Author(s): F. Githui; T. Thayalakumaran

Linked Author(s):

Keywords: Prediction uncertainty; Recharge; SWAT; SEBAL

Abstract: Hydrological models are often calibrated against point data such as measured flow. Due to the large number of model parameters, the information content of point data used in calibration is often insufficient to parameterize such models. Use of remotely-sensed evapotranspiration datasets can overcome this problem as they contain additional information that is spatially distributed. In this paper, we applied the semi-distributed hydrological model (SWAT) of an irrigated catchment in southeastern Australia to compare optimised model parameters, flow and evapotranspiration for the model calibrated using i) flow, and ii) flow and remotely-sensed evapotranspiration. Calibration to flow provided a good fit to simulated flow although the parameters were poorly estimated. In contrast, calibration to both flow and remotely-sensed evapotranspiration gave reasonable parameter estimates and adequate simulations of flow and evapotranspiration. A predictive uncertainty analysis carried out on model outputs showed a reduction in uncertainty in both the model parameters and predictions after the inclusion of remotely sensed evapotranspiration.

DOI:

Year: 2011