Precipitation estimation based on weighted Markov chain model

The prediction of precipitation plays an important role in the climate prediction, but the medium-long term precipitation prediction is still a difficult problem in current climate prediction. To solve the problem of predicting precipitation state, an applicable method was proposed, which was based on weighted Markov chain model. Through the use of smoothing pretreatment, the effects of individual extreme value can be eliminated, and the classification standard of precipitation state can be set up by applying mean-variance method. According to the classification standard, determining historical precipitation state and getting state transfer probability matrix. Then an approach called Markov chain with weights can be used to forecast the future precipitation state by regarding the normalized self-correlatives as weights. On this basis, a specific precipitation was obtained via combining the level Eigen value of fuzzy theory. Finally, with applying this model to a real observation station of hydrology with 47-year precipitation data in Weihai, the experimental results and forecast errors can be accepted.