ANALYSIS OF CLIMATE AND LAND USE/LAND COVER CHANGES AND THEIR IMPACTS ON RIVER DISCHARGE AND SOIL LOSS IN THE DIDESSA RIVER BASIN, SOUTH WEST BLUE NILE, ETHIOPIA

Abstract

Didessa River basin is the second largest River Basin in the Blue Nile Basin, which contributes about a quarter of the Blue Nile annual total flow. Number of residents in the River Basin is increasing at alarming rate. This in turn has increased pressure on land and atmospheric resources in the River Basin. However, sufficient scientific information on land use/land cover change (LULCC) and climate change and corresponding impacts on Didessa River discharge and soil loss is lacking. Lack of such information could impede undertakings of possible corrective measures when there are misuses and to maintain when positive. Therefore, this study was undertaken with four main objectives: 1) to analyze variability and trend of historical rainfall and temperature 2) to predict future climate variability and change 3) to detect LULCC and 4) to investigate effects of LULCC and climate change on river discharge and soil loss. Thirty one year (1984 to 2014) historical rainfall and temperature data were analyzed. Coefficient of Variation (CV), Standard Rainfall Anomaly (SRA) and Precipitation Concentration Index (PCI) were derived from the daily rainfall data. Rain season onset, cessation and length were analyzed through INSTAT+v3.37 software. Trend and magnitude of change in rainfall and temperature were tested through Mann–Kendall test and the Sen's slope estimator, respectively. Variability and trend of future climate under RCP scenarios were also described by the same methods as historical climate. Change between historical and future climate characteristics were considered as the mean difference between their values. Landsat 5, 7 and 8 images acquired in the years 1986, 2001 and 2015, respectively, were used to detect LULCC through Maximum Likelihood Algorithm of supervised classification scheme in ERDAS IMAGINE2014 software. Finally, river discharge and suspended sediment yield were simulated by Soil and Water Assessment Tool (SWAT2012) model. Results of historical climate analysis showed annual and seasonal rainfall coefficient of variation between 11% and 14%, and 28% and 94%, respectively. More than 70% of the years (1984 to 2014) had SRA in the category near normal to extremely wet condition. Annual rainfall distribution was found to be in moderate condition for Arjo, Bedele, Nekemte, and Sire stations and with high concentration for Dembi and Gimbi stations. Length of rain season was found to be between 208 and 185 days for all the sites. Annual and seasonal rainfall trend was increasing except at Dembi and Sire sites in winter and autumn season where decreasing trends were found. Temperature had shown increasing trend both annually and seasonally at all the stations. Results indicated possible decrease in average future winter season rainfall amount at Nekemte, Dembi and sire in the range of 2.6 to 127.6 mm per annum while that of Bedele, Gimbi and Arjo may increase by 7.5 to 67.6mm per annum under all the RCP scenarios. On the other hand, autumn season rainfall may increase at Nekemte, Gimbi and Arjo in the range of 27.9 to 33.8mm while that of Dembi, Bedele, and Sire may decrease in the range of 0.2 to 38.6 mm per annum under all the RCP scenarios. The study result also indicated late onset of rain season by 3 to 37, 1 to 37 and 2 to 36 days in the early (2018 to 2043), mid (2044 to 2069) and late (2070 to 2095) 21st century, respectively. Early cessation of rain season by 3 to 49, 2 to 45 and 10 to 45 days may occur under RCP2.6, 4.5 and 8.5. Length of rain season may decrease by 11 to 63, 2 to 80 and 10 to 80 days in the early, mid and late 21st century, respectively. In the early and mid 21st century maximum temperature may increase by to 2.2 oC under RCP2.6 and RCP8.5, respectively, while it is expected to rise by 3.58 oC in the late 21st century under RCP8.5. Cropland, builtups and water body coverage has increased by 30.7%, 343.5% and 4.2%, respectively, while forest, shrubland and grassland were decreased by 18.8%, 31.1% and 53.6%, respectively, between 1986 and 2015. Overall accuracy of the classified 1986, 2001, and 2015 LULC images were 92.36%, 93.06% and 96.53%, respectively. Simulated average monthly river flow at the final outlet point increased by 10.6 m3/s due to LULCC between 1986 and 2015 under historical climate. Predicted future climate under RCP8.5 may increase the monthly river flow by 59.1 m3/s when 2015 LULC is kept constant. LULCC between 1986 and 2015 increased average monthly sediment production by 3,096 ton under historical climate. Predicted future climate under RCP8.5 may increase the average monthly sediment production by 74,834.6 tons when 2015 LULC is kept constant. It is likely that 87.04% of the total area may experience high soil loss under RCP8.5. Therefore, it is important that land use in the River Basin be revisited to reduce inappropriate changes and their associated impacts on climate, river flow and soil loss.