Evaluating use of satellite imagery in mapping differences in geological fertility in the Mahikeng - Ramotshere Moiloa Local Municipalities based on vegetation cover

Abstract

In rural, predominantly communal areas of South Africa, historical factors have left a need for poverty alleviation that includes planning development through municipal spatial development frameworks. Mahikeng and Ramotshere-Moiloa Local Municipalities in the North West Province of South Africa are predominantly rural, and in need of development planning. Remote sensing can help provide data in support of the planning of development. This research aimed at establishing the extent to which differences in geological fertility can be delineated on satellite imagery, using the high spatial resolution (10m) SPOT 5 HRG and lower spatial resolution (250m) MODIS imagery as test case. March-April 2012 SPOT HRG images covering the Mahikeng and Ramotshere-Moiloa Local Municipalities were obtained from the South African National Space Agency (SANSA). Sections of twelve SPOT HRG image scenes were required to cover the two local municipalities, which presented problems in obtaining same date images due to cloud cover problems. A same period MODIS 16-day (21 March - 06 April 2012) NDVI composite image (MOD13Ql) was obtained from the Oak Ridge National Laboratory Distributed Active Archive Center (ORNL DAAC) in the USA. Rain season images were, therefore, utilised. The rain season was selected based on the fact that the rain season is the time when the vegetation (grass and trees inclusive) is at highest productivity and, therefore, differences in spectral response of the vegetation was judged to be most likely to manifest then. The images were projected to the Universal Transverse Mercator (UTM) projection and subset to extract the area covered by the two municipalities. A geological fertility map of the area was extracted from the 1: 1 million geology map of South Africa. Vegetation cover on the images was enhanced using the Normalized Difference Vegetation Index (NOVI). Field work at 41 sites in protected areas in the study area yielded ground truth training data on vegetation attributes of canopy cover, tree density, tree height ranges, grass cover, and dominant vegetation types. Field data indicated a weak but statistically significant correlation between geological fertility and canopy closure (r = 0.378, P < 0.02), which permitted the use of mapping vegetation density, using on the NDVI, as indicator of geological fertility. Supervised maximum likelihood classification was employed in the process. Using the geological map as reference data, the results showed that the higher resolution SPOT HRG images produced a more accurate classification (overall accuracy 68.3%, K = 0.63) than the MODIS image (overall accuracy 48.3%, K = 0.39). It was concluded that geological fertility can be inferred in the area on the basis of tree density, and using this characteristic, that remotely sensed imagery can contribute to agricultural development planning for poverty alleviation.