Analysing the Performance of Various Radar-Rain Gauge Merging Methods for Modelling the Hydrologic Response of Upper Thames River Basin, Canada

Accurate estimate of precipitation is of paramount importance for assessing the hydrologic response of a river basin. Weather radar data integrated with rain gauge measurements are applied to characterize the spatial feature of the storm event producing precipitation over the basin. Ordinary kriging of rain gauge data, mean field bias, Brandes spatial adjustment, conditional merging (CM), and local bias techniques are applied in this study to evaluate the performance of these radar-rain gauge merging methods for hydrologic modelling of the Upper Thames River basin (UTRb), southwestern Ontario, Canada. Singularity-sensitive Bayesian merging method (SSBM) with a fine spatial resolution was also applied to retain the singularity character of the rainfall event. Rainfall-runoff simulations were carried out for three major storm events recorded in the UTRb using the HEC-HMS 4.0 hydrologic model. River flow analysis was performed for the comparison of results of HEC-RAS 4.1 hydraulic model with the observed rating curve. A novel methodology involving a dual-storage system is proposed to model three sub-basins of UTRb which displayed skewed and spiked observed runoff hydrographs. Using this dual-storage system for the three sub-basins it is found that CM and SSBM merging methods yielded optimal Nash-Sutcliffe efficiency coefficients for the prediction of runoff from these sub-basins.