Quantifying the Spatiotemporal Resolution Limits of GNSS Water Storage Estimates Inferred from Earth Surface Displacements

In the western US, water resources used for agricultural, domestic, and industrial purposes are largely derived from high-elevation watersheds. As a result of increasingly severe annual drought and growing population, the need for finer-scale estimates of terrestrial water storage (TWS) for individual watersheds has grown. Current hydrologic models commonly underestimate TWS, as topography, climate, lateral heterogeneity in geology, and subsurface changes in storage are difficult to incorporate in hydrologic models [Argus et al., 2017]. The Gravity Recovery and Climate Experiment Follow-on (GRACE-FO) TWS measurements offer accurate measurements at a continental scale (~300-400 km), but coarse spatiotemporal resolution and a one-month data latency make it unreliable as a tool for water management at a local scale [Fu et al., 2015]. Other recent advances in space geodesy, such as Global Navigation Satellite Systems (GNSS), allow for higher-resolution estimates of TWS to be made from observing deformation of Earth’s surface resulting from the redistribution of water. Changes in water mass on/beneath Earth’s surface produce vertical and horizontal elastic deformation of the Earth that is observable by GNSS antennas. Before estimates of TWS can be made for mountain watersheds (e.g., Hydrological Unit Code (HUC)-4 and smaller) using real geodetic observations, it is crucial to determine the spatiotemporal resolution in which TWS can be estimated for a given network of GNSS antennas. The spatial scale in which TWS can be derived is dependent upon the density and configuration of GNSS antennas within a region, a-priori hydrologic information, and inversion algorithm/parameters. Previous studies have been shown to estimate TWS at a regional scale [e.g., Borsa et al. 2014]; however, TWS estimates have yet to be made for individual mountain watersheds. With new advances in approach, such as incorporating a priori information from existing hydrology models and strategic deployment of GNSS antennas, spatiotemporal resolution of TWS estimates is likely fine enough for estimates to be made for individual watersheds. Using LoadDef [Martens et al., 2019], a modeling software used to model elastic deformation of the Earth, a series of synthetic loads will be forward modeled and inverted for the pre-existing Selway-Lochsa GNSS network located in the Bitterroot and Sapphire Mountains along the Montana-Idaho border, the first GNSS network specifically designed to observe changes in TWS at a watershed-scale. Synthetic loads will take the form of checkerboards, containing square loaded and unloaded areas of varying scale, as well as real hydrologic datasets such as SNODAS snow water equivalent (SWE) estimates and NLDAS Soil Moisture estimates to determine how effectively the existing network configuration can return a given synthetic load with varying spatial resolution and distribution. Additionally, a synthetic GNSS network can be designed and rearranged, including additional synthetic GNSS stations, using LoadDef to determine the optimal network configuration for estimating TWS at a watershed-scale and to determine the impact of adding additional stations to the existing network. Placing quantitative limits on model resolution advances understanding of TWS distribution and uncertainties as well as optimal GNSS network design. By improving the ability to accurately estimate TWS in mountain watersheds, we improve the ability to monitor and manage freshwater resources in the context of a changing climate