Abstract: A fuzzy comprehensive evaluation method and TS (Threat Score) score were used to compare and analyse the forecast skills of OCF (Objective Consensus Forecasting) model based on areal rainfall over six reservoir basins in Zhejiang province during Meiyu period,persistent rainy weather in August,afternoon thundershowers,typhoon precipitation processes in 2014.The results show that during the frontal precipitation processes (i.e.,Meiyu rainfall and August continuous rain processes),OCF model has better forecast skills with higher fuzzy comprehensive and TS scores,and the false alarm ratios are far higher than the miss alarm rates.Prediction skill of OCF model is gradually improved with the lead time from 192 h to 24 h,and the differences predicted among reservoirs are reduced.For the afternoon thundershower cases,OCF model gets a secondary performance in fuzzy comprehensive and TS scores.When these reservoir basins gets higher false alarm rates,they has higher miss alarm rates.The forecast scores of leading time from 192 h to 24 h show an unstable change pattern.There are significant regional differences.For typhoons of Fungwong and Matmo,OCF model exhibits a poor performance both in fuzzy comprehensive and TS scores,and the false alarm rates are higher than those of the miss alarm.The forecast skill shows a distinct improving trend from 192 h to 24 h in the leading time.The differences among reservoirs are obvious.

Key words: OCF model, Area rainfall, Fuzzy comprehensive assessment, TS score