| Summary: | This paper evaluates precipitation forecasts by the operational limited area model ALADIN (grid spacing 10 km) in the region of the European Alps. The observational reference is a daily high-resolution rain-gauge analysis (grid spacing about 25 km) available for the Mesoscale Alpine Programme Special Observation Period (MAP SOP,Sept. 7 to Nov. 15, 1999). The daily analysis is disaggregated in time using data from hourly reporting rain stations. Accordingly, the evaluation is performed for hourly, daily, and SOP mean precipitation fields and at grid scales of 25, 50, and 100 km, since skill of both the ALADIN forecasts as well as the analysis increases with scale. Applying standard continuous statistics for comparisons of means, variances, and patterns the results are: ALADIN and analysis converge with increasing scale of comparison (e.g., at 25 km and hourly resolution ALADIN explains 14% of the spatial field variance and 79% at 100 km at the SOP mean, the global means are not separable), but surprisingly the ALADIN performance of hourly and daily grid box time series are comparable, and forecasts of time series are generally better than of spatial fields (e.g., daily ALADIN forecasts overestimates spatial field variance by almost 50 %, but grid box time variance by only about 10%). The smaller spatial skill of ALADIN is attributed to a spatial bias pattern.
|
|---|