Downscaling Of Passive Mircowave Soil Moisture Using Visible And Shortwave Infrared Drought Index

Soil moisture, as a key natural factor, provides material basis for terrestrial plants andsoil organisms by regulating the hydrothermal energy exchange. It has devastating impacts onecological environment construction and socio-economy, especially for arid and semi-aridregions. So far, the monitoring of soil moisture using remote sensing and geographicalinformation science (GIS) technology has been attracted wide attention. The passivemicrowave radiometer AMSR-E soil moisture data has advantages of high temporalresolution, simple and feasible data processing and preferable soil moisture sensitivity,whereas the low spatial resolution (25~40km) makes it only applicable for large-scale soilmoisture monitoring. Though the spaceborne visible light and thermal infrared data canachieve the spatial resolution of100m~1km, they are easily affected by weather conditions.Therefore, taking advantage of both specific spatial and temporal resolution effectively, howto use downscaling algorithm to integrate passive microwave and optical data has become oneof key problems of quantitative remote sensing of soil moisture.Choosing the western region of Jilin Province as the study area, this paper aims to use andownscaling algorithm to improve the spatial resolution of AMSR-E soil moisture data, and toachieve the monitoring and analysis of soil moisture at more accurate and smaller scale. Thebasic data includes SPOT-VGT images from2003to2012, AMSR-E soil moisture daily datain mid-April2005, and the relative and absolute soil moisture data measured frommeteorological stations. Firstly, multi-spectral data of SPOT-VGT are used to confirm thevalidity of SWIR for soil moisture detection.Through the correlation analysis among sixdrought indexes including VSDI (Visible and Shortwave Infrared Drought Index),LSWI(Land Surface Water Index), MSI(Moisture Stress Index), GVMI (Global VegetationMoisture Index),NDVI(Normalized Difference Vegetation Index), RVI (Ratio VegetationIndex) and measured soil moisture, VSDI can be the most suitable drought index for soilmoisture monitoring. Secondly, during mid-Apri2005, the curve estimation is constructedbetween VSDI and AMSR-E soil moisture data (with spatial resolution of25km), and theS-curve function fits the best with the correlation coefficient of0.673. Thirdly, soil moistureunder1km spatial resolution of study area is derived using the downscaling algorithm.According to the precision verification, the downscaling soil moisture has significantcorrelation with measured data. The high accuracy of prediction may imply that thedownscaling results effectively reflect the regional differences in soil moisture distribution,but the downscaling soil moisture may underestimate the actual values. Finally, the1kmspatial distribution of soil moisture data is analyzed. The results show that the average soilmoisture of western Jilin Province was under moderate drought status in mid-April2005. Soil moisture decreased from east to west obviously. From north to south, there were nosignificant differences. At the county scale, Fuyu had the highest soil moisture, while thelowest values in Tongyu indicated the worst drought condition. Soil moisture of wetland wasthe highest, followed by forest, farmland, unused land, while soil moisture in shrub areaminimum.