
coal-burning braziers.  South African Journal of Science,  113(3/4):1-11. 
Masekameni, D. 2017.  Influence of fire-ignition methods and stove ventilation rates on gaseous and 
particle emissions from residential coal braziers.  Journal of Energy in Southern Africa,  26(4):16-28.  
Massey, D., Masih, J., Kulshrestha, A., Habil, M. & Taneja, A. 2009.  Indoor/outdoor relationship of 
fine particles less than 2.5 um (PM2.5) in residential homes locations in central Indian region.  
Building and Environment  ,44(10):2037–2045.  
Mdluli, T. N. 2007.  The societal dimensions of domestic coal combustion: People’s perceptions and 
indoor aerosol monitoring. Johannesburg: University of the Witswatersrand.  (Thesis - PhD). 
Mehta, S. & Shahpar, C. 2004.  The health benefits of interventions to reduce indoor air pollution from 
solid fuel use: a cost-effectiveness analysis.  Energy for Sustainable Development  , 8(3):53–59.  
Meng, Q. Y., Turpin, B. J., Korn, L., Weisel, C. P., Morandi, M., Colome, S., Zhang, J. (Jim), Stock, T., 
Spektor, D., Winer, A., Zhang, L., Lee, J. H., Giovanetti, R., Cui, W., Kwon, J., Alimokhtari, S., 
Shendell, D., Jones, J., Farrar, C. & Maberti, S. 2005.   Influence of ambient (outdoor) sources on 
residential indoor and personal PM2.5 concentrations: Analyses of RIOPA data.  Journal of Exposure 
Analysis and Environmental Epidemiology,  15(1):17–28.  
Mkoma, S. L. & Mjemah, I. C. 2011.  Influence of Meteorology on Ambient Air Quality in Morogoro , 
Tanzania.   International Journal of Environmental Sciences, 1(6):1107–1115. 
Mohammadyan, M., Alizadeh-Larimi, A., Etemadinejad, S., Latif, M. T., Heibati, B., Yetilmezsoy, K., 
Abdul-Wahab, S. A. & Dadvand, P. 2017.  Particulate air pollution at schools: Indoor-outdoor 
relationship and determinants of indoor concentrations.  Aerosol and Air Quality Research,  
17(3):857–864.  
Morakinyo, O. M., Adebowale, A. S., Mokgobu, M. I. & Mukhola, M. S. 2017.  Health risk of 
inhalation exposure to sub-10 µm particulate matter and gaseous pollutants in an urban-industrial area 
in South Africa : an ecological study.  BMJ Open,  7:1–10.  
Morawska, L., He, C., Hitchins, J., Gilbert, D. & Parappukkaran, S. 2001.   The relationship between 
indoor and outdoor airborne particles in the residential environment.  Atmospheric Environment,   
35(20):3463–3473.  
103 
 
Msukaligwa Local Municipality.  2010.  Msukaligwa Local Municipality Spatial Development 
Framework. Final Report.  Ermelo: The Municipality. 
Nakai, S. & Tamura, K. 2008.  Relationship between indoor and outdoor particulate matter 
concentrations in Japan.  Asian Journal of Atmospheric Environment,  2(1): 68–74.  
Nkosi, N., Piketh, S. J., Burger, R. P. & Annegarn, H. J. 2017.  Variability of domestic burning habits in 
the South African Highveld : A case study in the KwaDela Township.  Paper presented at 2017 
International Conference on the Domestic Use of Energy (DUE), Cape Town, 3 Apr 2017.  
Norman, R., Barnes, B., Mathee, A. & Bradshaw, D. 2007.  Estimating the burden of disease attributable 
to indoor air pollution from household use of solid fuels in South Africa in 2000.   South African 
Medical Journal,  97(8):764–771. 
Oglesby, L., Künzli, N., Röösli, M., Braun-Fahrländer, C., Mathys, P., Stern, W., Jantunen, M. and 
Kousa, A.  2000.   Validity of ambient levels of fine particles as surrogate for personal exposure to 
outdoor air pollution - Results of the European EXPOLIS-EAS Study (Swiss Center Basel). Journal 
of the Air & Waste Management Association,  50(7):1251–1261.  
Oluwole, O., Arinola, G. O., Ana, G. R., Wiskel, T., Huo, D., Olopade, O. I. & Olopade, C. O. 2013.   
Relationship between Household Air Pollution from Biomass Smoke Exposure, and Pulmonary 
Dysfunction, Oxidant-Antioxidant Imbalance and Systemic Inflammation in Rural Women and 
Children in Nigeria.   Global Journal of Health Science,  5(4):28–38.  
Orru, H., Mikola, A., Upan, M. & Koiv, T. 2014.   Variation of Indoor / Outdoor Particulates in Tallinn , 
Estonia – the Role of Ventilation , Heating Systems and Lifestyle.  Journal of Environment Pollution 
and Human Health,   2(2):52–57.  
Pandey, R. 2012.  Disease burden of fuelwood combustion pollutants in rural households of the 
Himalayas, India.  Italian Journal of Public Health,   9(1):71–79.  
Payne-Sturges, D. C., Burke, T. a., Breysse, P., Diener-West, M. & Buckley, T. J. 2004.  Personal 
exposure meets risk assessment: A comparison of measured and modeled exposures and risks in an 
urban community.  Environmental Health Perspectives,  112(5):589–598.  
Petzer, G. 2009.  Indoor Air Quality Issues in South Africa. Paper Presented at the 2009 National 
Association for Clean Air Conference, South Africa, Oct 2009.  
104 
 
