The paper addresses the problem of the reconstruction of the rainfall field using weather radar observables. It is well known that at the C band and especially at the X band the reconstruction of the rainfall rate profile along the range using absolute (ZH) and differential (ZDR) reflectivity measurements is significantly affected by the attenuation coefficients (i.e. αH and αD. This problem has been long and extensively studied and iterative attenuation correction techniques based on a cumulative procedure were developed, in which the attenuation at nth cell is estimated using the attenuation corrected reflectivity values at previous cell. Usually the attenuation coefficients αH, αD are estimated using non linear parametrizations with (ZH, ZDR), or, if phase measurements are available, using linear parametrizations with the specific differential phase shift KDp. In this work novel approaches based on neural networks (N.N.) have been used.First, to estimate αH, αD from (αH, ZDR, and ZDP; the N.N. estimators have shown better performance (often, slightly better) in comparison to the best ones known.Second, N.N. have been implemented to extract the range rainfall rate profile. The input to the network is a vector containing the attenuated measurements of (αH, ZDR, and ZDP in a number of range cells while the output is the estimated profile of the rainfall rate. In this way a global compensation of the attenuation is implemented.
Read full abstract