Difficulties in Assessing Reliability of Groundwater Storage Changes from GRACE Satellite Data

International audienceThere is increasing interest in use of GRACE satellite data to monitor changes in groundwater storage. The objective of this study was to assess the reliability of storage change estimates from GRACE using comparisons with ground-based data. Groundwater basins with large signals from groundwater depletion due to irrigation were selected and include the High Plains and Central Valley aquifers in the US. The High Plains aquifer is the most straightforward because it is large (450,000 km2), and has only two storage components (soil moisture and groundwater). The aquifer is unconfined; therefore, converting water level changes to water storage is relatively straightforward. Good correlations were found between groundwater storage change estimated from GRACE and ground-based data from 2002 through 2007. The Central Valley aquifer is much more difficult to analyze because it is much smaller (52,000 km2), and has a large number of storage components (snow, surface water, soil moisture, and groundwater) with different spatial distributions. The aquifer varies from unconfined to confined; therefore, storage coefficients vary by orders of magnitude. Processing the larger river basins that include the Central Valley increases the area to 154,000 km2. Snow data are available from SNODAS and surface water reservoirs are monitored directly. Groundwater levels are monitored; however, it is difficult to determine which wells are penetrating unconfined, semiconfined, or confined portions of the aquifer, increasing uncertainties in water storage estimates from water level data. Groundwater depletion was generally restricted to the 2006 through 2009 drought and depletion from GRACE (25 - 26 km3) compared favorably with estimates ground-based data (23 km3). Improving estimates of groundwater storage changes will require quantification of inputs and outputs to aquifers, incorporating irrigation into land surface models to provide reliable data on soil moisture storage changes, and improved estimates of groundwater storage from water level or potentiometric surface data. While some suggest that satellite data may replace ground-based monitoring, reliable estimates of groundwater storage from GRACE require strong integration of satellite and ground-based data and modeling analyses