Potential of weight of evidence modelling for gully erosion hazard assessment in Mbire District – Zimbabwe

Abstract

Gully erosion is an environmental concern particularly in areas where landcover has been modified by human activities. This study assessed the extent to which the potential of gully erosion could be successfully modelled as a function of seven environmental factors (landcover, soil type, distance from river, distance from road, Sediment Transport Index (STI), Stream Power Index (SPI) and Wetness Index (WI)) using a GIS-based Weight of Evidence Modelling (WEM) in the Mbire District of Zimbabwe. Results show that out of the studied seven factors affecting gully erosion, five were significantly correlated (p<0.05) to gully occurrence, namely; landcover, soil type, distance from river, STI and SPI. Two factors; WI and distance from road were not significantly correlated to gully occurrence (p>0.05). A gully erosion hazard map showed that 78% of the very high hazard class area is within a distance of 250m from rivers. Model validation indicated that 70% of the validation set of gullies were in the high hazard and very high hazard class. The resulting map of areas susceptible to gully erosion has a prediction accuracy of 67.8%. The predictive capability of the weight of evidence model in this study suggests that landcover, soil type, distance from river, STI and SPI are useful in creating a gully erosion hazard map but may not be sufficient to produce a valid map of gully erosion hazard.