Uncertainty in observed coastal vertical ground motions matters for local sea-level projections

Abstract

<p>While the understanding and modelling of RSLC due to ocean density and mass changes have greatly improved over the past few decades, RSLC contributions due to vertical ground motions (VGMs) remain a major source of uncertainty, in particular for local scales. Here, VGMs relate to ground motions that have imprints of a few kilometres, as opposed to broad scale land motion such as Glacial Isostatic Adjustment (GIA). VGMs can be caused by processes such as groundwater extraction or the load of infrastructure on recent sediment deposits, all of which vary in space and time, and are difficult to project in the future.</p><p>A common technique to account for the VGM contribution in RSLC projections is to use GNSS stations to identify present days rates, identify causes and extrapolate rates where appropriate. However, previous works that followed this approach generally considered a unique GNSS solution (e.g., only NGL) together with a unique trend estimate algorithm, while often several GNSS solution based on different trend estimate techniques are available (i.e., ULR, NGL, JPL and GFZ). Ignoring these alternative estimates and their potential discrepancies may therefore ignore further sources of uncertainty.</p><p>Here, we analyse how VGMs derived from 4 GNSS records (ULR, NGL, JPL and GFZ) and trend estimates affect uncertainty in future RSLC at tide gauges. Residual local VGM trend distributions are obtained from a Monte Carlo procedure by removing the effect of GIA and solid earth deformation induced by contemporary mass redistribution. This procedure is repeated using three different trend estimate techniques, all based on the MIDAS robust trend estimator, where GNSS timeseries offsets are (i) automatically detected, (ii) initially declared or (iii) manually corrected. We find that the choices of GNSS solution or the trend estimate technique can both substantially affect estimates of the VGMs leading to differences of about 1mm/year, which is approximately one third of sea-level rise rates observed today. Accounting for the uncertainty and discrepancy in GNSS records can change AR6 RSLC median projections by more than -/+ 15% in 2100. This is larger than the differences between SROCC and AR6. Likely ranges and skewness of projected RSLC distributions are also affected. We suggest that sea-level scenarios, whether median, likely, probabilistic or high-end, can benefit substantially from advances in observing and projecting local vertical ground motions.</p>