An assessment of satellite derived total evaporation data as a data source to the ACRU hydrological model.

Abstract

Hydrological models and tools are often used as decision support systems to inform water resources management. The successful application of these systems is largely dependent on the quality of data being incorporated into them. Accurate information with regards to total evaporation is of paramount importance to water resources managers, as it is a key indicator in determining if water resources are being used for their specific purposes. Due to the inherent spatial limitations associated with conventional techniques to estimate total evaporation, the application of satellite earth observation as a tool to estimate total evaporation is being advocated more frequently. The focus of this Dissertation was to develop an approach which would allow for the incorporation of total evaporation estimates from an existing evaporation model that incorporates satellite earth observation data i.e. the SEBS model, into a hydrological simulation model i.e. ACRU, to simulate streamflow. The SEBS model was first validated in the Komatipoort study site against the surface renewal system. The results of this investigation indicated that the SEBS model over-estimated total evaporation by approximately 47% and produced R2 and RMSE values of 0.33 and 2.19, respectively, when compared to total evaporation estimates obtained from the surface renewal system. Once, the model had been validated, it was then applied to estimate total evaporation for quarternary catchment X23_A for the period 01st December 2011 to 25th November 2012. These estimates were used to create a continuous total evaporation time series, which was used as an input to ACRU to model streamflow. The EVTR3 approach was derived to allow for the incorporation of the aforementioned SEBS total evaporation estimates in ACRU and to estimate streamflow amongst other hydrological parameters. The simulated streamflow for this technique was under-estimated by approximately 10% and produced R2 and RMSE values of 0.41 and 1.05, respectively, when compared to observed streamflow. Although these results appear to be satisfactory at best, similar results were obtained when using the conventional evaporation routine in ACRU to estimate streamflow. This occurrence circuitously highlights the potential of utilizing satellite earth observation data as a data source for a hydrological model.