Land Use/Cover Change and their Impacts on Streamflow in Kikuletwa Catchment of Pangani River Basin, Tanzania

Abstract

Streamflow perturbation is highly prevalent in Kikuletwa catchment. However, little is known concerning land use/cover change (LULCC) with regard to streamflow perturbation in the catchment. This study aims to detect the historical and predict future LULCC and assess their impacts on streamflow amounts using the Soil and Water Assessment Tool (SWAT) model. Supervised classification of Landsat imagery data for 1985, 2000 and 2015 years was done in ERDAS 14 Imagine software. Future prediction of LULCC was done using Module for Land Use Change Evaluation (MOLUSCE) tool, a QGIS plug-in. An accuracy ranging from 79% to 82% was obtained for all steps. The results revealed that, from 1985 to 2000; 1985 to 2015; 1985 to 2030 and 1985 to 2050 the percentage of area change in cultivated land is +21.1%; +29.2%; +38.2% and +42.7%, respectively; forest is - 2.3%, -3.1%, -3.8% and -5.8%, respectively; and shrubland is -6.3%, -10%, -15.7% and - 16%, respectively. The performance of SWAT model during calibration were 0.74, 0.75, 0.51 and -0.5% for NSE, R2, RSR and PBIAS, respectively. The impacts of LULCC indicated that, between 1985 to 2000; 1985 to 2015; 1985 to 2030 and 1985 to 2050, the percentage increase in average simulated annual flow is 4.7%, 6.8%, 12.6% and 19.3%, respectively. Surface runoff increased from 25.2 mm (baseline) to 34.5 mm (36.9%); 36.2 mm (42.4%); 41.4 mm (64.3%) and 47.6 mm (88.9%), respectively. Base flow decreased marginally from 82.2 mm (baseline) to 79 mm (-3.8%); 77.8 mm (5.4%); 75.4 mm (-8.3%) and 73.9 mm (- 10.1%), respectively. Thus, apart from climate effects, streamflow perturbation in the catchment is also related to disturbances of catchment influences such as LULCC as revealed in this study. The study is useful for land planners and water resources managers and policy makers in managing resources sustainably. 

Keywords: ERDAS, Kikuletwa catchment, Land use land cover change, MOLUSCE tool, Streamflow, SWAT model.