On the Use of GPS to Estimate Terrestrial Water Storage Variations in the Western United States: A Comparison of Seasonal Hydrological Loading Inferred from GPS, GRACE, and the NASA Catchment Land Surface Model

Seasonal variations in terrestrial water storage (TWS) cause the Earth's crust to deform. Observations of surface displacements (or deformation), in turn, can be used to study terrestrial hydrology after properly accounting for non-hydrologic loadings. A comparison of hydrologic loading estimates derived from the Gravity Recovery and Climate Experiment (GRACE) mission, ground-based Global Positioning System (GPS) sensors, and the NASA Catchment Land Surface Model (Catchment), in terms of both vertical deformation and TWS, was conducted over the Western United States with a particular focus in snow-dominated basins. To assess vertical displacements derived from GPS, GRACE, and Catchment, the effects of atmospheric loading, non-tidal ocean loading, and glacier isostatic adjustment are first removed from the GPS vertical displacements observations. Statistics relating: 1) GRACE-derived versus GPS-derived vertical displacements, and 2) Catchment-derived versus GPS-derived vertical displacements are presented. Results show that over 75% stations provide a correlation coefficient R>0.7 between GPS-based and GRACE-based vertical displacements, indicating a good agreement. The inconsistencies are mostly associated with anthropogenic activities such as groundwater pumping. The study also investigates the inversion of vertical displacement variation from the GPS observations into TWS space in order to compare seasonal variations in TWS as derived from GPS, GRACE, and Catchment. Preliminary results show that GPS can be used as an independent measurement of TWS estimation with similar temporal and spatial trends as found in the GRACE-based TWS estimates. The relatively good agreement between GPS- and GRACE-derived vertical displacements, as well as the capability of GPS measurement for estimating TWS, suggests potential for merging ground-based GPS- and GRACE-based estimates into the Catchment model via Bayesian conditioning.