Spatial evaluation of soil erosion risk in the West Usambara Mountains, Tanzania

Abstract

Effective soil and water conservation programmes require the concentration of resources on limited areas. For that purpose regional-scale assessments of erosion risk are required. However, availability of good-quality spatial data for such assessments is often poor, especially in regions like sub-Saharan Africa. This study was conducted to evaluate the potential of analysing regional erosion risk using a limited amount of readily available data. The analysis was executed for the 70-km2 Baga watershed in the West Usambara Mountains, Tanzania. Principal data sources were a Landsat image and an SRTM DEM. Two qualitative mapping methods resulting in five erosion risk classes were evaluated against field-based erosion risk estimates. The first method derived slope classes from the DEM and fractional vegetation cover (FVC) classes from the Landsat image and FVC field estimates. Data integration was achieved with a decision tree. The second method did not apply field data for the map construction, but combined five equally sized NDVI classes and five slope classes using the minimum-operator. Resulting maps showed a high and comparable accuracy, 80 per cent for Method 1, and 81 per cent for Method 2, when allowing a maximum difference of one class with field erosion risk estimates. Spatial patterns of erosion risk were well represented by both methods and high erosion risk areas can be properly identified within the Baga Watershed. Due to the limited data requirement of Method 2, it has a high potential for quick identification of erosion risk in other areas of the East African Highlands. Copyright © 2006 John Wiley & Sons, Ltd.