Inversion techniques for ground-based microwave radiometric retrieval of precipitation columnar contents and path attenuation

Nonlinear inversion algorithms are developed to invert ground-based radiometric measurements for different sets of frequency channels and precipitation regimes. Both statistical regression estimators and feedforward neural networks are applied and compared using synthetic data sets from 6 to 50 GHz. An experimental validation is carried out using data collected by the ITALSAT ground-station (near Rome, Italy) equipped with 3 beacons at 19.7, 39.6, and 49.5 GHz together with a multi-channel radiometer at 13.0, 23.8, and 31.6 GHz. Results in terms of comparison between measurements and predictions for a rain event are finally discussed.