Time Trend And Prediction Model Of Meteorological Drought Over 1961-2016 In Shaanxi Province

Background:Drought is a major natural disaster that causes severe social and economic losses.Long-term variation help us to fully understand the occurrence and development of meteorological drought.The prediction of regional droughts may provide important information for drought preparedness and farm irrigation.The existing drought prediction models are mainly based on a single weather station.Efforts need to be taken to develop a new multistation-based prediction model.Understanding the mechanism of the effects of drought on vegetation can help us mitigate the damage of drought to crops.Objectives:The main purposes of this study include exploring the long-term trend of meteorological drought in Shaanxi Province in recent decades,optimizing the predictor selection process and developing a new model to predict droughts using past drought index,meteorological measures and climate signals from 32 stations during 1961 to 2016 in Shaanxi province,China.The effect of drought on vegetation index was also preliminarily analyzed.Methods:Firstly,we used Mann-Kendall trend test method to analyze the trend of frequency,duration and intensity of drought in Shaanxi province.In the prediction model of drought,we applied and compared two methods,including a cross-correlation function and a distributed lag nonlinear model(DLNM),in selecting the optimal predictors and specifying their lag time.Then,we built a DLNM,an artificial neural network model and an XGBoost model and compared their validations for predicting the Standardized Precipitation Evapotranspiration Index(SPEI)and drought categories 1-6 months in advance.In the study of the response of vegetation index to drought,we first calculated the correlation coefficient of three vegetation index and drought at 4 time scale based on 32 stations,and then obtained the correlation coefficient distribution map of the whole Shaanxi province by kriging.Results:In addition to individual stations,there is no significant change in the frequency,duration and intensity of drought in Shaanxi Province in the past 56 years.In the prediction of drought,the DLNM was better than the cross-correlation function in predictor selection and lag effect determination.The XGBoost model more accurately predicted SPEI with a lead time of 1-6 months than the DLNM and the artificial neural network,with cross-validation R2 values of 0.68-0.82,0.72?0.89.0.81?0.92,and 0.84?0.95 at 3?,6-,9-and 12-month time scales,respectively.Moreover,based on the SPEI 12,the XGBoost model had the highest prediction accuracy for overall droughts(89%?94%)and for three specific drought categories(i.e.,moderate,severe,and extreme)(89%?94%?88%?95%and 89%?97%).The correlation between vegetation health index and SPEI is the strongest,and it is positively correlated in most areas of Shaanxi.Conclusion:This study offers a new modeling strategy for drought predictions based on multistation data.The incorporation of nonlinear and lag effects of predictors into the XGBoost method can significantly improve prediction accuracy of SPEI and drought.By comparing the correlation between three vegetation indices and drought,a new idea is provided for the use of remote sensing vegetation index to monitor drought.