Quantifying Noise in Daily GPS Height Time Series: Harmonic Function Versus GRACE-Assimilating Modeling Approaches

Abstract

The Global Positioning System (GPS) is routinely used to measure the elastic response of Earth to continental hydrological mass changes, occurring at various temporal and spatial scales. While long-term and seasonal height changes are well observed in GPS height time series, the subseasonal deformations are less well resolved. A suite of predictions based on a remote sensing based Gravity Recovery and Climate Experiment (GRACE)-assimilating land surface model suggest significant high-frequency changes observable in GPS height time series (with periods between 15 and 90 days). This physical model is then adopted for separating the deterministic component of time series from the stochastic noise and is compared with the conventional harmonic functions modeling, a commonly used approach with predefined annual and semiannual periods. The noise parameters (spectral index and amplitude) associated with the two methods are estimated using the maximum likelihood estimation. We conclude that the GRACE-assimilated model output removes the effect of high-frequency hydrological deformations, producing less correlated residuals. Among other aspects, our results highlight the importance of assimilating GRACE-based remotely sensed total water storage data into hydrological models to obtain unbiased estimates of GPS vertical velocity and its uncertainty which is in demand in a range of applications such as the upcoming reference frame realizations.