EVALUATION OF PROBABILISTIC FORECASTS OF CONSECUTIVE DAYS WITHOUT MEASURABLE RAINFALL OVER TAIWAN

Demand for probabilistic forecasts of consecutive days without measurable rainfall has grown significantly by users in different sectors of society, especially in agriculture, livestock, and water resource management. The purpose of this study is to provide users with reliable and skillful forecasts, which help users obtain more economic benefits in decision making. In this study, Analog post-processing (AP) is applied in 20-year reforecasts of the National Centers for Environmental Prediction (NCEP) Global Ensemble Forecast System version 12 (GEFS v12) to produce calibrated and downscaled probabilistic forecasts of consecutive days without measurable rainfall over Taiwan land area. Long-term forecast evaluation indicates that: (1) the problem of under-dispersion of the raw forecasts is effectively mitigated through the AP. (2) The probabilistic forecasts of consecutive days without measurable rainfall have good reliability and discrimination (potential usefulness) within next four weeks. (3) The calibrated forecasts provide higher economic benefits for users with a much wider spectrum of cost-to-loss ratio compared to the raw forecasts.