Spatial modeling of soil erosion potential in a tropical watershed of the Colombian Andes

Soil erosion potential of a 58 km(2) watershed in the coffee growing region of the Colombian Andes was assessed using the Revised Universal Soil Loss Equation (RUSLE) in a GIS environment. The RUSLE factors were developed from local rainfall, topographic, soil and land use data. Seasonal erosivity factors (R) were calculated for six pluviographic stations (1987-1997) located within 22 km of the basin. Two regression models, one for the wet and one for the dry seasons, were created and used to estimate seasonal erosivity for 10 additional stations with pluviometric data. Erosivity was on average higher in the wet seasons (4686 MJ min ha(-1) h(-1) season(-1)) than the dry ones (2599 MJ min ha(-1) h(-1) season(-1)). Seasonal erosivity surfaces were generated using the local polynomial interpolation method, and showed increases from west to east in accordance with regional elevation. Soil erodibility was calculated from field measurements of water stable aggregates (> 2 mm) and infiltration, which were influenced by land use. Three erodibility scenarios were considered (high, average and low) to represent the variability in infiltration measurements within each land use. The topographic and land cover factors were developed from existing contour and land use data. Model results indicated that in the dry seasons, and under the average erodibility scenario, 534 ha (11%) of the basin's rural area were within the extreme erosion potential category (above 3.5 t ha(-1) season(-1)). During the wet seasons, this area increased to 1348 ha (28%). In general, areas under forest and shrub had low erosion potential values, while those under coffee and pasture varied according to topography. Modeling of probable land use change scenarios indicated that the erosion potential of the basin would decrease as a result of coffee conversion to pasture. (c) 2005 Elsevier B.V. All rights reserved.