Impact of Uncertainty of Polarimetric Radar-Rainfall Inputs on Distributed Hydrological Model in Urban River Basin

Integrating the high-resolution estimation of rainfall input from X-band polarimetric Doppler weather radar to the distributed hydrological model offers an improved storm event simulation in urban river basin. Uncertainties are inherent in radar-rainfall observation, and they may propagate into final peak flood estimates. This study investigated the influence of systematic and random error in radar-rainfall input on the performance of the distributed hydrological model. The analysis reveals that the model performance is less affected by systematic error of R(Z(KDP)) estimator. The Normalized Peak Error introduced by the feature of R(Z(H)) and R(KDP) is in the range of - 12.3% to 115.7%, and 5.8% to 66.1% respectively. Perturbed radar-rainfall data with random error intensity exceeding 0.3 is detrimental to the model performances. Analysis of sensitivity has shown that a higher sensitivity of hydrological model performance with respect to the overestimation than to the underestimation of rainfall input.