Pinto, J.P., Lefohn, A.S. & Shadwick, D.S. 2004.  Spatial Variability of PM2.5 in Urban Areas in the 
United States.  Journal of the Air and Waste Management Association,  54:440-449.  
Rainham, D., Krewski, D., Mcdowell, I., Sawada, M. & Liekens, B. 2008.   Development of a wearable 
global positioning system for place and health research. International Journal of Health Geographics,  
7:59. 
Raja, S., Chandrasekaran, S. R., Lin, L., Xia, X., Hopke, P. K. & Valsaraj, K. T. 2017.  Analysis of beta 
attenuation monitor filter rolls for particulate matter speciation.  Aerosol and Air Quality Research,  
17(1):14–23.  
Rajagopalan, S. & Brook, R. D. 2012.   Indoor-outdoor air pollution continuum and CVD burden: An 
opportunity for improving global health.  Global Heart, 7(3):207–213.  
Riekert, J. W. 2011.  Projecting the External Health Costs of a Coal-Fired Power Plant: The Case of 
Kusile.  Working Papers,  23(4):52–66. 
Rosenthal, J., Balakrishnan, K., Bruce, N., Chambers, D., Graham, J., Jack, D., Kline, L., Masera, O., 
Mehta, S., Mercado, I. R., Neta, G., Pattanayak, S., Puzzolo, E., Petach, H., Punturieri, A., 
Rubinstein, A., Sage, M., Sturke, R., Shankar, A., Sherr, K., Smith, K. & Yadama, G. 2017.  
Implementation science to accelerate clean cooking for public health.   Environmental Health 
Perspectives,  125(1):A3–A7.  
Sasol.  2015.  Aerosol Source Apportionment at KwaDela Township, Mpumalanga, South Africa.  
Johannesburg: Sasol. (Unpublished). 
Scorgie, Y., Kneen, M. a, Annegarn, H. J. & Burger, L. W. 2003.   Air pollution in the Vaal Triangle - 
Quantifying source contributions and identifying cost-effective solutions.  Clean Air Journal,  
13(1):5–18.  
Scorgie, Y. & Thomas, R. G. 2006.   Air Pollution Compliance Assessment and Health Risk Analysis of 
Cumulative Operations of Current , RTS and Proposed Eskom Power Station Located within the 
Mpumalanga and Gauteng Provinces. Project Done on Behalf of Eskom Holdings SOC Ltd.  
(Unpublished).  
105 
 
Shilton, V., Giess, P., Mitchell, D. & Williams, C. 2002.  The Relationships between Indoor and 
Outdoor Respirable Particulate Matter: Meteorology, Chemistry and Personal Exposure.  Indoor Built 
Environment,  11:266-274. 
Song, P., Wanga, L., Hui, Y. & Li, R. 2015. PM2.5 Concentrations Indoors and Outdoors in Heavy Air 
Pollution Days in winter. Precedia Engineering,  121:1902-1906. 
Smith, K. R. 1993.  Fuel Combustion, Air Pollution Exposure, and Health: The Situation In Developing 
Countries.  Annual Reviews Energy Environment,  18:529–566.  
Smith, K.R. 2000. National burden of disease in India from indoor air pollution. Proceedings of the 
National Academy of Sciences of the United States of America,  97(24):13286-13293. 
Smith, K. R. & Mehta, S. 2003.   The burden of disease from indoor air pollution in developing 
countries: comparison of estimates.  International journal of hygiene and environmental health,  
206(4–5):279–289.  
Song, P., Wanga, L., Hui, Y. & Li, R. 2015.  PM2.5 Concentrations Indoors and Outdoors in Heavy Air 
Pollution Days in Winter.  Procedia Engineering,  121:1902–1906.  
South Africa. 1996. Constitution of the Republic South Africa. Cape Town: Government Printer. 
South Africa.  Department of Environmental Affairs. 2007. Declaration of the Highveld as Priority Area 
in terms of Section 18(1) of the National Environmental Management: Air Quality Act, 20014, Act 
No 39 of 2004. (Government notice no. R1123).  Government gazette, 30518:3, Nov 2007. Pretoria: 
Government Printer. 
South Africa. Department of Environmental Affairs. 2009. National Ambient Air Quality Standards. 
(Government notice no. R1210).  Government gazette, 32816:6, Dec 2009. Pretoria: Government 
Printer. 
South Africa. Department of Environmental Affairs. 2012a.  A Survey of Energy-related Behaviour and 
Perceptions in South Africa.  The Residential Sector.  Pretoria: The Department of Energy. 
South Africa.  Department of Environmental Affairs. 2012b.  National Ambient Air Quality Standards 
for Particulate Matter with Aerodynamic Diameter less than 2.5 Micro Meters (PM2.5).  
106 
 
