Soil moisture retrieval using an index based approach and passive microwave data from AMSR-E

This poster addresses the retrieval of soil moisture using an index based approach and passive microwave data from the Advanced Microwave Scanning Radiometer (AMSR-E) for the Earth observing system on-board Aqua satellite. The approach is potentially capable of providing an indication of relative variation of surface wetness condition. Brightness temperature (T_B) at low frequencies, such as 6.9 GHz, and 10.7 GHz, has been using to develop an index using linear mixture model. The end members (here after as the T_B max and T_B min) are extreme value over particular area under driest and wettest conditions, respectively. Two different approaches to derive these end members are evaluated: First, the end members are obtained from all sub-pixels over a particular area (referred to as global) and second, the end members are obtained from pixel-per-pixel over a particular area (referred to as local). Accuracy of the index is assessed using statistics based on a regression coefficient (R). Based on regression analysis, the results indicate that index obtained from the end members derived at local level and using T_B at 10.7 GHz is capable of providing reasonable estimate of near surface soil moisture when compared with in-situ soil measurement. The good agreement (R=0.820, RSME=0.05) is achieved over Arizona (SMEX04) region with bare soil surface (mean NDVI<0.3). For the observed lowest correlation coefficient (R=0.416, RSME=0.07) is obtained over dense vegetation area (mean NDVI>0.5) over South Oklahoma. The impact of the vegetation effect on the index needs to be investigating. In addition, the index shows a good relationship with precipitation and volumetric soil moisture variation. However, results have shown that the index responds to precipitation with a lags time of about 1-2 days.