Abstract:

Using 24-hour precipitation data from 39 observation stations on the Xizang Plateau from 2017 to 2021, the SWC-WARMS model products of precipitation forecasts were evaluated according to the operational numerical prediction evaluation metrics. The results showed that precipitation on the Xizang Plateau is mainly concentrated in the warm season (rainfall period), with light rain as the dominant form. The annual averaged accuracy rate, hit rate, and success rate of the 24-hour cumulative precipitation forecast from the SWC-WARMS model are 78%, 83%, and 60%, respectively, indicating considerable prediction performance, which are however characterized by relatively high false alarm rate and bias. The averaged TS score for light rain and above during the warm season is up to 0.63, while the TS for moderate rain and above is relatively low. The false alarm rate and missing alarm rate increase with rising altitude differences. During the cold season (snowfall period), the number of precipitation days is fewer, with lower TS score and higher false alarm rate than those in the warm season. However, in certain regions, the TS for large-threshold snowfall forecasts reaches 0.30 in the SWC-WARMS model.

Key words: TS score, False alarm rate, Missing alarm rate