Abstract:
In this study, anomalies of terrestrial water storage (TWS) observed by the Gravity Recovery and Climate Experiment (GRACE) were assimilated into the NASA's Catchment land surface model in western and central Europe for a 7-year period, using a previously developed ensemble Kalman smoother (EnKS) which was designed to disaggregate GRACE observations vertically and horizontally while nudging modeled TWS towards GRACE TWS. Validation against gauged stream flow data showed that GRACE data assimilation improved runoff correlations in 17 out of 18 hydrological basins, even in basins smaller than the foot print of GRACE. Improvements in root zone soil moisture are less conclusive primarily due to the short in situ data period. In addition to improving temporal correlation, GRACE data assimilation also corrected the large increasing trend in monthly TWS and runoff, which was initiated by inaccurate trends in monthly precipitation forcing data and the model's response to the trend of increasing precipitation. The assimilated root zone soil moisture and TWS exhibited significant changes in their dryness ranks, suggesting that GRACE data assimilation could have a substantial impact on drought monitoring. Comparisons of GRACE TWS with MODIS Normalized Difference Vegetation Index (NDVI) showed that GRACE TWS identified droughts with more clarity and persistency than NDVI whose detection of water shortage may be affected by seasons. While both sources detected the same droughts, they may disagree in severity and recovery time.