| description abstract | AbstractWith the advance of the weather radar technology, dual-polarization (dual-pol) radar data are now available for hydrological studies, which go beyond the traditional rainfall products relying purely on rain gauge data. Previous studies have focused on the evaluation of rainfall products and their hydrological responses using point-based observational data; however, spatial patterns of simulated hydrological variables are equally important to be considered in order to fully address the distributed effect of the precipitation estimates. In the present study, we compare three rainfall estimations based on rain gauge, single-polarization, and dual-pol radar data. Special attention is given to the use of the two radar products and their corresponding hydrological simulations of both surface water and groundwater. Performance of the hydrological simulations is evaluated based first on traditional point-based observations of stream discharge and groundwater head, and second on remotely sensed land surface temperature data. For the latter, the empirical orthogonal function analysis, which quantifies spatial pattern similarities, is employed. The Skjern River catchment in western Denmark is selected as the study site, and the results show that all three models perform equally well in terms of the traditional aggregated evaluation criteria, such as Nash?Sutcliffe efficiency (NSE) and RMSE on time series data. It is found that the differences of simulated hydrological spatial patterns are sensitive to rainfall signal intensity, as well as the simulation scale in space (<100 km2) and time (subdaily). Our study suggests that the currently available observational data have limited capabilities to clearly differentiate the performance of the three applied models due to the low resolution. | |
