Dissolving better: what can Earth System models learn from 60 years of in situ carbonate mineral dissolution measurements

Carbonate mineral dissolution has the potential to neutralise anthropogenic CO2 and act as a buffer against ocean acidification. To accurately quantify these effects and predict how this process will respond to a changing climate, we need to know what its drivers are. Still, our current understanding of dissolution is incomplete: for instance, excess alkalinity production in the mesopelagic suggests that carbonates dissolve where seawater is thought to be supersaturated with respect to calcite. In situ measurements of the dissolution rate can help to determine which other environmental factors - apart from the saturation state - drive dissolution. However, those measurements are scarce and even though they measure the same phenomenon, the resulting rates differ by up to two magnitudes between the available studies. Additionally, the dissolution patterns with depth are also not consistent with each other. Possible explanations for these variances include differences in methodologies and sample types, or the respective physical and chemical environments. This work aims to disentangle those factors and determine the real qualitative and quantitative value of the existing datasets. This is achieved through review of the literature and the training and interpretation of a supervised regression model (XGBoost), exploring the blind spots in our current conception of carbonate mineral dissolution. Based on these results, changes to the implementation of calcite and aragonite dissolution in Earth System Models are recommended.