(Government notice no. R486).  Government gazette, 35463:7, Jun 2012. Pretoria: Government 
Printer. 
South Africa. Department of Environmental Affairs. 2016.  Draft Strategy to Address Air Pollution in 
Dense Low-Income Settlements, 2016.  (Government notice no.  R356).  Government gazette, 
40088:1008, 24 Jun. Pretoria: Government Printer. 
South Africa. 2016.   The Medium-Term Review of the 2011 Highveld Priority Area (HPA): Air Quality 
Management Plan. Draft Review Report.  Pretoria: The Department. 
South Africa. Department of Health. 2017. Guideline for Monitoring Domestic Indoor Air Quality. A 
guide for South African health practitioners. Unpublished. 
Statistics South Africa.  2012a.  Census 2011.  Census in Brief.  Pretoria:  Statistics South Africa 
Statistics South Africa.  2012b.  Census 2011.  Statistical Release.  Pretoria:  Statistics South Africa. 
Statistics South Africa.  2015.  Census 2011: Income Dynamics and Poverty Status of Households in 
South Africa. Pretoria: Statistics South Africa. 
Taylor, E. T., Wirmvem, M. J., Sawyerr, V. H. & Nakai, S. 2015.  Diurnal Concentrations and Variation 
of Carbon Monoxide in Indoor and Outdoor Air of Residential Homes in Western Sierra Leone.  
Environment and Pollution,   4(3):10-18.  
Terblanche, P., Opperman, L., Nel, C. M. E., Reinach, S. G., Tosen, G. & Cadman, A. 1992a.   
Preliminary results of exposure measurements and health effects of the Vaal Triangle Air Pollution 
Health Study.  South African Medical Journal,  81(11):550–556. 
Terblanche, P., Nel, R., Reinach, G. & Opperman, L. 1992b.  Personal exposures to total suspended 
particulates from domestic coal burning in South Africa.  The Clean Air Journal, 8(6):15–17. 
Thatcher, T. L. & Layton, D. W. 1995.  Deposition, resuspension, and penetration of particles within a 
residence.  Atmospheric Environment,  29(13):1487–1497.  
Thomas, E., Wickramasinghe, K., Mendis, S., Roberts, N. & Foster, C. 2015.  Improved stove 
interventions to reduce household air pollution in low and middle-income countries: a descriptive 
systematic review.  BMC Public Health,  15(1):1-15. 
107 
 
Tyson, P.D.  1988.  Transport and recirculation of aerosols off southern Africa – Macroscale plume 
structure.  Atmospheric Environment,  32(9):1511-1524. 
Umoh, V., Peters, E., Erhabor, G., Ekpe, E. and Ibok, A.  2013.  Indoor air pollution and respiratory 
symptoms among fishermen in the Niger delta of Nigeria.  African Journal of Respiratory Medicine,  
9(1):17–21.  
U.S. EPA. Guidelines for Exposure Assessment.  1992.  Risk Assessment Forum. U.S. Environmental 
Protection Agency, Washington, DC. 
U.S. EPA. Exposure Factors Handbook.  2011.  Edition (Final Report). U.S. Environmental Protection 
Agency, Washington, DC. 
Vette, A. F., Rea, A. W., Lawless, P. A., Rodes, C. E., Evans, G., Highsmith, V. R. & Sheldon, L. 2001.  
Characterization of Indoor-Outdoor Aerosol Concentration Relationships during the Fresno PM 
Exposure Studies.  Aerosol Science and Technology,   34(1):118–126.  
Wang, Y. Q., Zhang, X. Y., Sun, J. Y., Zhang, X. C., Che, H. Z. & Li, Y. 2015.  Spatial and temporal 
variations of the concentrations of PM10, PM2.5 and PM1 in China.  Atmospheric Chemistry and 
Physics,  15(23):13585–13598. 
Wichmann, J. 2006.  Air pollution epidemiological studies in South Africa : Need for freshening up. 
Reviews on Environmental Health,  20(4):265–301. 
Wilson, W. E., Mage, D. T. & Grant, L. D. 2000.  Estimating separately personal exposure to ambient 
and nonambient particulate matter for epidemiology and risk assessment: why and how. Journal of 
the Air & Waste Management Association,  50(7):167–1183.  
Wilson, J. G. 2006. Spatial variability of intraurban particulate air pollution: Epidemiological 
implications and applications. Canterbury: University of the Canterbury.  (Thesis - PhD). 
World Health Organisation.  2006.  Fuel for Life. Household Energy and Health.  Geneva:  World 
Health Organisation. 
World Health Organization.  2016.  Ambient (outdoor) air quality and health.  Fact sheet.  
http://www.who.int/mediacentre/factsheets/fs313/en/ Date of access: 23 Sep. 2017. 
Xu, G., Limin, J., Suil, Z. & Jiaqi, C. 2016. Spatial and Temporal Variability of PM2.5 
108 
 
