Estimating Vertical Land Motion and Residual Altimeter Systematic Errors Using a Kalman‐Based Approach

Abstract

Vertical land motion (VLM) of the Earth's crust occurs over diverse spatial and temporal scales, driven by surface mass-loading, anthropogenic influences, and tectonic processes (e.g., Pfeffer et al., 2017). VLM provides the key connection between relative sea-level (RSL) derived from a tide gauge (TG) and absolute sea-level (ASL) derived from a satellite altimeter (ALT, e.g., Church & White, 2011). ASL data is mainly comprised of ALT-inferred time series with a near-global coverage, while RSL is typically observed along well-developed coasts (Neumann et al., 2015). In the areas with sparse long-term TG records, improved understanding of VLM is critical to help plan better sea-level rise adaptation pathways at local and regional scales (Cazenave et al., 2014;Church et al., 2013;Woodworth et al., 2011).VLM at TG locations is obtained from geodetic observations or, in many cases, inferred from models of glacial isostatic adjustment (GIA). VLM is directly observable using space-borne techniques such as GNSS (of which global positioning system (GPS) is the most frequently used, for example, Hammond et al., 2016;Santamaría-Gómez et al., 2011), and TG VLMs are often interpolated from GPS velocities (e.g., Bouin & Wöppelmann, 2010). In the absence of nearby GPS, GIA rates are alternatively applied (e.g., Hay et al., 2015;Spada & Galassi, 2012), with an often-untested assumption that GIA is the dominant contributor to VLM.

Vertical deformation and residual altimeter systematic errors around continental Australia inferred from a Kalman-based approach

“…Aim: Optimize a posteriori estimates of nonlinearity in VLM and altimeter systematic error by Rezvani et al (2021) to simultaneously estimate the underlying parameters through recursive forward and backward solutions of the refined multi-stage approach as summarized in Fig. 2.…”

“…We removed residual annual and semi-annual terms using the harmonic analysis of all ATG observations relating to each CP, while the altimeter-specific intercepts were estimated simultaneously. For computational efficiency, we followed thresholds suggested by Rezvani et al (2021) to limit our ATG combinations to a maximum of 8 mission-specific ATG series per TG. For later analysis, the candidate CPs were flagged as being 'on' or 'off' the continental shelf based on their relative location to the 200 m depth contour.…”

“…We then constructed measurements of tandem/dual ALT crossover (designated as AXO), through differencing the ASL data of the two overflying missions at the respective CPs that were located within 500 km of the coastal gauges. For computational efficiency, we reduced the numbers of tandem and dual AXO CPs using the same criteria proposed by Rezvani et al (2021) as well as applying the additional root mean squared error (RMSE) and distance thresholds given in Table S2. We adopted the sampling times of AXO series as the average times of ASL measurements from each set of overflying missions.…”

“…The potential nonlinearity in ALT systematic error and/or VLM parameters is also investigated. Figure adapted from Rezvani et al (2021) These methods have also been affected by short-scale oceanic processes that act differently at coastal TG and offshore ALT measurement locations and thus potentially bias results.…”

“…We address these questions by advancing the space-time Kalman filtering and smoothing method set out by Rezvani et al (2021) who applied an early version of the framework to estimate linear-only signals of VLM and regional systematic error in the Baltic Sea region using the reference-mission altimetry records. Here, we consider multi-mission alongtrack datasets and simultaneously estimate the time-fixed and time-variable components of location-specific VLMs and residual altimeter-specific systematic errors around Australia since the early 1990s.…”