Characterization of Radar Rainfall Estimation Error due to Variations of Raindrop Size Distribution and Vertical Storm Structure

This study attempts to determine the relative significance of two key error sources of the ground-based volume-scanning radar rainfall estimation, namely: (1) the effect of vertical storm structure (i.e., vertical profile of radar reflectivity; VPR) combined with the radar beam height, and (2) the variability of the raindrop size distribution (DSD). Three different statistical models are combined that characterize (1) the rainfall distribution in space and time, (2) the variability of DSD, and (3) the VPR effect. The rainfall fields are generated using a space-time stochastic model. The DSD variability is modeled using a statistical parameterization scheme, which assures quantitative consistency between rainfall rate and the DSD, and simultaneously allows for plausible variability in the DSD parameters. The VPR effect is represented using a bimodal statistical scheme. The model parameters will be estimated based on data collected by the Tropical Rainfall Measuring Mission's (TRMM) space-borne radar during overpasses over the Goodwin Creek research watershed in northern Mississippi. The TRMM radar data will be used to estimate the parameters of the VPR model, while the raindrop spectra data collected simultaneously by the Joss-Waldvogel disdrometer within the Goodwin Creek catchment will be used to estimate the parameters of the DSD model. Using the TRMM radar based precipitation type classification (e.g., convective, stratiform, mixed rain), model parameters will be estimated as a function of storm type to "physically" link the various statistical models. Multiple realizations of the simulation framework will be used to evaluate the relative error statistics of the two investigated error sources for varying radar range and storm characteristics.