Concentration in China.  Wuhan University Journal of Natural Sciences, 21(4):358-368. 
Yadav, R., Sahu, L. K., Jaaffrey, S. N. A. & Beig, G. 2014.  Temporal variation of Particulate Matter 
(PM) and potential sources at an urban site of Udaipur in Western India.  Aerosol and Air Quality 
Research,  14(6):1613–1629.  
Yangyang, X., Bin, Z., Lin, Z. & Rong, L. 2015.  Spatiotemporal variations of PM2.5 and PM10 
concentrations between 31 Chinese cities and their relationships with SO2, NO2, CO and O3.  
Particuology,  20:141–149.  
Zhang, J.J. & Smith, K.R. 2007. Household Air Pollution from Coal and Biomass Fuels in China: 
Measurements, Health Impacts, and Interventions.  Environmental Health Perspectives,  115(6):848-
855. 
Zhao, L., Chen, C., Wang, P., Chen, Z., Cao, S., Wang, Q., Xie, G., Wang, Y., Wang, Y. & Lu, B. 2015. 
Influence of atmospheric fine particulate matter (PM2.5) pollution on indoor environment during 
winter in Beijing.  Building and Environment,  87:283-291. 
Zubkova, G.  2003.  Analysis of Ambient Fine Particulate Matter, PM2.5, in Pittsburgh using Time-
series Techniques and Meteorology. Ohio: University of Ohio. (Thesis - PhD). 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
109 
 
 Annexures 
 
7.1 Annexure A –  Hourly maximum concentrations measured per case in KwaDela and 
KwaZamokuhle per campaign 
 
 
 
The maximum hourly value for each day for which measurements existed was derived and plotted for 
every PM type for every case for every seasonal sampling campaign. Indoor and personal hourly 
maximum PM concentrations were, where possible, compared to the maximum hourly value 
concentrations for the same hour of the day for ambient PM concentrations. Indoor and personal PM4 
measurements do not exist for every house for every season, and hence not every graph shown in this 
section of the report will display results for the same set and the same number of houses. The terms “house” 
and “case” may be used interchangeably in this section.  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 7.1.  Hourly maxima per PM type compared at household level KwaDela Winter 2013 
 
111 
 
 
Figure 7.2.  Hourly maxima per PM type compared at household level KwaDela Summer 2014 
112 
 
 
Figure 7.3.  Hourly maxima per PM type compared at household level KwaDela Winter 2014 
 
113 
 
 
Figure 7.4.  Hourly maxima per PM type compared at household level KwaDela Summer 2015 
 
114 
 
 
Figure 7.5.  Hourly maxima per PM type compared at household level KwaZamokuhle Spring 2015 
 
115 
 
 
Figure 7.6.  Hourly maxima per PM type compared at household level KwaZamokuhle Summer 2016 
 
116 
 
 
Figure 7.7.  Hourly maxima per PM type compared at household level KwaZamokuhle Winter 2016 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7.2 Annexure B – Regression analyses between ambient, indoor and personal PM 
concentrations at a community level 
 
 
 
 
Figure 7.8.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of that 
same regression analysis when plotting overarching, community-level daily average particulate matter 
concentrations in KwaDela 
118 
 
 
Figure 7.9.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of that 
same regression analysis when plotting overarching, community-level daily maximum particulate matter 
concentrations in KwaDela 
 
119 
 
 
Figure 7.10.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of 
that same regression analysis when plotting overarching, community-level hourly average particulate 
matter concentrations in KwaDela 
 
