An object-based approach for areal rainfall estimation and validation of atmospheric models

Abstract

Integral radar volume descriptors (IRVD) are introduced and discussed as a pathway to an object-based characterization of precipitation systems. A set of IRVD values characterize the temporal development of precipitation systems which constitute the objects. The IRVDs are based on the temporal evolution of the three-dimensional distribution of radar reflectivities produced by the objects. In a first step a set of descriptors, i.e. potential IRVDs, are postulated, which characterize a precipitating system observable by a scanning radar e.g. the mean echo-top-height or the temporal change of the bright band depth of a raining system. In a second step a statistical analysis identifies those descriptors, which bear the most significant information about system surface precipitation yield, which are called IRVDs the values of which describe the objects. IRVDs are derived both from pseudo-radar observations retrieved from a weather forecast model and from real radar observations. Since different sets of IRVDs suggest also different precipitation generation mechanisms acting in the model and reality, the IRVD concept is proposed as a more process-oriented approach to model validation. Finally, the potential of IRVDs to improve estimates of radar-derived precipitation system yields when used on top of Z–R relations is demonstrated.