Towards Generating Long-term AMSR-based Soil Moisture Data Record

Wednesday, 17 December 2014
Iliana E Mladenova, Thomas J Jackson, Rajat Bindlish and Michael H Cosh, U. S. Dept. of Agriculture, Beltsville, MD, United States
Research done over the past couple of years, such as Jung et al. (Nature, 2010) among others, demonstrates the potential for using soil moisture as an indicator and parameter for identifying long-term changes in climate trends. The study mentioned links the reduction in global evapotranspiration observed after the 1998 El Nino to decline in moisture supplies in the soil profile. Due to its crucial role in the terrestrial cycles and the demonstrated strong feedback with other climate variables, soil moisture has been recognized by the Global Climate Observing System as one of the 50 Essential Climate Variables (ECVs). The most cost and time effective way of monitoring soil moisture at global scale on routine basis, which is one of the requirements for ECVs, is using satellite technologies. AMSR-E was the first satellite mission to include soil moisture as an operational product. AMSR-E provided us with almost a decade of soil moisture data that are now extended by AMSR2, allowing the generation of a consistent and continuous global soil moisture data record. AMSR-E and AMSR2 are technically alike, thus, they are expected to have similar performance and accuracy, which needs to be confirmed and this the main focus of our research. AMSR-E stopped operating at its optimal rotational speed about 6 months before the launch of AMSR2, which complicates the direct inter-comparison and assessment of AMSR2 performance relative to AMSR-E. The AMSR-E and AMSR2 brightness temperature data and the corresponding soil moisture retrievals derived using the Single Channel Approach were evaluated separately at several ground validation sides located in the US. Brightness temperature inter-comparisons were done using monthly climatology and the low spin AMSR-E data acquired at 2 rpm. Both analyses showed very high agreement between the two instruments and revealed a constant positive bias at all locations in the AMSR2 observations relative to AMSR-E. Removal of this bias is essential in order to ensure consistency between both instruments. The corresponding soil moisture retrievals from AMSR-E and AMSR2 demonstrated reasonable agreement relative to in situ data. A detailed discussion that focuses on this analysis as well as possible approaches for removing the observed bias in the brightness temperature observations will be presented.