 
120 
 
 
Figure 7.11.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of 
that same regression analysis when plotting overarching, community-level hourly maximum particulate 
matter concentrations in KwaDela 
 
 
 
 
121 
 
 
Figure 7.12.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of 
that same regression analysis when plotting overarching, community-level daily average particulate 
matter concentrations in KwaZamokuhle 
122 
 
 
Figure 7.13.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of 
that same regression analysis when plotting overarching, community-level daily maximum particulate 
matter concentrations in KwaZamokuhle 
 
123 
 
 
Figure 7.14.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of 
that same regression analysis when plotting overarching, community-level hourly average particulate 
matter concentrations in KwaZamokuhle 
124 
 
 
Figure 7.15.  Scatter plot showing the R2 value of a regression analysis against the slope of the line of 
that same regression analysis when plotting overarching, community-level hourly maximum particulate 
matter concentrations in KwaZamokuhle 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
125 
 
7.3 Annexure C – Overview of personal PM4 concentrations measured per micro-
environment 
 
 
Table 7.1.  Overview of personal PM4 concentrations measured per micro-environment (M-E) identified 
between 21 August 2013 and 28 August 2013 (I=Inside a house; DO=Directly outside a house; DR=On 
a dirt road; TR=On a tar road; OF=On an open field) 
 
21 August 
2013 
22 August 
2013 
23 August 
2013 
24 August 
2013 
25 August 
2013 
26 August 
2013 
27 August 
2013 
28 August 
2013 
Avera
ge 
time  
Averag
ed PM4 
(µg/m³) 
M
- 
E 
Averag
ed PM4 
(µg/m³) 
M 
- 
E 
Averag
ed PM4 
(µg/m³) 
M
- 
E 
Averag
ed PM4 
(µg/m³) 
M
- 
E 
Averag
ed PM4 
(µg/m³) 
M
- 
E 
Averag
ed PM4 
(µg/m³) 
M
- 
E 
Averag
ed PM4 
(µg/m³) 
M
- 
E 
Averag
ed PM4 
(µg/m³) 
M
- 
E 
8:30-  
8:40 
                            148.72 D
O 
8:40-  
8:50 
                            1441.44 D
O 
8:50-  
9:00 
        108.68 D
O 
                158.73 D
O 
9:00-  
9:10 
    2.86 D
O 
88.66 I             50.77 D
O 
    
9:10-  
9:20 
    3.58 O 76.51 D
O 
114.40 D
O 
        50.77 D
O 
148.72 I 
9:20-  
9:30 
    3.58 I 96.53 D
O 
87.95 D
O 
        45.76 D
O 
  
9:30-  
9:40 
    4.29 D
O 
127.27 I                   
9:40 -
9:50 
    64.35 I 85.80 T
R 
                56.49 I 
9:50-
10:00 
        35.04 I     699.99 I         53.63 I 
10:00-
10:10 
        38.61 D
O 
    62.92 I         52.20 I 
10:10-
10:20 
                        49.34 D
O 
    
10:20-
10:30 
                    50.05 D
O 
65.78 D
O 
    
10:30-
10:40 
        37.90 I         50.05 D
O 
        
10:40-
10:50 
        37.90 I                 56.49 I 
10:50-
11:00 
        67.21 D
R 
                53.63 D
O 
11:00-
11:10 
        70.07 I                 52.20 D
O 
11:10-
11:20 
        75.79 I                 50.05 D
O 
11:20-
11:30 
        106.54 I 100.10 I 72.22 I             
11:30-
11:40 
        115.12 D
O 
100.82 I 61.49 I 52.20 D
O 
    54.34 D
O 
11:40-
11:50 
        105.11 D
O 
        47.91 D
O 
    51.48 D
O 
11:50-
12:00 
    2.86 D
R 
    199.49 I     52.91 I     59.35 D
O 
12:00-
12:10 
    3.58 D
O 
148.72 I 103.68 D
O 
62.92 D
O 
        75.08 O
F 
12:10-
12:20 
    3.58 D
O 
114.40 D
O 
77.22 I 42.90 D
O 
129.42 I 47.91 D
O 
50.05 I 
12:20-
12:30 
    4.29 D
O 
87.95 D
O 
64.35 I 42.90 D
O 
94.38 I     50.05 I 
12:30-
12:40 
    64.35 I 78.65 D
O 
61.49 D
O 
71.50 D
O 
        51.48 I 
 
