Study On The Spatiotemporal Evolution Characteristics And Drought Prediction Of Drought In Yulin Of Northern Shaanxi

In the context of climate warming,drought problems have become increasingly prominent.Yulin in northern Shaanxi is an important area of ecological protection strategy in the Yellow River Basin and an important agricultural area in Shaanxi Province.Frequent drought has a serious impact on local agricultural production and ecological protection.Therefore,the appropriate drought index is selected to analyze the spatial and temporal evolution and distribution characteristics of drought in Yulin,northern Shaanxi,and to predict the future drought,so as to provide theoretical support for the construction of drought prediction and early warning system and the study of drought prevention and disaster reduction measures in Yulin.Based on the monthly precipitation,air temperature and soil moisture data with a spatial resolution of 0.25° × 0.25° from 1951 to 2022,this paper compares the applicability of standardized precipitation index SPI,standardized precipitation evapotranspiration index SPEI,standardized soil moisture index SSMI,relative humidity index MI and precipitation anomaly percentage PA in drought assessment of Yulin City on 3-month and 6-month scales.The temporal and spatial evolution characteristics of drought in Yulin City from 1951 to 2000 were revealed.The prediction effects of the least squares support vector machine model LSSVM based on the preferred index construction,the pelican optimization algorithm combined with the feedforward neural network model POA-BPNN,and the variational mode decomposition combined with the long-term and short-term memory neural network model are analyzed.On this basis,based on the SPEI6 sequence of Yulin City from 2001 to 2022,the VMD-LSTM model is used to predict the drought development trend of Yulin City from 2023 to 2027 and conduct spatio-temporal pattern mining.The main research results are as follows:(1)A comparison was made on the recognition ability of the spatial and temporal distribution of historical drought disasters in Yulin City from 1951 to 2000 based on five drought indices.In terms of the recognition ability of the temporal distribution of historical drought disasters,SPEI3 and SPEI6 were able to identify 25 out of 30 major drought disasters from 1951 to 2000,with a recognition rate of 83%,which is significantly better than the other four indices(SSMI,SPI,MI,PA can identify 18 out of 30,15 out of 30,12 out of 30,and 8 out of 30,respectively).In terms of the ability to identify the spatial distribution of historical drought disasters,taking 1997 as an example,SPEI shows that the spatial distribution of drought in Yulin City is characterized by severe drought in most areas,and only moderate drought in Fugu County,which is more in line with reality.The description of drought distribution by MI and PA does not match the literature;SPI and SSMI describe the drought situation slightly.(2)Based on SPEI3 and SPEI6,the spatiotemporal evolution characteristics of drought in Yulin City from 1951 to 2000 were explored.The M-K trend test results of SPEI3 and SPEI6 sequences from 1951 to 2000 showed a drying trend throughout the city,and the drying trend gradually weakened from east to west.The regions with the most significant trend of worsening drought are in the northern part of Fugu County,some areas in the northern part of Shenmu City,and Qingjian County.The SPEI3 and SPEI6 sequences from 1951 to 2000 both have the first main cycle 39a and the second main cycle 8a;The periodic characteristics of SPEI6 are more significant.The SPEI3 and SPEI6 sequences from 1951 to 2000 underwent a mutation from wet to dry in 1982.From 1951 to 2000,the drought in Yuyang District of Yulin City was the most severe(with an average drought intensity of-0.71 and a frequency of 178 droughts),and the drought lasted for a long time(with an average annual drought duration of 3.2 months);The degree of drought in Hengshan District and Jingbian County is relatively severe,and the duration of the drought is relatively short;The drought in Shenmu City and Fugu County lasted for a long time,but the degree of drought was relatively mild.Qingjian County has the shortest duration of drought and the lightest degree of drought(drought intensity is-0.52,drought frequency is 151,and drought duration is 2.7 months).(3)Based on the SPEI3 and SPEI6 sequences from 1951 to 2000,a comparison was made on the prediction performance of three models:LSSVM,POA-BP,and VMD-LSTM.Among the three models,VMD-LSTM had the smallest prediction error,and the five model validation indicators were better than the other two models(MAE,MSE,RMSE,MAPE,0.074099,0.014953,0.13466,0.56551,respectively).The model had the highest prediction accuracy,followed by LSSVM,and POA-BPNN had the worst prediction accuracy.The test results of the three models based on SPEI6 have errors smaller than SPEI3,and the model validation indicators are better than SPEI3,indicating that the SPEI6 sequence is more suitable for drought prediction.(4)Based on the SPEI6 sequence from 2023 to 2027,a spatiotemporal cube model was established to predict the drought development trend in Yulin City from 2023 to 2027.In the next 5 years,there will be a trend of drying in the eastern part of Dingbian County and Suide County,and the trend is relatively stable;The western part of Dingbian County,Hengshan District,and Yuyang District are developing towards a trend of reducing drought,and the degree of relief is increasing over time,which may enter a humid period.The southern part of Jingbian County,Qingjian County,Jia County,central part of Shenmu City,and Fugu County are experiencing an increasing trend of drought,and the trend is becoming increasingly severe.The overall drought situation in Jingbian County,Qingjian County,Jia County,and Shenmu City is relatively severe in terms of time and space;The western part of Dingbian County,Hengshan District,and a small part of Yuyang District are relatively severe compared to their neighboring areas(12 adjacent pillars)in terms of time and space.