This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Remotely sensed terrestrial water storage changes (TWSC) from the past Gravity Recovery and Climate Experiment (GRACE) mission cover a relatively short period (≈15 years). This short span presents challenges for long-term studies (e.g., drought assessment) in data-poor regions like West Africa (WA). Thus, we developed a Nonlinear Autoregressive model with eXogenous input (NARX) neural network to backcast GRACE-derived TWSC series to 1979 over WA. We trained the network to simulate TWSC based on its relationship with rainfall, evaporation, surface temperature, net-precipitation, soil moisture, and climate indices. The reconstructed TWSC series, upon validation, indicate high skill performance with a root-mean-square error (RMSE) of 11.83 mm/month and coefficient correlation of 0.89. The validation was performed considering only 15% of the available TWSC data not used to train the network. More so, we used the total water content changes (TWCC) synthesized from Noah driven global land data assimilation system in a simulation under the same condition as the GRACE data. The results based on this simulation show the feasibility of the NARX networks in hindcasting TWCC with RMSE of 8.06 mm/month and correlation coefficient of 0.88. The NARX network proved robust to adequately reconstruct GRACE-derived TWSC estimates back to 1979.

The terrestrial water storage anomaly (TWSA) from the previous Gravity Recovery and Climate Experiment (GRACE) covers a relatively short period (15 years) with several missing periods. This study explores the boosted regression trees (BRT) and the artificial neural network (ANN) to reconstruct the TWSA series between 1982 and 2014 over the Yangtze River basin (YRB). Both algorithms are trained with several hydro-climatic variables (e.g., precipitation, soil moisture, and temperature) and climate indices for the YRB. The results from this study show that the BRT is capable of reconstructing TWSA and shows Nash–Sutcliffe efficiency (NSE) of 0.89 and a root-mean-square error (RMSE) of 18.94 mm during the test stage, outperforming ANN in about 2.3% and 7.4%, respectively. As a step further, the reliability of this technique in reconstructing TWSA beyond the GRACE era was also evaluated. Hence, a closed-loop simulation using the artificial TWSA series over 1982–2014 under the same scenarios for the actual GRACE data shows that BRT can predict TWSA (NSE of 0.92 and RMSE of 6.93 mm). Again, the BRT outperformed the ANN by approximately 1.1% and 5.3%, respectively. This study provides a new perspective for reconstructing and filling the gaps in the GRACE–TWSA series over data-scarce regions, which is desired for hydrological drought characterization and environmental studies. BRT offers such an opportunity for the GRACE Follow-On mission to predict 11 months of missing TWSA data by relying on a limited number of predictive variables, hence being adjudged to be more economical than the ANN.