126 
 
 21 August 
2016 
22 August 
2016 
23 August 
2016 
24 August 
2016 
25 August 
2016 
26 August 
2016 
27 August 
2016 
28 August 
201 
Avera
ge 
time  
Averag
ed PM4 
(µg/m³) 
M
-E 
Averag
ed PM4 
(µg/m³) 
M
-E 
Averag
ed PM4 
(µg/m³) 
M
-E 
Averag
ed PM4 
(µg/m³) 
M
- 
E 
Averag
ed PM4 
(µg/m³) 
M 
- 
E 
Averag
ed PM4 
(µg/m³) 
M 
- 
E 
Averag
ed PM4 
(µg/m³) 
M 
- 
E 
Average
d PM4 
(µg/m³) 
M 
- 
E 
12:40-
12:50 
    4.29 D
O 
75.79 I 58.63 I 40.76 I 129.42 I 44.33 I 51.48 I 
12:50-
13:00 
    4.29 I 77.22 D
O 
62.92 I 178.04 D
O 
94.38 D
O 
47.19 I     
13:00-
13:10 
    8.58 I 74.36 D
O 
    35.75 D
O 
        55.06 I 
13:10-
13:20 
    12.16 I 78.65 I     56.49 I 49.34 D
O 
    52.91 I 
13:20-
13:30 
    12.16 I 84.37 I     72.22 I 69.36 I         
13:30-
13:40 
    19.31 D
O 
96.53 I     61.49 I 60.78 I 50.05 D
O 
    
13:40-
13:50 
        101.53 I     48.62 I 61.49 I         
13:50-
14:00 
    12.16 I 96.53 D
O 
    45.05 D
O 
48.62 D
O 
        
14:00-
14:10 
    12.16 I 95.81 D
O 
178.04 D
O 
44.33 D
O 
50.77 D
O 
45.76 I     
14:10-
14:20 
        102.25 I 35.75 D
O 
82.23 I 50.77 I         
14:20-
14:30 
    7.15 D
O 
100.10 I 56.49 D
O 
    45.76 D
O 
        
14:30-
14:40 
    2.86 I 100.82 I 72.22 D
O 
100.10 I 45.76 D
O 
62.92 I     
14:40-
14:50 
        199.49 I 61.49 D
O 
97.24 I 42.90 D
O 
55.77 I     
14:50-
15:00 
    77.22 I 103.68 I 48.62 D
O 
100.82 I 48.62 D
O 
        
15:00-
15:10 
        77.22 I 45.05 D
R 
75.79 D
O 
46.48 D
O 
        
15:10-
15:20 
    1.43 I 64.35 I 44.33 I 70.07 D
O 
48.62 D
O 
        
15:20-
15:30 
    2.15 I 61.49 O 82.23 I 56.49 D
O 
49.34 D
O 
        
15:30-
15:40 
        58.63 O     175.18 D
O 
69.36 I         
15:40-
15:50 
    3.58 I 62.92 O     51.48 D
O 
60.78 I         
15:50-
16:00 
    3.58 D
O 
699.99 O     40.76 D
O 
            
16:00-
16:10 
        62.92 O     41.47 D
O 
            
16:10-
16:20 
27.17 D
O 
1.43 D
O 
        47.19 D
O 
            
16:20-
16:30 
27.17 D
O 
3.58 I         188.05 D
O 
    59.35 D
O 
    
16:30-
16:40 
40.76 D
O 
    70.07 D
O 
    171.60 D
O 
    61.49 D
O 
    
16:40-
16:50 
33.61 D
O 
20.02 I         135.85 D
O 
61.49 D
R 
117.98 I     
16:50-
17:00 
31.46 I 60.06 I 106.54 D
O 
64.35 IF 106.54 D
O 
    587.02 I     
17:00-
17:10 
30.03 I                             
17:10-
17:20 
                                
17:20-
17:30 
                                
17:30-
17:40 
                289.58 D
O 
            
18:00-
18:10 
                2780.64 D
O 
            
 
127 
 
7.4 Annexure D – Integrated exposure concentration and integrated potential dose 
calculation  
 
Table 7.2. Derivation of time-weighted, integrated exposure concentrations and potential doses for an 
individual carrying a SidePak and GPS monitor between 21 and 28 August 2013 (U.S. EPA, 2011) 
 
 
21 August 2013 
Micro-environment  Average PM4 
concentration (C 
µg/m3) 
Time 
fraction 
(t)* 
C x t 
(µg/
m3) 
Microenviron
ment 
contribution 
(%) 
Potenti
al dose 
(µg/ 
day) 
PD*t 
(µg/d
ay) 
PD*t 
(mg/da
y) 
Inside a house (I) 189.64 0.33 63.21 73% 3276.9
8 
1092.
33 
1.0923
26 
Directly outside a 
house (DO) 
35.21 0.67 23.47 27% 608.43 405.6
2 
0.4056
19 
On a dirt road (DR) 0.00 0.00 0.00 0% 0.00 0.00 0 
On a tar road (TR) 0.00 0.00 0.00 0% 0.00 0.00 0 
On an open field (OF) 0.00 0.00 0.00 0% 0.00 0.00 0 
  Time-weighted exposure 
concentration 
86.69 
 
