Drought Monitoring In Shandong Based On Multi-source Remote Sensing Data

Due to the increasing frequency of drought caused by global warming,It has become one of the key issues in the world.Drought monitoring can reduce the disasters caused by drought.Therefore,it is of great practical importance to select an appropriate drought index to achieve accurate monitoring of drought.This paper takes Shandong Province as the study area.We study the meteorological drought and agricultural drought in Shandong Province from 2001 to 2017.The main research contents are as follows:(1)Based on the TRMM Precipitation Data of Shandong Province from 2001 to2017,the accuracy of TRMM was verified on three scales of month,year and station.In this paper,the variation coefficient of monthly average precipitation,seasonal average precipitation and the number of non-rainy months are calculated to analyze the variation of precipitation in Shandong Province.The meteorological drought index PCI of Shandong Province from 2001 to 2017 was calculated as the data set to study the variation characteristics of meteorological drought in Shandong Province.The results show that it has high accuracy and good applicability in different scales of TRMM data in Shandong Province.Summer is the season with the most precipitation in Shandong,while winter is the least,which makes Shandong prone to severe drought in winter.From 2001 to 2017,the northwest of Shandong Province was prone to meteorological drought,and the frequency of meteorological drought increased gradually,and the period became shorter.(2)The correlation coefficient R between Standardized Precipitation Evapotranspiration Index(SPEI)and Soil Moisture(SM)at different time scales was calculated.It was found that the R of SPEI-3 and SM at 3-month time scale was the largest,which could monitor agricultural drought.The three machine learning methods including the bias corrected random forest(BRF),support vector regression(SVR)and Cubist model were used to integrate multiple drought factors to simulate(SPEI-3),and according to the best model,the relative importance of each drought factor was obtained.Then the spatial distribution map of Multi-factor Synthetic Drought Index MSDI was drawn.The results show that among the three models,the accuracy of SPEI-3 value of BRF model is the highest.Most of the simulated values are consistent with the observed values.MSDI spatial distribution map can accurately monitor the agricultural drought in Shandong Province on site and spatial scales.(3)According to the spatial distribution of MSDI in Shandong from 2001 to 2017,the drought situation in Shandong was analyzed from different time scales and spatial scales.The results show that from 2001 to 2017,the number of severe drought months in Shandong Province began to appear scattered.The drought situation changed from no drought to light drought,and the frequency of light drought is also increasing.In addition,the degree of drought in Shandong Province fluctuated greatly from 2001 to 2007,and the degree of drought in Shandong Province was relatively stable from 2008 to 2017.On a spatial scale,in the growing season of winter wheat,the central and northwest parts of Shandong Province are prone to drought.From 2001 to 2017,Shandong Province is prone to drought in winter.The frequency of severe drought and moderate drought is the most in winter,followed by spring and autumn.For dry years,severe drought will occur in summer due to high temperature.