Machine Learning Solutions to Regional Surface Ocean δ18O‐Salinity Relationships for Paleoclimatic Reconstruction.

Stable isotope‐based reconstructions of past ocean salinity and hydroclimate depend on accurate, regionally constrained relationships between the stable oxygen isotopic composition of seawater (δ18Osw) and salinity in the surface ocean. An increasing number of δ18Osw observations suggest greater spatial variability in this relationship than previously considered, highlighting the need to reassess these relationships on a global scale. Here, we use available, paired δ18Osw and salinity data (N = 11,119) to create global interpolations of each variable. We then use a self‐organizing map, a specialized form of machine learning, to define 19 regions with unique δ18Osw‐salinity relationships in the surface (<50 m) ocean. Inclusion of atmospheric moisture‐related variables and oceanic tracer data in additional self‐organizing map experiments indicates global surface δ18Osw‐salinity spatial patterns are strongly forced by the atmosphere, as the SOM spatial output is highly similar despite no overlapping input data. Our approach is a useful update to the previously delimited regions, and highlights the utility of neural network pattern extraction in spatiotemporally sparse data sets. Plain Language Summary: Our understanding of past changes in the ocean and atmosphere often comes from information stored in biological marine carbonate archives. Stable oxygen isotope measurements from such archives record past ocean temperature as well as the stable oxygen isotopic composition of seawater. In the surface ocean, stable oxygen isotope values have a strong, linear, relationship with salinity, thought to be the result of evaporation and precipitation affecting stable isotope ratios and salinity in a similar manner. This permits reconstruction of past ocean salinity from marine carbonates, which may be used to infer large‐scale hydroclimate variability, in some cases. However, this manner of reconstruction relies on the linear isotope‐salinity relationship, which remains poorly constrained and is influenced by a combination of oceanic and atmospheric processes. In this work, we define regional isotope‐salinity relationships and broadly assess whether ocean or atmospheric processes play a larger role in the spatial distribution of these regions. We identify new regions compared to previous work, and find that over most of the ocean, precipitation, precipitation isotope values, and evaporation are important in setting seawater isotope‐salinity patterns. Key Points: Mixed‐layer regions with distinct salinity‐seawater δ18O paired values are identified by self‐organizing mapInclusion of atmospheric or oceanic tracers in the self‐organizing map shifts the spatial distribution of salinity‐seawater δ18O patternsRegion identification is still dependent on available observational data, but the throughput offered here enables updating as data increase [ABSTRACT FROM AUTHOR]