Time-weighted 
dose 
1.50 
 
 
22 August 2013 
Micro-environment  Average PM4 
concentration (C 
ug/m3) 
Time 
fraction 
(t)* 
C x t 
(ug/
m3) 
Microenviron
ment 
contribution 
(%) 
Potentia
l dose 
(µg/ 
day) 
PD*t 
(µg/d
ay) 
PD*t 
(mg/d
ay) 
  
       
Inside a house (I) 259.37 0.59 152.0
4 
91% 4481.91 2627.
33 
2.63 
Directly outside a 
house (DO) 
36.78 0.38 13.95 8% 635.56 241.0
7 
0.24 
On a dirt road (DR) 36.78 0.03 1.27 1% 635.56 21.92 0.02 
On a tar road (TR) 0.00 0.00 0.00 0% 0.00 0.00 0.00 
On an open field (OF) 0.00 0.00 0.00 0% 0.00 0.00 0.00 
  Time-weighted exposure 
concentration 
167.2
6 
 
Time-weighted 
dose 
2.89 
 
 
 
 
 
 
 
128 
 
23 August 2013 
Micro-environment  Average PM4 
concentration C 
(ug/m3) 
Time 
fraction 
(t)* 
C x t 
(ug/
m3) 
Microenviron
ment 
contribution 
(%) 
Potenti
al dose 
(µg/ 
day) 
PD*t 
(µg/d
ay) 
PD*t 
(mg/d
ay) 
  
       
Inside a house (I) 207.70 0.49 101.4
3 
84% 3589.06 1752.
79 
1.75 
Directly outside a 
house (DO) 
39.14 0.47 18.20 15% 676.34 314.5
8 
0.31 
On a dirt road (DR) 39.14 0.02 0.91 1% 676.34 15.73 0.02 
On a tar road (TR) 39.14 0.02 0.91 1% 676.34 15.73 0.02 
On an open field (OF) 0.00 0.00 0.00 0% 0.00 0.00 0.00 
  Time-weighted exposure 
concentration 
121.4
6 
 
Time-weighted 
dose 
2.10 
 
 
24 August 2013 
Microenvironment  Average PM4 
concentration (C 
ug/m3) 
Time 
fraction 
(t)* 
C x t 
(ug/
m3) 
Microenviron
ment 
contribution 
(%) 
Potenti
al dose 
(µg/ 
day) 
PD*t 
(µg/da
y) 
PD*t 
(mg/da
y) 
  
       
Inside a house (I) 147.80 0.43 63.34 73% 2553.9
8 
1094.
56 
1.0945
65 
Directly outside a 
house (DO) 
40.07 0.48 19.08 22% 692.41 329.7
2 
0.3297
19 
On a dirt road (DR) 40.07 0.05 1.91 2% 692.41 32.97 0.0329
72 
On a tar road (TR) 0.00 0.00 0.00 0% 0.00 0.00 0 
On an open field (OF) 40.07 0.05 1.91 2% 692.41 32.97 0.0329
72 
  Time-weighted exposure 
concentration 
86.24 
 
Time-weighted 
dose 
1.49 
 
 
 
 
 
 
 
 
 
 
 
 
 
129 
 
25 August 2013 
Microenvironment  Average PM4 
concentration (C 
ug/m3) 
Time 
fraction 
(t)* 
C x t 
(ug/
m3) 
Microenviron
ment 
contribution 
(%) 
Potenti
al dose 
(µg/ 
day) 
PD*t 
(µg/da
y) 
PD*t 
(mg/da
y) 
  
       
Inside a house (I) 237.24 0.37 88.12 77% 4099.5
1 
1522.
67 
1.5226
74 
Directly outside a 
house (DO) 
42.06 0.63 26.44 23% 726.80 456.8
4 
0.4568
44 
On a dirt road (DR) 0.00 0.00 0.00 0% 0.00 0.00 0 
On a tar road (TR) 0.00 0.00 0.00 0% 0.00 0.00 0 
On an open field (OF) 0.00 0.00 0.00 0% 0.00 0.00 0 
  Time-weighted exposure 
concentration 
 
114.5
6 
 
Time-weighted 
dose 
1.98 
 
 
26 August 2013 
Microenvironment  Average PM4 
concentration (C 
ug/m3) 
Time 
fraction 
(t)* 
C x t 
(ug/
m3) 
Microenviron
ment 
contribution 
(%) 
Potenti
al dose 
(µg/ 
day) 
PD*t 
(µg/da
y) 
PD*t 
(mg/d
ay) 
  
