Epsilon Archive for Student Projects

Abstract

In the Upper Blue Nile Basin land use is rapidly changing due to agricultural development and intensification with potential effects on the water availability both up- and downstream. This study is aimed to explore the impact of agricultural land use change on a landscape water balance. The studied watershed in the Fagita Lekoma District, Ethiopia has undergone agricultural change from an annual cropping system to an afforestation system with Acacia decurrens plantations. The land use has significantly changed from dominant crop and pasture on almost 90% of the area to permanent cultivated or natural forest on >45% of the area over the last 20 years.
The soil water assessment tool (SWAT), a distributed dynamic model was set up to analyse the effects of land use change on the water balance and sediment yields. Three scenarios were run; a control, a control with improved soil water conservation (SWC) practices and the land use with the afforestation practices.
The modelling resulted in some unexpected changes on the actual evapotranspiration (ETa) and surface runoff between the control and afforestation practices. Due to a high biomass density of the plantations a higher ETa and a lower surface runoff were expected. But the A. decurrens dominated landscape showed less ETa than the control dominated by annual crops. The water balance is sensitive to the designed growth response parameters of the A. decurrens. Between the control and the control with improved SWC practices there was no significant change in the landscape water balance. The implementation of the afforestation practice and SWC practices reduced the sediment yields in the watershed from 26.5 to 15.5 t ha-1 y-1. Cropland was the land use that in all scenarios was most prone to soil erosion.
It is likely that the A. decurrens crop description has affected the modelled water balance. Therefore, next steps should be field verification of growth response parameters and water and nutrient uptake parameters. Furthermore, higher resolution input data needs to be acquired to support calibration and validation of the model.