Spring Drought Analysis And Forecasting In Typical Area Of Shandong Province Based On Monte Carlo

Drought disaster is one of the most nonnegligible natural disasters in Shandong Province, especially more prominent in recent years and causing heavy losses to the national economy. One of the drought characteristics of Shandong Province is severe drought,because of the distribution of precipitation is very uneven in the year, spring only accounted for 14% of the annual precipitation. Spring(March-May) is a period that main grain crops wheat is reviving,growing and maturing in Shandong Province, water demand is big,lack of precipitation. The spring drought is obviously divergent in different areas in Shandong Province, among them, Binzhou, Dezhou, Dongying locate in northwest and Weifang, Jinan this five places because of small precipitation and big evaporation become cities that drought occurred more frequently and caused harm more seriously. In this background, this paper studies spring drought conditions in this several typical areas in Shandong Province, and to predict the spring drought in the future.This paper mainly studies the spring drought in five typical areas of Shandong Province. First, Pa, Z and SPI three drought indexs were selected as the index of drought to analyze the monthly precipitation and total precipitation in spring of five areas in 1955-2014.According to the different indexes divide the different level of drought,we can obtain monthly and annual times of spring drought and compare with the actual drought situation, then the adaptability of the indexes in the typical area of Shandong Province was summarized. Secondly, using Monte Carlo simulation to study the spring precipitation. Monte Carlo simulation can be accomplish by Risk Simulator(?) 2013. It contains distribution fitting tool and analysis tools that can simulate five typical areas precipitation distribution pattern and calculate the probability density function(PDF), the cumulative probability density function(CDF) and the inverse cumulative probability function (ICDF) of five typical areas. By PDF,CDF and ICDF,we obtain occurrence probability of spring drought of different levels, and compared with the occurrence probability of spring drought through index, illustrate the advantages of Monte Carlo for the distribution of precipitation expression; then we use Monte Carlo simulation to nonlinearly predict spring precipitation in five typical areas. Finally, five typical areas of Shandong Province of spring precipitation time series data are established ARIMA model, and predict precipitation in the future by the model then infer when appear to spring drought in the next five years. Comparatively analyzing the forecasting spring drought of Monte Carlo and ARIMA,to optimize drought forecasting results.The research finds that:analyzing Pa, Z index and SPI three kinds of drought indices in a single month scale we can discover that Z index has the best adaptability to divided into drought grade of five typical areas; on the seasonal scale, Pa although has less performance to express severe drought ability, but it is more accurate, and better than the other two index to express the light drought. Rainfall distribution pattern in five typical areas based on Monte Carlo simulation are obtained, respectively as Weifang Erlang distribution, lognormal Jinan, Binzhou maximum value distribution, the normal distribution Texas,Dongying maximum value distribution. The probability of spring drought occurrence around respectively are 32.87%,37.41%,37.33%,31.44% and 38.10%. Compared with the typical areas spring drought probability described in drought index, severe drought probability described in Monte Carlo simulation get closer to the reality, the accuracy is higher than drought index in the distribution fitting, so Monte Carlo simulation better described precipitation distribution characteristics.The five typical areas of ARIMA model respectively are Weifang ARIMA (5,1,3) model, Jinan ARIMA (2,1,7) model, Binzhou ARIMA (2,1,3) model, Dezhou ARIMA (5,1,8) model, Dongying ARIMA (9,1,9) model. Comparison of the two prediction methods, Monte Carlo prediction with higher precision than ARIMA.