The microphysical structure of extreme precipitation

Improved capability for measuring and predicting extreme precipitation would provide significant economic and societal benefits. Although an appreciable effort has been devoted to modeling and observing the storm-scale structure of extreme precipitation phenomena, their microphysical structure has received relatively little attention. Yet, it is the spatio-temporal variability of the hydrometeor size spectra in extreme precipitation which can learn us more about the physical processes involved in producing it. Moreover, an improved understanding of the microphysical structure of extreme precipitation is crucial in developing improved techniques for its (remote) measurement. Since accurate measurement is the basis for reliable prediction, this seems a particularly relevant issue.

We present an extensive analysis of existing datasets of raindrop size distributions corresponding to rain rates in excess of 100 mm h^-1. The datasets have been collected using different types of instruments, i.c. the Illinois State Water Survey raindrop camera, the Joss-Waldvogel disdrometer and the Knollenberg optical array probe. They involve both measurements at the ground and at different altitudes aloft. These data have been analyzed in the framework of a relatively novel formalism, in which raindrop size distributions are parameterized in terms of a scaling law. This formulation allows a separation of the effects of the spatio-temporal variability of bulk rainfall variables (rain rate, reflectivity, etc.) from changes in the shape of an intrinsic raindrop size distribution. This so-called general raindrop size distribution, in contrast to the conventional definition, does no longer depend on the actual values of bulk rainfall variables. It is therefore directly related to the physical processes producing (extreme) precipitation. The effects of the spatio-temporal variability of bulk rainfall variables are entirely contained in the values of the so-called scaling exponents. The values of these exponents determine whether it is the variability of the raindrop sizes or the variability of their concentrations (or some combination thereof) which controls the variability of the raindrop size distribution.

A preliminary analysis of the discussed dataset reveals that extreme rain rates tend to be associated with atmospheric conditions in which the variability of the raindrop size distribution is strongly number-controlled. This implies that bulk rainfall variables would be roughly proportional to each other, something which would have profound implications for the remote measurement of extreme rainfall. Moreover, the associated intrinsic raindrop size distributions display a tendency towards multi-modality. Both observations happen to be consistent with the concept of the so-called equilibrium raindrop size distribution.