Error Structure of Multiparameter Radar and Surface Measurements of Rainfall Part I: Differential Reflectivity

Abstract

Fluctuations in the radar measurements of ZDR are due to both signal power fluctuations and the cross-correlation between the horizontal and vertical polarized signals. In Part I of this study, these signals are simulated for an S-band radar for backscatter from rain media, which is characterized by a gamma model of the raindrop size distribution (RSD). The parameters N0, D0, m of the gamma RSD are then varied over the entire range found in natural rainfall. Thus, the radar simulations contain the effects of both statistical fluctuations and physical variations. We also simulate sampling of raindrops by disdrometer. The sampling errors are related to the Poisson statistics of the total number of drops in the fixed sample volume and to the statistics that govern the gamma distribution of drops as a function of size. We simulate disdrometer RSD samples over the entire range of N0, D0, m values found in rainfall, so that the effects of statistical fluctuations and physical variations are introduced. It is shown that ZDR, computed from disdrometer RSD samples, is correlated with Z and with other moments of the RSD when the same disdrometer data is used. This correlation is purely statistical and is independent of the physical correlation. We use the radar and disdrometer simulations to intercompare the rain rate as derived by the radar ZDR-method with the rain rate estimated by the disdrometer. Our simulation results are used to explain the correlation and error structure of radar/disdrometer-derived rain rate intercomparison data reported in the literature.