Validation of surface soil moisture from AMSR-E using auxiliary spatial data in the transboundary Indus Basin

Information on soil moisture is vital to describe various hydrological processes. Soil moisture parameters are normally measured using buried sensors in the soil. Alternatively, spatial and temporal characteristics of surface soil moisture are estimated through satellites. Advanced Microwave Scanning Radiometer on the Earth Observing System (AMSR-E) is one of such satellites that estimate surface soil moisture in an operational context. These estimates need validation prior to use in various hydrological and water management applications. Such validations are normally carried out using field measurements of soil moisture. This is not technically feasible in vast river basins such as the Indus Basin and for pixel sizes of 25 km x 25 km with non-homogeneous soils and land use. Therefore, AMSR-E data interpreted with Njoku model and posted by the National Snow and Ice Data Center (NSIDC) for the Indus Basin is evaluated by comparing it against auxiliary spatial data. The auxiliary data exists of (i) land use, (ii) rainfall from the Tropical Rainfall Measuring Mission (TRMM) satellite, (iii) seasonality of vegetation from SPOT-Vegetation and (iv) saturated water content (theta(sat)) inferred from soil maps. A strong relationship was observed between rainfall and surface soil moisture in the land use class "rainfed". Spearman's rank correlation coefficient (r(s)) between the soil moisture and rainfall ranged from 0.14 to 0.55 with a mean of 0.36. For irrigated land uses, r(s) ranged from 0.04 to 0.52 with a mean of 0.29 due to control of soil moisture by irrigation water supply. The temporal analysis of soil moisture data with vegetation time series showed resemblance with growth phenology. Higher Pearson's correlation coefficient (r) between the soil moisture and vegetation development was found for time lags of a few weeks. The daily maximum values estimated by AMSR-E ranged from 0.08 to 0.38 cm(3) cm(-3). The maximum values were near, but below theta(sat) limits for different soil types. AMSR-E captured the flooding processes during July and August 2010 by showing the soil moisture values to approximate the saturated soil moisture content for areas that are reported to be flooded. This suggests that the absolute AMSR-E soil moisture data from NSIDC are accurate in the upper range of land wetness. It is concluded that AMSR-E surface soil moisture data exhibits spatio-temporal behavior, and the trends agree with auxiliary spatial data sets. (C) 2011 Elsevier B.V. All rights reserved.