Improving Radar-Based Estimation of Rainfall over Complex Terrain

Abstract

This paper investigates a multicomponent radar-based rainfall estimation algorithm that includes optimum parameter estimation and error correction schemes associated with radar operation over mountainous terrain. Algorithm preprocessing steps include correction for terrain blocking, adjustment for rain attenuation, and interpolation of reflectivity data from polar radar coordinates to a three-level (1, 2, and 3 km) vertically integrated Cartesian grid. The error correction schemes investigated herein include a simple but efficient approach to correct for the vertical variation of reflectivity and a stochastic filtering approach for mean-field radar-rainfall bias adjustment. The primary algorithm parameters are estimated through a global optimization scheme. Eight major flood-inducing storm events observed coincidentally by a C-band weather radar and 39 rain gauge stations over an alpine region of northeast Italy are used. We describe sensitivity analysis of the parameter values obtained from global optimization, the improvements in accuracy owing to the implementation of the different preprocessing and error correction schemes, and the overall improvement achieved as compared with previous algorithm studies in the area. The advantages of performing vertical integration versus using the lowest-available-elevation radar field, applying stochastic filtering versus the deterministic approach for mean field bias, and estimating the algorithm parameters through optimization are demonstrated.