A pedogenetic method for land type survey disaggregation into soil association maps

Abstract

There is an increasing demand worldwide for spatial soil information. Unfortunately, gathering new soil data is expensive, leading to a focus on extracting detailed information from existing datasets. The most extensive dataset in South Africa is that of the land type survey. This paper proposes a method of disaggregating land type inventories into a soil association map, using a pedogenetic approach. The method is illustrated by the disaggregation of land types in landscapes with simple (Cathedral Peak) and complex (Ntabelanga) soil distribution patterns. The maps were validated with independent datasets. The Cathedral Peak map was produced with a Kappa value of 0.66, indicating a substantial agreement with reality, but the Ntabelanga soil association map only achieved a Kappa value of 0.20, indicating a slight agreement with reality. These values, comparable to results obtained using the automated disaggregated algorithm DSMART, show that the method is useful in landscapes with a simple soil distribution pattern, but not in areas where there is a complex soil distribution pattern. The pedogenetic method would be useful to soil surveyors with a moderate level of GIS skills and access to pedogenetic knowledge of the study area. Sites consisting of numerous land types are better analysed by the automated DSMART method