A remote sensing based approach to determine evapotranspiration in the Mbire District of Zimbabwe.

Abstract

Water availability and consumption assessments are important for integrated water resources management. Such assessments are usually hindered by the lack of hydro-meteorological data to close the water balance. Remote sensing based methods and hydrological modelling techniques provide tools to obtain estimates of spatial and temporal variation of key water balance components such as evaporation, transpiration and soil moisture which account for about 65 % of the global water budget. In this study, twenty atmospherically corrected MODIS images taken between 2005 and 2015 in the Mbire District in Zimbabwe were processed using the Surface Energy Balance System (SEBS) algorithm to determine the actual evapotranspiration (ETA). The study revealed an increasing trend of actual evapotranspiration from 2005 to 2015, with values of ETA of around7 mm/day being observed in the densely forested regions as compared to 3 mm/day for grassland. Results also showed that the SEBS estimated mean ETA values (3.79 mm/day) were within reasonable range of the mean Potential ET (3.31 mm/day) for the 3 automatic stations which were setup in the district during the period of study. The Topographic driven rainfall runoff model (TOPMODEL) whose land surface inputs were obtained from remote sensing techniques was calibrated with runoff data and was extended to simulate soil moisture patterns as well as ETA from September 2008 to August 2010. The model simulation yielded Nash Sutcliffe (NSE) values of 0.4 and 0.1 for Manyame and Angwa respectively during the calibration period. Results for soil moisture estimation by TOPMODEL showed high levels of soil moisture (mean value of 80 %) along river channels, valleys and floodplains. The point to pixel comparison of soil moisture simulated by the model and the one retrieved by the data logger at Chidodo station showed a difference of 12 %. A water balance analysis for the basin was conducted by subtracting all outflows from the basin inflows, the inputs driven by SEBS and/or TOPMODEL. Results of the precipitation and evapotranspiration estimates reveal that, the Mbire District received an average of 4.37 mm/day and 4.66 mm/day respectively from November to the end of March 2016. Refinement of the analysis was achieved by dividing the Mbire District into 3 hydrological sub-catchments and this saw an improved SEBS derived water balance closure by 7.2 %, 16.5 % and 12.8 % as compared to 11 %, 20 % and 16.1 % by TOPMODEL for the Angwa, Manyame and Musengezi basins respectively. Ultimately a combination of GIS, hydrological modelling and remote sensing techniques have proven to be an affordable and efficient means of determining water balance parameters, closure of water balance, determine water availability and hence facilitate easier and more efficient analyses in areas where point data may be scarce or unavailable.