Properties of the Threshold Method on a Radar Rain Cluster Basis

Abstract

The average rain rate ?R? is estimated with radar data at the radar beam height by measuring the fractional area F(τ) of pixels above a preset rain-rate threshold τ. This work applies the ?threshold method? to smaller areas. The threshold method typically requires large observing areas (104 km2?105 km2) to achieve good sampling of a well-behaved probability density function (PDF) of rain rate. The present work, however, shows that the threshold method can be applied to smaller observing areas (103 km2) when fractional area is calculated on a rain cluster, or rain cell, defined as a connected set of rainy pixels. The results show general agreement with those reported by other authors. The slope of the ?R??F(τ) regression line relation ?(τ), is approximately linear with the threshold τ. Seasonal values of ?(τ) are obtained by applying the method to seasonally grouped months of radar data from areas surrounding Paris, France, during 1994. Since measuring fractional area on a cell basis allows one to establish an appropriate ?R? versus F(τ) relationship for different cell populations defined in a prescribed manner, a test is conducted using lightning data to isolate lightning-related cells in data from July and August. The conditional mean rain rate above the threshold and ?(τ) are found to be higher in cells with lightning. For the lightning-related cells a tighter ?R??F(τ) relation exists, indicating a similarity in the PDF of rain rate of these cells. A scheme is explored that uses the slope S(τ) as a criterion to distinguish real echoes from ground clutter echoes.