Ocean acidification, driven by anthropogenic carbon emissions, significantly impacts marine life, particularly in vulnerable Arctic regions. This study employs unsupervised machine learning to analyze Arctic surface acidification simulations from two advanced climate models. Four sub-regions are identified, their boundaries influenced by current and projected sea ice patterns, consistent across models and emission scenarios. The central Arctic shows stronger trends towards corrosive waters due to early summer warming and late summer freshening. Salinity and total alkalinity reductions are key drivers of Arctic pH changes, highlighting the importance of sub-regional analysis of surface water properties.

John P. Krasting, Maurizia De Palma, Maike Sonnewald, John P. Dunne, Jasmin G. John