2012 AGU Chapman Conference on Remote Sensing of the ...

Gan, Thian Y.Soil Moisture Retrieval From Microwave andOptical Remotely Sensed DataGan, Thian Y. 1 ; Nasreen, Jahan 11. Dept Civil & Enviro Engineerin, Univ Alberta, Edmonton,AB, CanadaThe objective of this research is to investigate thepotential of using the newly available, quad-polarized,RADARSAT-2 synthetic Aperture Radar (SAR) data in nearsurface soil moisture retrieval. 11 Radarsat-2 images have sofar been acquired over the Paddle River Basin (PRB), Alberta,Canada and 1575 soil samples, from 9 sites (agricultural,herbaceous and pasture land sites) have been collectedwithin the basin on those days when the RADARSAT-2satellite flew over the study site to obtain actual soilmoisture information. The popular theoretical IntegralEquation model (IEM), linear and nonlinear regressions wereused to retrieve soil moisture from the RADARSAT-2 SARdata. Normalized Difference Vegetation Index (NDVI) andLand Surface temperature (LST) from the optical sensor ofthe Moderate resolution Imaging Spectroradiometer(MODIS) have also been used as additional predictors in theregression algorithms. The combined use of HH, VV, and HVradar backscatters, LST and NDVI as the predictorsproduced more accurate soil moisture retrievals than usingonly individual/multiple radar backscatters as thepredictors. This is probably because the HH polarizedbackscatters can penetrate more than the VV counterpartsand hence together they provide more information about thesoil moisture. On the other hand the VV polarizedbackscatters are useful in determining vegetation growthstage, height, type and health while HV and VH polarizedbackscatters provide complementary information aboutvegetation structure. Therefore radar and optical datatogether could provide more information about the surfacecharacteristics and the effects of vegetation on soil moisturethan individual radar backscatters alone. Compared to fieldmeasurements, soil moisture retrieved from RADARSAT-2SAR data by the best regression and the IEM modelsachieved correlation coefficients of 0.89 and 0.91,respectively, at the watershed-scale when soil moisture wasaveraged over all 9 sites. . Retrieve soil moisture usingArtificial Neural Network and Support Vector Machine gavebetter results than that using regression and IEM models.Garg, R. D.Estimating Snow Water Equivalent (SWE) in thePart of North West Himalayan Catchment of BeasRiver, using Synthetic Aperture Radar (SAR) dataThakur, Praveen K. 1 ; Aggarwal, S. P. 1 ; Garg, P. K. 2 ; Garg, R.D. 2 ; Mani, Sneh 31. Water Resources Division, Indian Institute of RemoteSesning (IIRS), Dehradun, India2. Geomatics Eng., Indian Institute of Technolgy (IIT),Roorkee, India3. Avalanche forecasting group, Snow and AvalancheEstablishment SASE), Chandigarh, IndiaThe Snow Water Equivalent (SWE) of the seasonal snowcover can be an important component of the water cycle inmountainous areas, and the knowledge of this temporarystorage term may for example be very valuable for predictingseasonal discharge, for making short-range dischargeforecasts and also for assessing water quality aspects (Braun1991). The present study has been done to estimate the SWEby thermal inertia approach by using ENVISAT-ASAR data.The study area is the catchment area of Beas River up toManali, in part of North West Himalaya, with area of ~350km2. The algorithm used to recover the SWE from SAR datais made of two equations (Bernier and Fortin 1998, Bernieret al 1999). The first equation is the linear relationshipbetween the snow thermal resistance (R) and thebackscattering ratio between a winter image and a reference(snow-free) image. The second equation of the algorithminfers the SWE from the estimated snow thermal resistance(R) and a function of the mean density of the snow pack ().The current study has concludes that this approach can beused for bare soil and grassland land use class of study areaand the snow density is most important and sensitiveparameter for SWE estimation using thermal inertiaapproach.65