Semiarid Drought Characteristics In The Evolution Analysis And The Drought Prediction

Water resources gradually reducing has become a global people need attaches great importance to the problem,how to efficiently save water resources is more and more is also brought to the attention of the world.Global warming and other global problems as climate,drought occurred more and more frequent.The increase in population and economic progress is one of the reasons for water loss and bad environment gradually all over the world.The problems caused by the drought and the survival of human are closely linked,drought also led to huge economic losses,also affect the normal operation of human society.Ye Baishou area located in the west of liaoning province,belongs to the typical semi-arid sub-humid areas of China,uneven distribution of rainfall leads to frequently drought,affects the spring so wing,too.So we must according to the meteorological data of the past to explore characteristics of drought in the region so as to provide theoretical basis for the development of agricultural economy.This paper first analyzed the Palmer drought index of Ye Baishou regional drought conditions in 1952-2015,and uses wavelet decomposition and apply the PDSI multiresolution analysis,and then study the drought occurred in the region level,frequency,coverage,cycle,such as mutation time evolution characteristics and laws.After using the weighted markov chain to predict drought in Ye Baishou state reuse the gray residual error based on BP neural network model and time series prediction method for Ye Baishou area rainfall prediction.The main conclusions include:(1)Use the Palmer drought index for Ye Baishou region variation characteristics of annual drought in 64 years,to calculatethe drought occurrence frequency,drought coverage calculation,and try to analysis the drought characteristics.The analysis results show:Ye Baishou region has 11 years without the drought occurred in 64,light light drought and waterlogging in the year for 11 years and 9 years respectively,the drought and drought year of 7 years and 6 years respectively,the rest of the year of drought degree of are below 5 years and five years;Drought frequency average of 5.72 months,spring,summer,autumn and winter seasons the drought occurrence frequency were 1.3,1.59,1.56,and 1.41 times respectively;More than the light drought coverage up to 100%of the year for 37 years,average coverage of 82.6%,in 1978 and 1979,only six light drought monitoring station has not been happened.(2)On the basis of five standard weighted markov flood and drought,establish four standard weighted markov model,both two models can accurately predict Ye Baishou region of drought condition and predict the future state of the drought in the region but for the sake of drought resistance and disaster mitigation,we choose the first level 4 and drought indexes weighted markov model first(3)Using BP neural network to modify gray residual sequence,a new model.The results showed that the BP neural network gray residual error correction model is more suitable for Ye Baishou area rainfall forecast.(4)Adopt the method of time series,after the stationarity test,refers to AIC and SC determine using ARIMA(2,1,3)for Ye Baishou area rainfall prediction.With the rainfall forecast in 2014 and 2015,after with the actual result comparing the relative error is only 3.8%and 3.8%.And forecast the Ye Baishou area rainfall in 2016 and 2016,it comes to 406.1 mm and 416.91 mm respectively.