Abstract's details

A Kalman-based approach to simultaneously estimate vertical land motion and altimeter-specific systematic errors using altimeter, tide gauge, and GPS measurements

Mohammad-Hadi Rezvani (School of Geography, Planning, and Spatial Sciences, University of Tasmania, Australia)


Christopher Watson (School of Geography, Planning, and Spatial Sciences, University of Tasmania, Australia); Matt King (School of Geography, Planning, and Spatial Sciences, University of Tasmania, Australia)

Event: 2022 Ocean Surface Topography Science Team Meeting

Session: Regional and Global CAL/VAL for Assembling a Climate Data Record

Presentation type: Type Forum only

Contribution: PDF file


Many studies have sought to combine altimeter (absolute sea level, ASL) and tide gauge (relative sea level, RSL) data to infer systematic errors in altimetry, or vertical land motion (VLM). Regionally correlated errors have not been rigorously addressed, and most studies assume VLM is linear and inferred reliably from non-co-located GPS sites. We develop a Kalman filtering and smoothing approach to simultaneously quantify ongoing variability in VLM as well as regional altimeter-specific systematic errors. We assimilated data from multi-mission altimeters, long-running TGs, and permanent GPS sites, while considering the space-time covariances and time-correlated errors within observational series of tandem and dual crossovers, altimeter minus TG, and GPS heights. We investigated the approach using case study regions including the Baltic Sea, Australia, and South America. We quantified localized variability of VLM trends at TGs, typically ~1.1 mm/yr (but up to ~4.5 mm/yr), that otherwise cannot be inferred from spatially interpolated GPS velocities or predicted GIA rates. We detected small but significant systematic errors in regional altimetry data within a typical range of ~0.5-2.5 mm/yr and evaluated the possibility of their time-variability reaching up to ~3.0 mm/yr over 3.5 years during the mission-specific lifespan. These narrowed the deviation between ASL estimates from TG and ALT records, and reduced the RMSE of their geographical variability by up to ~40%. We further investigated the capabilities of the method to derive time-variable VLM as a result of co- and post-seismic deformation, and ice-mass loss in the Antarctic Peninsula. We discuss the key limitations that arise primarily due to differential oceanographic signals between ALT and TG locations in the presence of complex coastal processes. This approach advances the ALT minus TG technique to estimate localized VLM at TG locations, while simultaneously estimating altimeter-specific systematic errors in a regional context, with potential applicability to other global- and local-scale studies.
Mohammad-Hadi Rezvani
School of Geography, Planning, and Spatial Sciences, University of Tasmania