Comparative Study On Agricultural Drought Monitoring Methods Based On Multi-source Remote Sensing Data

Agricultural drought is a natural disaster closely related to human production and life,which has the most direct impact on agricultural production and rural economic development.At present,drought index based on meteorological observation data can not effectively monitor agricultural drought characteristics at regional scale.While remote sensing technology can effectively provide the continuously distributed data in time and space,which has become an important means of drought monitoring at regional scale.Therefore we first selected 8 remote sensing drought indices that can characterize drought conditions.Then,comparing them with the drought index calculated from station measured data and the field measured crop yield data.Last,we determined the most suitable remote sensing drought index to monitoring the monthly drought of cultivated land in the Loess Plateau.And the index was used to research the temporal and spatial variation characteristics of agricultural drought in the Loess Plateau.In this paper,eight remote sensing drought indices including vegetation state index(VCI),temperature state index(TCI),precipitation state index(PCI),soil water state index(SMCI),drought stress index(DSI),temperature index state index(TVDI),scale drought condition index(SDCI)and land water storage change(TWSC)at monthly time scale during the crop growing season(April-October)in the Loess Plateau from 2003 to 2015 were calculated based on the TRMM precipitation(3B43)data,AMSR-E soil moisture data,GRACE terrestrial water reserves data and MODIS surface temperature(LST),evapotranspiration(PET,ET),normalized vegetation index(NDVI)remote sensing data from 2003 to 2015,and monthly mean temperature and precipitation data from 92 meteorological stations in and around the Loess Plateau from 1960 to 2015.Then,the reliability of the above remote sensing drought index was evaluated using the standardized precipitation index(SPI)and the standardized evapotranspiration index(SPEI)based on the precipitation and temperature data from meteorological stations.In addition,the vegetation state index(VCI),temperature state index(TCI),drought stress index(DSI)and temperature index state index(TVDI)of Shanxi Province in the growing season(April-October)of 2003-2015 were calculated based on the remote sensing data of surface temperature(LST),normalized vegetation index(NDVI)and evapotranspiration(PET,ET).The drought monitoring performance of the above four remote sensing drought indices in cultivated land of Shanxi Province was evaluated using the data of grain crop yield per unit area,winter wheat yield per unit area,drought disaster area and crop yield per unit area in Shanxi Province from 2003 to 2015.The drought distribution characteristics in typical drought years of Shanxi Province and the effects of agricultural drought on winter wheat yield were further discussed.Through this study,the following main conclusions are drawn:(1)Within cultivated land,SMCI,PCI and TWSC can not monitor the agricultural drought conditions of the Loess Plateau very well.VCI,TCI,TVDI,SDCI and DSI have their own advantages in monitoring the agricultural drought conditions of the Loess Plateau.Comprehensive comparison shows that DSI and SDCI have better drought measurement performance,but SDCI which takes precipitation factors into account more than DSI has more obvious advantages in space-time scale.So DCI is most suitable for monitoring agricultural drought conditions of the Loess Plateau.(2)The degree of drought in the growing season of the Loess Plateau from 2003 to 2015 was moderate and gradually increases from southeast to northwest in space.The interannual change of cultivated land drought showed a slowly decreasing trend.The degree of drought in 2006 was the most serious and in 2014 is the least.The annual change of cultivated land drought showed a sustained reduction from April to August.The drought continueed to worsen from August to October,with the lightest in August and the worst in April.(3)We compared the four remote sensing drought indices,VCI,TCI,TVDI and DSI of the existing 8 day time scales with the disaster area,the area of crop failures,the yield per unit area of grain crops and the yield per unit area of winter wheat.DSI is the best remote sensing drought index for monitoring agricultural drought in Shanxi Province on an 8-day time scale.TVDI performs well,while TCI and VCI have the worst drought measurement ability in this area.(4)From 2003 to 2015 in Shanxi Province,drought had the most significant effect on winter wheat yield from mid-April to mid-May.From 113 to 137 days,winter wheat was in jointing,heading and filling period which was the most critical period for its growth.Due to the winter-spring drought lasted for a long time(57-121 days)in Shanxi Province in 2005,spring drought had a great impact on the yield of winter wheat.