       
Inside a house (I) 86.82 0.38 33.39 62% 1500.2
5 
577.0
2 
0.58 
Directly outside a 
house (DO) 
32.67 0.58 18.85 35% 564.54 325.6
9 
0.33 
On a dirt road (DR) 32.67 0.04 1.26 2% 564.54 21.71 0.02 
On a tar road (TR) 0.00 0.00 0.00 0% 0.00 0.00 0.00 
On an open field (OF) 0.00 0.00 0.00 0% 0.00 0.00 0.00 
  Time-weighted exposure 
concentration 
 
53.50 
 
Time-weighted 
dose 
0.92 
 
 
 
 
 
 
 
 
 
 
 
 
 
130 
 
27 August 2013 
Microenvironment  Average PM4 
concentration (C 
ug/m3) 
Time 
fraction 
(t)* 
C x t 
(ug/m
3) 
Microenviron
ment 
contribution 
(%) 
Potential 
dose 
(µg/ 
day) 
PD*t 
(µg/
day) 
PD*t 
(mg/da
y) 
  
       
Inside a house (I) 124.62 0.44 54.52 73% 2153.43 942.
13 
0.94 
Directly outside a 
house (DO) 
36.50 0.56 20.53 27% 630.72 354.
78 
0.35 
On a dirt road (DR) 0.00 0.00 0.00 0% 0.00 0.00 0.00 
On a tar road (TR) 0.00 0.00 0.00 0% 0.00 0.00 0.00 
On an open field 
(OF) 
0.00 0.00 0.00 0% 0.00 0.00 0.00 
  Time-weighted exposure 
concentration 
 
75.05 
 
Time-weighted 
dose 
1.30 
 
 
28 August 2013 
Microenvironment  Average PM4 
concentration (C 
ug/m3) 
Time 
fraction 
(t)* 
C x t 
(ug/
m3) 
Microenviron
ment 
contribution 
(%) 
Potenti
al dose 
(µg/ 
day) 
PD*t 
(µg/d
ay) 
PD*t 
(mg/d
ay) 
  
       
Inside a house (I) 272.10 0.52 142.5
3 
83% 4701.9 2462.
9 
2.4628
94 
Directly outside a 
house (DO) 
60.57 0.43 25.96 15% 1046.6 448.6 0.4485
64 
On a dirt road (DR) 0.00 0.00 0.00 0% 0.0 0.0 0 
On a tar road (TR) 0.00 0.00 0.00 0% 0.0 0.0 0 
On an open field (OF) 60.57 0.05 2.88 2% 1046.6 49.8 0.0498
4 
  Time-weighted exposure 
concentration 
 
171.3
7 
 
Time-weighted 
dose 
2.96 
 
 
Notes:  
Time fraction = Fraction of time spent in each environment     
Micro-environment contribution = Percentage that each microenvironment contributes to the total time, time-weighted, 
integrated exposure (Ej) 
Time-weighted integrated exposure (E) (µg/m³) = ∑ C x t    
Potential dose (PD) = C x IR        
Time-weighted integrated potential dose E (µg/m³) = ∑ C x t    
Inhalation rate = 17.28 m3/day   
 
 
(U.S.EPA, 2011) 
131 
 
7.5 Annexure E – Daily average timeseries (after QA/QC) 
 
 
 
Figure 7.16 Timeseries of daily average PM concentrations (µg/m3) in KwaDela across the campaigns 
 
 
Figure 7.17 Timeseries of daily average PM concentrations (µg/m3) in KwaZamokuhle across the 
campaigns 
 
 
 
132 
 
 
 
 
Figure 7.18 Timeseries of ambient daily average PM concentrations (µg/m3) in KwaDela in winter 2013 
 
 
Figure 7.19  Timeseries of ambient daily average PM concentrations (µg/m3) in KwaDela in summer 
2014 
 
133 
 
 
Figure 7.20  Timeseries of ambient daily average PM concentrations (µg/m3) in KwaDela in winter 2014 
 
 
 
Figure 7.21  Timeseries of ambient daily average PM concentrations (µg/m3) in KwaDela in summer 
2015 
 
 
134 
 
 
Figure 7.22   Timeseries of ambient daily average PM concentrations (µg/m3) in KwaZamokuhle in 
spring 2015 
 
 
Figure 7.23   Timeseries of ambient daily average PM concentrations (µg/m3) in KwaZamokuhle in 
summer 2016 
 
135 
 
 
Figure 7.24   Timeseries of ambient daily average PM concentrations (µg/m3) in KwaZamokuhle in 
winter 2016