Fitting And Prediction Of Station Drought Index And Establishment Of Comprehensive Drought Index In Shandong Area

Drought,which is a natural disease caused by various factors,has severely affected China's social and economic development and people's lives.Shandong Province is seleted as study area.And the ARMA-GARCH model is used to simulate and predict Standardized Precipitation Index(SPI)calculated by precipitation data derived from five topical meteorological station in study area;the multivariable linear regression model is used to establish composite remote sensing drought index MCDI to detect Shandong drought.The main research contents and conclusions of this dissertation include the following aspects:(1)Autoregressive moving average(ARMA)model is used to simulate and predict site index SPI,and the process of determining parameters is systematically illustrated.The Generalized Auto-Regressive Conditional Heteroscedasticity(GARCH)model is introduced in this dissertation to eliminate the heteroscedasticity effect,and the coupled model(ARMAGARCH)is established.We compare results of simulation and prediction generated by ARMA and ARMA-GARCH models and find that the performance of ARAM-GARCH model is better than ARMA model in both simulation and prediction,especially in prediction,ARMAGARCH model has significant advantage.(2)The correlation coefficients between the four single remote sensing drought indices Precipitation Condition Index(PCI),Soil moisture Condition Index(SMCI),Temperature Condition Index(TCI)and Vegetation Condition Index(VCI)and Standardized Precipitation Evapotranspiration Index(SPEI)are calculated to assess the capability of these single drought indices in monitoring Shandong drought.The results show that the correlation between PCI and 1-month SPEI(SPEI-1)is highest compared with other SPEIs,and the TCI also has the same characteristic: the r decreased as the SPEI time scale increased,indicating they are suitable to detect short-term drought of study area;SMCI and VCI are lagging on response to drought,and VCI has a longer lag time than SMCI,which suggests they can provide more effective information for monitor long-term drought.(3)The multivariable linear regression model is used to regress the four single remote sensing drought indices and different time scales SPEI to establish composite remote sensing drought index MCDI.Additionally,mediator and moderator variables are introduced to optimize the model.And,we compare the r values between the composite indices and SPEIs and the highest r values between single drought indices and SPEIs,the results show the former is higher than latter and improvement is from 0.04 to 0.12.(4)In-site meteorology indices MI and SPI and SMOS soil moisture are used to further verify the feasibility and reliability of MCDIs in monitoring drought in study area.The results show MCDI-1 has the highest correlation with SPI-1 and MI,suggesting MCDI-1 is suitable to detect meteorology drought;the correlation between MCDI-9 and soil moisture is the highest indicating MCDI-9 is fit to monitor agricultural and vegetation drought.