Simulation of SPI Time Series Stochastic Behavior by Using Markov Chain Model for Early Warning of Fars Province Drought

The standard precipitation index (SPI) as a selected index was computed based on the same period in 32 years (1343-78) in the time scale of 12 months for 26 stations located inside and outside of Fars Province. In the next stage, drought severity data resulting from SPI method was combined with Markov chain, and a probability transition matrix and steady matrix of region were developed. Ultimately, region long term condition in terms of drought duration and severity was predicted for coming 10 years. The result showed that in probability transition matrix of drought, highest probability was observed in diagonal part of matrix indicating the stability of environmental condition. In probability steady matrix of drought, the percentage of the periods persisting fr long term is similar to the probabily shown by SPI, indicating Markov chine model capability of predicting draoughts and disemminating warnings in advance in Fars Province.