Radar Rainfall estimation using different Polarmetric Algorithms

The polarimetric algorithm for rainfall estimation based on the power relation between rain rate R and specific differential phase K_DP R=A K_DP^b tends to underestimate / overestimate rainfall if small / large drops dominate the drop size distribution (DSD), although to a lesser degree than the conventional algorithm utilizing radar reflectivity factor Z.

In order to mitigate effects of DSD variability, we suggest the extension of the algorithm R(K_DP) with the coefficient A depending on differential reflectivity ZDR averaged over sufficiently large spatial / temporal domain. Different functions A(Z_DR) are recommended for light, moderate, and heavy rain. These functions have been derived empirically by comparing one-hour rain accumulations obtained from the Cimarron polarimetric radar and 42 gauges comprising a dense micronetwork in Central Oklahoma. Twenty rain events have been examined.

The use of average values of Z_DR instead of point Z_DR estimates greatly reduces the influence of statistical errors in Z_DR measurements on the overall performance of the suggested algorithm. The result is a much less biased estimate of rainfall compared to the algorithm R(K_DP) without noticeable increase in statistical errors.

We have compared also polarimetric variables measured by the radar with the ones computed from 2D-video-disdrometer assuming the Pruppacher - Pitter shapes of raindrops. It was found that specific differential phase measured by the radar is somewhat smaller on average than K_DP estimated from the disdrometer. The reasons for such a discrepancy are analyzed.