Application of a Time-Series-Based Methodology for Soil Moisture Estimation From AMSR-E Observations Over India

A time-series-based methodology has been used for the estimation of soil moisture from AMSR-E 10.7-GHz H-polarized brightness temperature over India. The estimated soil moisture has been validated with in situ soil moisture and compared with the NSIDC AMSR-E soil moisture product. The soil moisture estimated using the present algorithm shows improvement over NSIDC soil moisture products over India. The present algorithm shows an root-mean-squared error of 8.8% with in situ soil moisture, which is 12.9% for the NSIDC products over Indian agromet stations. The sensitivity in the soil moisture is captured well from the present algorithm; however, it is limited to regions with low to moderate vegetation cover having low temporal variability. The algorithm is location specific that is tuned over India to provide better estimates than achievable using the NSIDC global algorithm. This will help in improving the mesoscale monsoon prediction of rainfall over India.