An Assessment of Global Positioning System Velocity Uncertainty in California

We analyze data from 580 continuous global positioning system (GPS) stations in California to quantify differences in published velocity estimates from five analysis centers. Horizontal and vertical rates for individual stations can differ up to 5 mm/yr, with systematic differences in some areas comparable to deformation rates. Published velocity uncertainties vary between analysis centers and are systematically underreported in the horizontal relative to empirical uncertainties calculated from the scatter of analysis center velocities. In the vertical, published velocity uncertainties are both over and underreported and vary more widely between centers. An interpolated ensemble vertical velocity field shows high-subsidence regions in the Central Valley and Salton Trough have the largest empirical uncertainties, while station density has a modest impact on uncertainties. Applications that rely on subcentimeter GPS accuracy should consider the possibility that formal errors published with velocity rate estimates understate true velocity uncertainties in both the horizontal and vertical. Plain Language Summary The continuous recordings from geodetic grade global positioning system (GPS) sensors provides high resolution ground motion measurements. Multiple analysis centers process the raw GPS data into daily station positions and provide high quality data to the scientific community. Each analysis center applies different processing techniques and model corrections that produce differences in the final time series product. We analyze the GPS positions and published velocities for five analysis centers and develop a composite velocity dataset with uncertainties for 580 stations in California. We also reanalyze station velocities using two standard methods to assess if the velocity differences between analysis centers arise from the underlying positions or the time series analysis. We find that horizontals positions are consistent, but vertical positions, which are an order of magnitude smaller, vary by analysis center and exhibit their greatest discrepancies in areas of largest observed subsidence. We further evaluate the vertical velocity field from all five analysis centers and develop an ensemble velocity field to characterize the spatially varying uncertainty. Our results demonstrate the importance of assessing position uncertainty using multiple analysis centers when informing geophysical models of observed ground motions.

Automated summary

The study found discrepancies in GPS station velocity estimates among different analysis centers, with vertical differences being more significant. Actual velocity uncertainties are often underestimated in the horizontal direction and may be over or underestimated in the vertical direction. Subsidence rates vary widely in different regions, with station density having a modest impact on uncertainties.