A new relative sea level database for Norway tested against glacial isostatic adjustment models with an ensemble of physics-based history-matched Eurasian Ice Sheet chronologies.

We present a new relative sea level (RSL) database for Norway and for modelling studies. The total database contains 1011 data points, of which 558 (55%) are index points and 453 (45%) limiting dates. The new RSL database differs from earlier efforts in two key ways. Firstly by having fewer limiting dates as we have removed redundant data. Secondly, it contains new RSL data collected over 2018-2024 which are largely index point data.

The new RSL database is compared to 9,900 ice-Earth combinations from a 1-D glacial isostatic adjustment (GIA) model. From these combinations, the ice models tested come from a high-variance subset of 10 Eurasian Ice Sheet chronologies. These (GLAC3) chronologies are from a last glacial cycle history matching of the physics-based Glacial Systems Model against a diverse set of constraints. The 10 Eurasian ice chronologies are combined with 3 different reconstructions of global ice changes (i.e., a total of 30 ice models). 

We show how data-model fits vary for the ice chronologies and Earth model parameters explored. Results indicate relatively weak upper mantle viscosities for Norway. While some ice-Earth model combinations can reproduce the general RSL trends and show features of the Younger Dryas and Tapes transgressions, no model parameter sets provide quality fits to all the data or can follow all the observed RSL fluctuations. This suggests inaccuracies in the model and/or the need to explore a larger parameter space.

RSL uncertainties are calculated using a nominal Bayesian approach and capture ~80% of the Norwegian RSL data. By splitting the data into 3 subregions, we show how data-model fits vary geographically and which ice-Earth model combinations are preferred where. This reveals that data-model fits are poorest in South Norway, where only 40% of the RSL are captured (and only 22% of the index point data). We hypothesise that the poor fits in this region are due to inaccuracy in the regional and/or background (global) ice models considered.