Polarimetric weather radar : from signal processing to microphysical retrievals

Accurate modelling of liquid, solid and mixed-phase precipitation requires a thorough understanding of phenomena occurring at various spatial and temporal scales. At the smallest scales, precipitation microphysics defines all the processes occurring at the level where precipitation is a discrete process. The knowledge of these microphysical processes originates from the interpretation of snowfall and rainfall measurements collected with various sensors. Direct sampling, performed with in-situ instruments, provides data of superior quality. However, the development of remote sensing (and dual-polarization radar in particular) offers a noteworthy alternative: large domains can in fact be sampled in real time and with a single instrument. The drawback is obviously the fact that radars measure precipitation indirectly. Only through appropriate interpretation radar data can be translated into physical mechanisms of precipitation. This thesis contributes to the effort to decode polarimetric radar measurements into microphysical processes or microphysical quantities that characterize precipitation. The first part of the work is devoted to radar data processing. In particular, it focuses on how to obtain high resolution estimates of the specific differential phase shift, a very important polarimetric variable with significant meteorological importance. Then, hydrometeor classification, i.e. the first qualitative microphysical aspect that may come to mind, is tackled and two hydrometeor classification methods are proposed. One is designed for polarimetric radars and one for an in-situ instrument: the two-dimensional video disdrometer. These methods illustrate the potential that supervised and unsupervised techniques can have for the interpretation of meteorological measurements. The combination of in-situ measurements and polarimetric data (including hydrometeor classification) is exploited in the last part of the thesis, devoted to the microphysics of snowfall and in particular of rimed precipitation. Riming is shown to be an important factor leading to significant accumulation of snowfall in the alpine environment. Additionally, the vertical structure of rimed precipitation is examined and interpreted.