Global analysis of approaches for deriving total water storage changes from GRACE satellites and implications for groundwater storage change estimation

Increasing interest in use of GRACE satellites and a variety of new products to monitor changes in total water storage (TWS) underscores the need to assess the reliability of output from different products. The objective of this study was to assess skills and uncertainties of different approaches for processing GRACE data to restore signal losses caused by spatial filtering based on analysis of 1°×1° grid scale data and basin scale data in 60 river basins globally. Results indicate that scaling factors from six land surface models (LSMs), including four models from GLDAS-1 (Noah 2.7, Mosaic, VIC, and CLM 2.0), CLM 4.0, and WGHM, are similar over most humid, sub-humid, and high-latitude regions but can differ by up to 100% over arid and semi-arid basins and areas with intensive irrigation. Large differences in TWS anomalies from three processing approaches (scaling factor, additive, and multiplicative corrections) were found in arid and semi-arid regions, areas with intensive irrigation, and relatively small basins (e.g., ≤ 200,000 km2). Furthermore, TWS anomaly products from gridded data with CLM4.0 scaling factors and the additive correction approach more closely agree with WGHM output than the multiplicative correction approach. Estimation of groundwater storage changes using GRACE satellites requires caution in selecting an appropriate approach for restoring TWS changes. A priori ground-based data used in forward modeling can provide a powerful tool for explaining the distribution of signal gains or losses caused by low-pass filtering in specific regions of interest and should be very useful for more reliable estimation of groundwater storage changes using GRACE satellites.