Abstract: Index selection was basic for forecast and prediction,mitigation and prevention to agro-meteorology disasters.A case study on crop drought-flood index(K) and precipitation anomaly percentage(Pa) was conducted.A series of time sequences on K and Pa about three growth stage of winter wheat(sowing to mature,regreen to jointing,jointing to mature) during 1981-2000,1981-2001,1981-2002,was established in Yuncheng,Shanxi province.Drought and flood disasters during 2001-2003 were predicted by using Markov model and probability distribution function.The results showed that prediction based on K was consistent with the real situation from 2001 to 2003,as the relative veracity was only 55.5% based on Pa.It was indicated that the K as crop drought and flood index was more effective than Pa.The method of selection agro-meteorology disaster index based on risk prediction was simple and convenient,which could recommend to a new method for selecting agro-meteorology index.

Key words: Index of drought and flood, Index of drought and flood, Markov random process, Transfer probability, Risk prediction