Modelling of soil erosion and sustainable land management measures in the Usambara Mountains, Tanzania

Abstract

Tanzania’s population growth increases the stress on arable land, making it vulnerable to erosion especially in sloped areas like the Usambara Mountains. The construction of terraces and the growing of vegetative barriers are known to reduce erosion. Sustainable Land Management (SLM) measures are often modelled to evaluate its effects on runoff and erosion, but only few researches focus on the implementation of SLM types in erosion modelling. This research aims at incorporating SLM in KINEROS2 in order to develop a method to predict the effects of construction of SLM types on runoff and erosion beforehand. KINEROS2 is a distributed, event and physics based runoff and erosion model that represents the catchment by different connected and homogenous elements. For the research, 6 farmer plots in the Usambara Mountains in Tanzania were selected with the SLM types supported being terraces, agroforestry and grass trips. Input parameters required for KINEROS2 were measured in the field or determined from literature. The spatial dependency of cohesion, Ks, bulk density and soil texture on distance to trees was determined by taking measurements from a transect between two Grevillea trees. The control plots were used for calibration of the runoff and erosion parameters Ks, G, C and S. The SLM types were then incorporated in the KINEROS2 model, by representing each riser, bed and grass strip by a single element with specific parameter values depending on slope, soil type, vegetation, etc. The incorporation of SLM was validated by comparing the model results with the actual runoff and erosion data measured using Gerlach throughs. The Gerlach data did not show the expected effect of SLM on runoff. It did show a larger increase in erosion with increase in runoff on plots with poor or without SLM; the plot with grass strips showed the smallest increase in erosion. For the transect, no significant relationship of the measured parameters with tree distance was found. Therefore trees were not represented by single elements but its vegetation parameters were averaged over the entire plot. KINEROS2 modelled runoff as was expected from the model, but performance of the model compared to actual data resulte d in moderate predictions because of the actual data not always showing the expected trend of decreasing runoff on plots with SLM. Erosion was overestimated for the plots with terraces, which was caused by the large increase in erosion on the terrace’s risers. Further research on the effect of erosion on very steep slopes and the parameters influenced, and the adaption of the KINEROS2 model for rainfall on steep slopes, is needed before modelling of terraces can be accomplished in KINEROS2.