Improving operational measurement of precipitation using radar in mountainous terrain-Part II: verification and applications

This is the second letter of the series on improving operational measurement of precipitation using radar in mountainous terrain. Four major rainfall events (11 days) that affected the Western Alps between autumn 1994 and autumn 2000 are analyzed. In the first letter, the focus was on the advantages/disadvantages of the adjustment techniques, the importance of the training, and quality check of the radar-gauge couples used for the training itself. In this letter, our objective is to verify the improvement that can be eventually be obtained by applying in real time the adjustment techniques previously described. The first day of each event was used in the training, and the remaining seven days provided the dataset for the verification (up to 70 gauges in a ∼12000-km² area within 157 km from the radar). A simple bulk adjustment is able to reduce the root-mean-square differences between the daily radar and gauge amounts. A further improvement can be obtained by using, instead of a single bulk-adjusted coefficient, the four correction coefficients derived through a weighted multiple regression (WMR). Thanks to the four WMR-derived coefficients, it is possible to correct several errors in one step (not only "bias," but also beam-broadening, shielding, and orographic enhancement). As a consequence, it represents an inexpensive alternative to more sophisticated methods (e.g., vertical reflectivity profile correction), and at least during intense "long-lasting" (one day or more) events, it could be used operationally to improve-in real time-precipitation estimates over mountainous terrain.