Research Projects

Soil moisture is a crucial factor in water and energy exchanges, and satellite-based surface soil moisture is expected to gain more attention in the next decade. This is supported by the upcoming NASA-ISRO Synthetic Aperture Radar (NISAR) and the ESA Soil Moisture and Ocean Salinity (SMOS) mission and the NASA Soil Moisture Active/Passive (SMAP) mission. However, these products show less correlation with ground truth over the Indian region and weak relation with root zone soil moisture due to their reliance on global datasets. The low penetration power of these sensors also limits their use in topographically complex regions like India. To address this issue, regional soil moisture techniques are needed for the Indian region and land surface models for sub-surface soil moisture conditions. State correction can be provided using satellite soil moisture and brightness temperature through data assimilation, improving the capability of WRF-LSMs for surface and sub-surface soil moisture conditions. Deep learning techniques can be integrated with mesoscale models through data assimilation to enhance prediction skills of surface and sub-surface soil moisture conditions. Several data assimilation approaches will be considered for integrating remotely-sensed soil moisture information into the LSMs model. Simplified Kalman filtering methodology, Convolutional Neural Network (CNN), and Long Short Term Memory (LSTM) will be used to correct rainfall input. Ensemble Kalman filter (EnKF) or smoother (EnKS) can also be applied to correct LSM soil moisture states based on available remotely-sensed surface soil moisture retrievals.