The Gravity Recovery and Climate Experiment (GRACE) is a NASA Pathfinder Mission that has been approved for launch in 2001, with the goal of measuring the Earth's gravity field with unprecedented accuracy at two to four week intervals. Because mass movements of water at and below the Earth's surface are a major contributor to the time dependent component of the gravity field, it is believed that satellite-based measurements obtained by GRACE could be interpreted to estimate changes in continental water storage, which includes soil moisture, groundwater, snow and ice, lakes, rivers, and vegetative water storage, at regional (200,000 km2) and larger scales. These measurements will provide an independent constraint on components of terrestrial water storage that are typically modeled with very little measured data for validation.
The goal of this research is to explore the ability of GRACE to observe the temporal and spatial variability of continental water storage and its component stocks, so that a framework for extracting hydrological information from the mission will already be in place when the satellites are launched in 2001. To that end, we have set forth three objectives: 1) To identify regions and time periods of water storage detectability in light of the expected accuracy of GRACE measurements at varying space-time scales, by establishing the spatial and temporal ranges of variability in terrestrial water storage as seen in currently available modeled and observational data; 2) Using modeled data and field observations, to characterize the contributions of the component stocks of terrestrial water storage and to determine if characteristic signatures of individual stocks exist, and to use this information to discover which components will be detectable by GRACE over what regions and time periods; 3) To determine the spatial and temporal resolutions of auxiliary hydrologic observations necessary to decompose GRACE-derived estimates of water storage variations into estimates of changes in individual stocks at different confidence levels, and to determine the errors that would be expected given existing and hypothetical monitoring networks