Agricultural Drought Monitoring Combining Ground-based Observations And Satellite-derived Data

Extreme climate change has a significant impact on the global environment.As a typical agro-meteorological disaster,drought greatly affects social and economic activities,food security and human survival.Therefore,it is particularly important to engaged in agricultural drought monitoring research.At present,there has been some studies about agricultural drought monitoring in remote sensing community,and multivariate integrated drought indices that based on multi-satellite sensor have been proposed,which provides new methods for agro-meteorological disaster monitoring.However,there are few researches on rice agricultural drought monitoring based on remote sensing technology.Therefore,this paper selects five provinces along the middle and lower reaches of the Yangtze River which includes Jiangsu Province,Anhui Province,Hubei Province,Hunan Province,and Jiangxi Province,as research areas,and carries out agricultural drought monitoring research in single-cropping rice planting areas with 3S technology and data fusion methods.The main research contents and results are as follows:(1)Precipitation estimates based on rain gauge observations and satellite product.This paper uses the ATPK+GWRK method to obtain high-quality monthly precipitation estimates from2000 to 2017,based on rain gauge observation data,TRMM3B43 precipitation data and geographic auxiliary data.The results show that,when the ATPK method is used to process TRMM precipitation estimates,its accuracy will improved,compared with the original data,among which RMSE and MAE are 36.46 mm,27.27 mm,and R~2 is 0.694;Then the GWRK method was applied to further discuss the precipitation fusion method;In the comparison of different parameter interpolation methods,Kriging is selected to process the parameters to reach higher accuracy;Finally,seven fusion models composed of multivariable combinations are verified to obtain the most suitable model.Results show that model 2(DTRMM?GWRK?Model2)has higher accuracy,RMSE,MAE,Bias and R~2 are 28.43 mm,20.69 mm,-0.07%and 0.772,respectively.(2)Research on MODIS LST interpolation method.Based on MODIS land surface temperature products,MODIS vegetation index products and geographic auxiliary data,this paper uses adaptive window interpolation to interpolate the gap pixels of land surface temperature products.The results are verified using air temperature and land surface temperature data,observed by meteorological stations,R2 are about 0.9,and relatively stable;In the verification of land surface temperature,the RMSE is about 4.5?.The results can be used to construct the remote sensing agricultural drought index.(3)Research and application of agricultural drought monitoring based on remote sensing.Based on the data obtained in the previous chapter,a series of drought indices including univariate and multivariate synthesis are constructed to monitor agricultural drought in the study area from2003 to 2017.The results show that the area affected by single-cropping rice was the largest in2013,accounting for 15.14%of the total planted area,of which Jiangsu and Anhui Provinces had the highest proportion,reaching 20.84%and 19.28%of their province planted area,respectively.The agricultural drought in 2004 and 2011 were second,and the proportion of affected areas reached 13.49%and 13.91%,respectively.Among them,the proportion of affected areas in all provinces was more than 10%in 2011.