Utility of GPS vertical displacements in terrestrial water storage monitoring

Permanent GPS (Global Positioning System) stations measure land surface vertical displacements (dz) resulting from changes in terrestrial water storage (ΔS). Like GRACE (Gravity Recovery and Climate Experiment), GPS dz senses changes in mass. However, GPS dz has considerably higher spatial resolution where dense station networks exist. ΔS induces elastic deformation of the solid Earth, which is sensed by GPS satellites as millimeter-scale changes in the vertical positions of ground-based receivers. Dz time series from GPS station networks can be inverted to recover gridded ΔS estimates, providing novel information regarding the seasonal dynamics and distribution of ΔS. However, GPS dz cannot differentiate between ΔS in separate hydrologic reservoirs (e.g. groundwater versus snowpack). Moreover, GPS ΔS is subject to uncertainty related to spatial regularization of the inversion, rendering GPS dz unsuitable for estimating ΔS in basins smaller than ~10,000 km².

To overcome the issues with GPS-inverted ΔS, we evaluate the potential use of GPS dz as a mass constraint for land surface models (LSMs). We forward model expected dz at GPS stations in California, Oregon, and Washington, using input ΔS from the VIC (Variable Infiltration Capacity) and Noah LSMs in NLDAS-2 (North American Land Data Assimilation System). We find both spatial and temporal variability in the agreement between modeled and observed dz. Specifically, our results indicate that LSM ΔS is underestimated in mountainous regions and in wet years. While increasing mountain precipitation inputs improves dz agreement, it does not completely solve the problem. This work points to the potential future use of GPS dz to improve LSM parameterization.