Abstract
The central objective of this analysis is to significantly enhance the quality of radar-derived precipitation estimates by as fully as possible exploiting the information contained in the spatial and temporal variability of 3D radar volume data. The results presented are based on pseudoradar data and rain rates of a regional weather forecasting model and 12 true radiosoundings as well. Two approaches are pursued: the first approach estimates total rainfall from an individual storm over its lifetime, whereas the second approach assesses the areawide instantaneous rainfall from a multiplicity of such storms by the use of measurements of the areal coverage of the storms exceeding a threshold radar reflectivity. The concept is extended by adding more predictors to significantly enhance the rainfall estimates. The horizontal expected value and the horizontal standard deviation of enclosed reflectivities at the ground, the mean brightband fraction and its trend, the fractional area with reflectivities exceeding a threshold τ, and an orographic rainfall amplifier provide relative errors smaller than 10% in approximately 75% of the considered rain events in the first approach. In the second approach, a relative error is achieved that is below 10% in approximately 63% elements of the test set.
Corresponding author address: Silke Trömel, Meteorological Institute, University of Bonn, Auf dem Hügel 20, D-53121 Bonn, Germany. Email: silke.troemel@uni-bonn.de
