Precipitation attribution distance

Abstract

Precipitation is one of the most important meteorological parameters and is notoriously difficult to measure, predict, and also to verify. Distance measures are one of the five classes of spatial verification metrics that try to address the problems of traditionally used non-spatial methods (which only compare values at collocated grid points). Distance measures provide an estimate of spatial distance between the events in the two fields, which is relatively easy to interpret and understand. Our goal was to develop a new distance measure for precipitation that would not require thresholding and provide meaningful results that would not be too inconsistent with subjective forecast evaluation. At the same time, it would be computationally fast and numerically stable and could be used with fields provided on irregular grids. The new metric is called the Precipitation Attribution Distance or PAD and is based on a random nearest-neighbor attribution concept - it works by sequentially attributing randomly selected precipitation in one field to the closest precipitation in the other and is nondeterministic. We analyzed the new metric’s behavior in many idealized and real-world situations and compared it with the behavior of existing distance metrics.