The Study Of Downscaling Microwave-derived Soil Moisture

Satellite-based passive microwave remote sensing technique has outstandingadvantages and extensive application foreground in the monitoring of soil moisture.The advantages include the high soil moisture sensitivity, the splendid penetrateability (especially for low vegetation and near-surface soil), and the excellent dataacquisition ability in all weather all-time.The disadvantage of passive microwave remote sensing technique mainly lies inthe low spatial resolution. In partial, the mean spatial resolution of the SMOS (SoilMoisture and Ocean Salinity) is40km, and AMSR-E (Advanced Microwave ScanningRadiometer-EOS) is25km, FY (FengYun) is also25km. The soil moisture parameterin regional scales, extracted from low resolution passive microwave remote sensingdata, can’t meet the needs of land surface model and hydrologic model. Andconsidering all these factors, the downscale research of passive microwave remotesensing soil moisture data means a lot both in scientific field and real application.In this paper, the study area is Zhangye, which located in the middle reaches ofHeihe river. Soil moisture data of FY-3B(spatial resolution is25km), optical data ofMODIS(Normalized Difference Vegetation Index， NDVI; and Land SurfaceTemperature, Ts; spatial resolution is1km), and weather station data were used.Specifically, the FY-3B data are used to derive soil moisture base, and the MODISdata are used to estimated daily variation of land surface soil moisture over a pixel of25km×25km. At last, we get the1-km resolution soil moisture through thedownscaling algorithm which is presented by Merlin. Verify the trend of1-kmresolution soil moisture with TVDI(Temperature Vegetation Drought Index) throughthe scatter diagram. And then verify the accuracy with practical measured data fromthe BNUNET observation dataset in the middle reaches of the Heihe River Basin.Combined with the accuracy evaluation results of this inversion algorism we can get the conclusion：1The soil moisture data and TVDI have good consistency, the correlationcoefficient was0.66. It can accurately reflect the surface soil moisture. The soilmoisture retrial from the downscaling algorithms and the practical measured data allhave a tendency to rise.2The distribution of soil moisture is consistent with the distribution of land usetype that the irrigated farmland has the high soil moisture for the high water retentionability when the bare land, grass land, and sparse vegetation area have the relativelylow soil moisture.3The distribution of soil moisture is continuous in both sides of originalmicrowave image pixels. And eliminate the influence of block structure,This research demonstrates that the soil moisture value got from downscalemodel has high confidence level that can make the demands of parameter changeresearch in small-scale. And with the help of MODIS data and high temporalresolution passive microwave data we can get the soil moisture dynamic oflong-time-series in a simple way. At the same time, we should consider that theimprovement in passive microwave data quality, and the correct selection of bothparameters and quality evaluation index may narrow the gap between real-data andexperimental result.