On the use of real‐time radar rainfall estimates for flood prediction in mountainous basins

Abstract

Abstract. This paper investigates the effect of systematic mean-field and rangedependent radar rainfall errors on the accuracy of runoff simulation in mountainous basins. Statistical analysis of radar rainfall and runoff simulation error is performed on six flood events for two medium size watersheds in northern Italy, located at 38 and 60 km basin-to-radar distances, respectively. We show significant range-related rainfall biases, which are due to the high elevation radar scans used to minimize the interception of the radar beam with the topography. These biases are corrected by converting radar reflectivity measurements at a given altitude into their equivalent surface values, using real-time identification of the mean vertical reflectivity profile. The mean-field bias is adjusted using a multiplicative factor determined based on real-time radar-rain gauge comparisons. The impact of the above radar rainfall biases, and the improvements obtained from the proposed corrections, on areal-rainfall estimation and runoff simulation are evaluated. It is shown that radar rainfall biases magnify through the rainfall-runoff transformation and preclude the accurate simulation of runoff, particularly for the distant basin. The combined correction procedure results in significant reduction of the hydrologic prediction error, especially at the distant basin.

Advancing Precipitation Estimation and Streamflow Simulations in Complex Terrain with X-Band Dual-Polarization Radar Observations

“…Despite the random differences shown in Figures 3b and 4b between the XPOL and in situ rainfall for the high elevation angle (3.5 • ), the comparison of accumulated rainfall depicted in Figures 3c and 4c, shows that the VPR correction significantly reduces the difference between radar and in situ ground observations. Removing this systematic error in accumulated rainfall is a critical factor in hydrological runoff modeling [68], which will be discussed later in the paper.…”

A Bayesian technique for conditioning radar precipitation estimates to rain-gauge measurements

“…While their conclusions on the effect of radar and raingauge combination are generally positive in terms of bias reduction (Borga et al, 2000), they seem rather negative in terms of reduction of variance. In the case of uncertain covariance, Bayesian techniques have been used in Kriging for parameter estimation and the assessment of the uncertainty induced on the Kriging spatial functions (Kitanidis, 1986;Omre and Halvorsen, 1989;Le and Zidek, 1992;Handcock and Stein, 1993).…”

Investigating the spatial scaling effect of the non-linear hydrological response to precipitation forcing in a physically based land surface model

“…The impact of precipitation retrieval error on the simulation of hydrological variables (hereafter called error propagation) has been the subject of a number of studies (Sharif et al, 2002;Borga et al, 2000). These studies by Borga et al and Sharif et al focused on error propagation in runoff prediction driven by radar rainfall observations.…”