Reconstructing GRACE-like TWS anomalies for the Canadian landmass using deep learning and land surface model

Terrestrial water storage (TWS) is an essential part of the global water cycle. Long-term information of observed and modeled TWS is fundamental to analyze water resources, meteorological extreme events (e.g., droughts and floods), and the climate change impacts. Over the past several decades, hydrologists have been applying physically-based hydrological model (GHM) and land surface model (LSM) to simulate TWS and its components (e.g., groundwater storage). However, the reliability of these physically-based models is often affected by uncertainties in climatic forcing data, model parameters, model structure, and mechanisms for physical process representations. Launched in March 2002, the Gravity Recovery and Climate Experiment (GRACE) satellite mission exclusively applies remote sensing techniques to measure the variations in TWS on a global scale. The mission length of GRACE, however, is too short to meet the requirements for analyzing long-term TWS. Therefore, lots of effort have been devoted to the reconstruction of GRACE-like TWS data for the pre-GRACE era. Data-driven methods, such as multilinear regression and machine learning, exhibit a great potential to reconstruct TWS data by integrating GRACE observations and physically-based model simulations. The advances in artificial intelligence enable adaptive learning of correlations between variables in complex spatiotemporal systems. However, the applicability of various deep learning techniques has not been adequately studied for GRACE TWS reconstruction. In this study, three deep learning-based models are developed to reconstruct the historical TWS using LSM outputs for the Canadian landmass from 1979 to 2002. The performance of the models is evaluated against the GRACE-observed TWS in 2002–2004 and 2014–2016. The trained models achieve a mean correlation coefficient of 0.96, with a mean RMSE of 53 mm. The results show that the LSM-based deep learning models significantly improve the correlations between original LSM simulations and GRACE observations.