Constraining local ocean dynamic sea level projections using observations

Abstract. The redistribution of ocean water volume under ocean-atmosphere dynamical processes results in sea level changes. This process, called Ocean Dynamic Sea Level (ODSL) change, is expected to be one of the main contributors to sea level rise along the western European coast in the coming decades. State-of-the-art climate model ensembles are used to make 21^st century projections for this process, but there is a large model spread. Here, we use the Netherlands as a case study and show that ODSL rate of change for the period 1993–2021 correlates significantly with ODSL anomaly at the end of the century and can therefore be used to constrain projections. Given the difficulty to estimate ODSL changes from observations on the continental shelf, we use three different methods providing seven observational estimates. Despite the broad range of observational estimates, we find that half of CMIP6 models have rates above and one below the observational range. We consider the results of those models as implausible and compare projections of ODSL with all models and the plausible selection. The difference is largest for the low emission scenario SSP1-2.6 for which the median and 83^rd percentiles are reduced by about 25 % when the plausible selection is used. This method results in reduced uncertainty in sea-level projections. Additionally, having projections that are compatible with the observational record increases trust in their century-scale accuracy. We argue that this model selection is better than using all models to provide sea level projections suited to local users in the Netherlands and that the same method can be used elsewhere.