Comparing data from radar, rain gauges and a NWP model

Data from rain gages, radar and a high resolu- tion Numerical Weather Prediction (NWP) model are being analyzed. The accuracy of daily, operational, radar-derived precipitation amounts is verified during heavy rain in the Alps, using in situ measurements (71 gages within 157 km from the radar). Independent data were used for the train- ing and the verification of two adjustment techniques: even a simple bulk-adjustment (based on one correction coeffi- cient, hence, space-independent) leads to a significant im- provement; a Weighted Multiple Regression (WMR) is well worth the additional effort of deriving four coefficients to replicate the influences of calibration, beam-broadening, visibility and orography. Good agreement is found between daily radar/gauges amounts during severe precipitation. This confirms previous findings: in days with strong weather signal, the WMR is able to correct several errors in one step (calibration, beam-broadening, shielding and orographic enhancement). Since it does not require full volume 3D-data, it represents an inexpensive alternative to more sophisticated methods (e.g. reflectivity profile correction). The average forecast (over a 12 000 km 2 area) was in reasonable agree- ment with the average observations while the intense cells were shifted in space (they were forecast upstream, while they occurred upslope).