Erosion hazard assessment in the upper ewaso Ng’lro basin of Kenya: application of GIS, USLE and EUROSEM

Abstract

A methodology was developed for assessing soil erosion hazard in the Upper Ewaso Ng’iro basin of Kenya, using Geographic Information Systems (GIS), the Universal Soil Loss Equation (USLE) and the European Soil Erosion Model (EUROSEM). The USLE was used in a GTS environment by creating thematic maps of R, K, L, S, C and P and then calculating soil loss by raster-grid modelling with Arc/Info GRID. The rainfall erosivity factor (R) was derived from relationships between rainfall amount and erosivity using erosion plot data from within the catchment. The nature of the relationship was found to be a function of agro-climatic zones of the region. Mean annual erosivities ranged from 145 to 990 J m"2 hr"1 For a given amount of rainfall, erosivity was higher in zone IV than in the wetter zones 11-111. The soil erodibility factor (K) was estimated using the USLE nomograph and data from laboratory analysis of field samples collected from representative major soil mapping units. The K-values were low to medium, ranging from 0.10 to 0.25 over 84 percent of the basin. The topographic factor (LS) was obtained by creating Digital Elevation Models (DEMs) of the basin with TOPOGRIDTOOL of Arc/Info. These were then used to determine the slope steepness and length factor values, calculated with raster-grid modelling. Although DEMs proved a useful tool, maximum values of both steepness and length had to be set in this reconnaissance study to achieve reasonable results. A finer resolution of input data and a smaller grid cell size are needed for accurate determination. The cover and management factors (C) were obtained by determining the land cover types within the basin using remotely sensed data (SPOT 1 colour composite prints) and ground truthing studies. The factor values were estimated from USLE guide tables and measurements of cover from plots and test sites. Some 70 percent of the basin is covered by rangelands. The conservation practice (P) factor values were estimated from USLE guide tables and then applied to areas where soil conservation had been introduced according to maps obtained from the Ministry of Agriculture. The USLE was validated using data from erosion plots. A value of R2 = 0.645 was obtained between predicted and measured values but the standard error was rather high (e = 5.745 t ha"1 yr"1). Using an annual soil loss of 9.0 t ha"1 yr"1 as tolerance level, some 36 percent of the basin was found to experience unacceptably high erosion rates. Most of this area was communal grazing land and cropland where soil conservation measures had not been applied. A critical land cover type within the grazing land is shrubland, where vegetation cover is less than 40 percent and high erosion risk was predicted and confirmed by field surveys. EUROSEM could not be integrated within a GIS in the time available for research. It was therefore simulated outside GIS environment, where it was applied to Embori and Mukogodo plot data using separate data sets for calibration and validation. Calibration was used to obtain input parameters for saturated hydraulic conductivity, cohesion and Manning’s roughness coefficients. Validation gave correlation coefficients of 0.907 and 0.840 for predictions of storm runoff and soil loss respectively at Embori; the corresponding values for bare soil plots at Mukogodo were 0.895 and 0.577. However, EUROSEM predicted runoff poorly (R2 = 0.570) and failed to predict soil loss at all the vegetated plots at Mukogodo. The model was applied to simulated vegetation covers of barley, maize, grass and forest for a 36.7 mm rainstorm at Embori. The simulated soil losses showed an exponential decrease with increasing cover. At a threshold cover of 70 percent, soil loss diminished to zero under grass and forest and decreased to a minimum value under barley and maize. These results support the USLE simulations, which showed that areas with more than 70 percent cover (such as forest) had a low erosion hazard, even with steep slopes and high rainfall erosivities. This research has demonstrated that GIS can be used with the USLE to assess and quantify erosion hazard, giving results that can be used for conservation planning. EUROSEM can be applied successfully to bare soil and cropland, but application to other land covers requires further investigation. Land cover and topography are the main factors controlling the spatial distribution of soil loss in the Upper Ewaso Ng’iro basin. Future conservation activities should be concentrated on the rangelands.