An evaluation of the WRF Mei-yu rainfall forecasts in Taiwan, 2008–2010: differences in elevation and sub-regions

This study explores differences in the performance of the Mei-yu quantitative precipitation forecast (QPF) in Taiwan by use of the Weather Research and Forecasting (WRF) model in continuation of an earlier study, with particular detailed focus on those among three elevation ranges and four sub-regions at different stages of the season and within the diurnal cycle. Against data from about 400 rain gauges, both subjective and objective verifications are performed to complement each other. The Mei-yu QPF's skill is shown to vary with evaluation periods (every half-month) within the season, with elevation (categorized as plain (0–250 m), slope (0.25–1 km) and mountain (>1 km) areas), and also with sub-region (northern, central, southern and eastern Taiwan). The diurnal cycle of rainfall produced by the model is strong over the mountains, but too weak over the plains. Overall, the model exhibits the highest skill in southern and central Taiwan, where more rain occurs due to forced uplift of the prevailing southwesterly flow over the windward slopes (mainly during the first half of June). In general, the more rainfall there is, the better the model performs (with higher skill) during the Mei-yu season.