Modelling the Impact of Climate and Land Use/Land Cover Change on Streamflow and Sediment Yield in the Black Volta River Basin

Abstract

This study investigated the impacts of climate change and the sensitivity of land use/land cover (LULC) changes on river flow and sediment yield in the Black Volta River Basin (BVRB) using a modelling approach, with the aim of providing water resources and basin managers with science-based evidence of impacts of global change at basin level to inform sustainable planning and management of water resources in the basin. Climate change impact and sensitivity of LULC change on streamflow and sediment yield in the studied basin were assessed using the Soil and Water Assessment Tool (SWAT) model. Sensitivity analysis, calibration, validation and uncertainty analysis of the SWAT model were performed in SUFI-2 algorithm of SWAT-CUP. Calibration and validation were done at monthly time step for river flow and total sediment yield at Bui and Chache, respectively. Due to lack of long term data of good quality, calibration of flow was limited to 6 years (1995-2000) while that of total sediment yield was from 2000 to 2004 (5 years). Validation of flow was for 5 years (2002-2006) and for sediment, 3 years (2005-2007). The calibrated SWAT model was driven with three RCM/GCM pairs (RCA4/MPI-ESM-LR, RACMO22T/ICHEC-EC-EARTH and RCA4/CanESM2) of downscaled rainfall and temperature data analyzed in this study. The model was also driven with the ensemble mean values of the RCM/GCM pairs. The RCM/GCM data were obtained from the CORDEX archives and consisted of data forced by two IPCC emission scenarios- Representative Concentration Pathways (RCPs) 4.5 and 8.5 for two future time slices: 2051-2075 - representing the late 21st century (the 2060s) and 2076-2100 - representing the end of the 21st century (the 2080s). The period 1984- 2010 was set as the baseline. The sensitivity of streamflow and sediment yield to changes in LULC in the BVRB was assessed based on the basin’s LULC maps of 1990 and 2000. The SWAT model results showed that the curve number (CN2) and canopy storage (CANMX) were, respectively, the most and least sensitive model parameters, with respect to streamflow. The model performance in reproducing monthly streamflow and sediment yield of the basin during calibration was rated as “good”. The Nash-Sutcliffe efficiency (NSE), coefficient of determination (R2), percent bias (PBIAS) and root mean square error (RSR) values for streamflow calibration were 0.85, 0.86, 8.1% and 0.38 respectively. Sediment calibration results yielded NSE of 0.68, R2 of 0.74, PBIAS of 27.5% and RSR of 0.59. In general, the model validation results were “satisfactory” for both flow and sediment yield. The NSE, R2, PBIAS and RSR values for streamflow validation were 0.60, 0.62, 20.1% and 0.64 respectively while that for sediment yield were 0.66, 0.74, 39.1% and 0.59 respectively. The p- and r-factor values for the model calibration and validation indicated that low levels of uncertainties existed in the model results. The projected precipitation over the basin by the ensemble members of the models showed a mixed pattern of positive and negative trends. However, both the positive and negative future trends in the rainfall were statistically non-significant. Analyses of the average annual, intra-annual and seasonal precipitation indicated high uncertainty regarding the direction of the future rainfall. Mean annual precipitation change for the late 21st century ranged between -16% and +6% under the RCP4.5 scenario and between -27% and +14% under the RCP8.5 scenario. The end of the 21st century projections showed changes in mean precipitation amounts ranging between -23% and +2% and between -33% and +13% under the RCP4.5 and RCP8.5 scenarios, respectively. With regards to temperature, average annual projections by the ensemble runs showed increases over the basin under both RCP scenarios and for both time periods. Warming over the basin is projected to be higher under the RCP8.5 scenario than under the RCP4.5 scenario, with the end of 21st century period being warmer than the late 21st century. Average annual mean temperature increase across the model run ranged between 2.2oC and 2.6oC under the RCP4.5 scenario and between 3.5oC and 3.7oC under the RCP8.5 scenario for the end of the 21st century. Analysis of dry - (November to March) and wet - (August to October) period streamflow and sediment yield showed mainly increases for the 2060s and 2080s under both scenarios. The model runs also projected increases in mean annual streamflow and sediment yield for the basin. The change in streamflow is projected to range between +40% and +42%, and between +100% and +143% for sediment yield during the late 21st century under the RCP4.5 scenario. For the end of the 21st century, the projected change ranges between -6% and +78% while sediment yield is between +100% and +216%. Under the high emission RCP8.5 scenario, streamflow is projected to range between +48% and up to +148% across the models. For sediment yield the projected change ranges from +249% to +335%. The end of century projections of flow is between +69% and +243% while total sediment ranges between +358% and 412% across models. Analysis of the trends in the future annual flow and sediment yield indicated that the increasing trends observed are statistically significant at the 5% level of significance. Sensitivity analysis of streamflow in the basin based on a 10-year land use/land cover change showed statistically insignificant changes.