A Physical–Statistical Approach to Match Passive Microwave Retrieval of Rainfall to Mediterranean Climatology

A physical-statistical approach to simulate cloud structures and their upward radiation over the Mediterranean is described. It aims to construct a synthetic database of microwave passive observations matching the climatological conditions of this geographical region. The synthetic database is conceived to train a Bayesian maximum a postetioti probability inversion scheme to retrieve precipitating cloud parameters from spaceborne microwave radiometric data. The initial microphysical a priori information on vertical profiles of cloud parameters is derived from a mesoscale cloud-resolving model. In order to complement information from cloud models and to match simulations to the conditions of the area of interest, a new approach is proposed. Climatological constraints over the Mediterranean are derived on a monthly basis from available radiosounding profiles, rain-gauge network measurements, and colocated METEOSAT infrared measurements. In order to introduce the actual surface background in the radiative-transfer simulations, a further constraint is represented by the monthly average and variance maps of surface emissivity derived from Special Sensor Microwave Imager (SSM/I) clear-air observations. A validation of the forward model is carried out by comparing a large set of brightness temperatures measured by the SSM/I with the synthetic cloud radiative database to asses its representativeness and range of variability. The marginal contribution of each constraint source, used in the matching process, is also evaluated. Finally, surface rain rates, retrieved from SSM/I data using the new synthetic database, are compared with colocated rain rates measured by a rain-gauge network along the Tiber River basin in Italy throughout 1995. This comparison is performed both for selected case studies and in a statistical way, discussing the major advantages and limitations of the proposed approach.