Vertical Displacements of the Amazon Basin From GRACE and GPS

Abstract

The extent to which Gravity Recovery and Climate Experiment (GRACE)-recovered gravity anomalies can improve our understanding of Global Positioning System (GPS)-measured vertical displacements is currently uncertain. To address this issue, we compared vertical displacements measured by 23 GPS stations in the Amazon basin with displacements estimated from GRACE geopotential fields. We show that despite poor correlation (r(2) = 0.15) between rate estimates in GPS and GRACE-derived displacement time series, further analyses reveal low bias between annual amplitude estimates and a scaling near 1. There is higher correlation (r(2) = 0.78) between annual periodic motions, with near 1 to 1 agreement, but there is poor correlation (r(2) = 0.02) and little agreement between semiannual amplitude estimates. Subtracting GRACE displacements from the GPS time series flattens the GPS power spectra, reducing the spectral index magnitude, on average, from -1.2759 +/- 0.0007 ("fractional Brownian motion") to 0.3346 +/- 0.0006 ("fractional Gaussian noise"), suggesting that some fraction of the apparent GPS error correlation derives from mass loading signals that are not completely characterized by secular trends or seasonal periodic motions. From March 2011 to November 2016, we find a GPS and GRACE-derived displacement combined average uplift of the Amazon basin of 1.20 +/- 0.26 mm/yr and combined average annual periodic motion of 10.22 +/- 0.57 mm. Deviations from a standard trajectory model for site motion are apparent in both data sets and appear to coincide with various flooding and drought events between 2011 and 2016, which suggests that the GPS coordinate time series record displacements driven by large-scale climate oscillations.