Predicting gully densities at sub-continental scales: a case study for the Horn of Africa

Gully erosion is a major cause of land degradation in many regions, due to its negative impacts on catchment
hydrology, its associated losses of land and damage to infrastructure, as well as its often major contributions to
catchment sediment yields. Mitigation and prevention of gully erosion requires a good knowledge of its spatial
patterns and controlling factors. However, our ability to simulate or predict this process remains currently very
limited. This is especially the case for the regional scale. Whereas detailed case studies have provided important
insights into the drivers of gully erosion at local scales, these findings are often difficult to upscale to larger
regions.
Here we utilized a simple and cheap method to predict patterns of gully density at the sub-continental scale. By
means of a random sampling procedure, we mapped gully densities for over sixty study sites across the Horn of
Africa, using freely available Google Earth imagery. Next, we statistically analyzed which factors best explained
the observed variation in mapped gully density. Based on these findings, we constructed a multiple regression
model that simulates gully density, based on topography (average slope), soil characteristics (percentage silt) and
land use (NDVI-value). Although our model could benefit from further refinement, it succeeds already fairly well
in simulating the patterns of gully density at sub-continental scales. Over 75% of the predicted gully densities
differ less than 5% from the observed gully density, while over 90% of the predictions deviate less than 10%.
Exploration of our results further showed that this methodology may be highly useful to quantify total gully
erosion rates at regional and continental scales as well as the contribution of gully erosion to catchment sediment
yields.