Skip to main content

Complexity of Arctic Ocean water isotope (δ18O, δ2H) spatial and temporal patterns revealed with machine learning

Scientific Literature
Peer Reviewed Literature
Authors

Eric S. Klein, , Andrew P. Baltensperger, and Jeffrey M. Welker 

Date
December 4th, 2023
Abstract

The stable isotope compositions of water (d18O, d2H, deuterium-excess) are important tracers that help illuminate the changing Arctic water cycle and how Arctic-sourced water can influence lower latitudes. We present simultaneous boundary layer water vapor and ocean water isotope data that were measured continuously in the western Arctic Ocean. Sea surface water isotopes varied between the shallower continental Chukchi Shelf and the deeper Chukchi Borderlands to the north. The Chukchi Borderlands surface waters were less saline than the offshore Chukchi Shelf, as the Borderlands are influenced by greater sea ice cover and contribution of sea ice melt to surface freshwater. This greater contribution of sea ice melt resulted in lower deuterium-excess (d2H  8*d18O) in surface water values in the deeper Chukchi Borderlands than on the shallower Chukchi Shelf. Additionally, the sea ice melt contributions to freshwater were less prominent than river runoff, but freshwater from both sources decreased substantially below 70 m depth in the Chukchi Borderlands. Our observed water isotope values provided the foundation for producing water isotope maps (isoscapes) based on remote sensing and machine learning which incorporate parameters that can influence ocean circulation and thus water isotopes (e.g., salinity, sea surface temperature, water depth). These isoscapes suggest spatial complexity in the distribution of stable isotopes in Arctic waters, including sharp gradients in the distribution of the isotopes in seawater that were studied. These isoscapes can be improved in future iterations, for example, with the availability of more spatially continuous, remotely sensed oceanic variables or continuous ship-based measurements to use as additional predictors. As a result, the generation of these isoscapes could become a useful tool for understanding the past, present, and future Arctic water cycle in the context of the global hydrologic cycle.