Gaussian mixture modeling describes the geography of the surface ocean carbon budget.

Jones, Dan ORCID: https://orcid.org/0000-0002-8701-4506; Ito, Takamitsu. 2019 Gaussian mixture modeling describes the geography of the surface ocean carbon budget. In: Proceedings of the 9th International Workshop on Climate Informatics: CI 2019, Paris, France, October 2-4, 2019. University Corporation for Atmospheric Research (UCAR), 108-113.

Before downloading, please read NORA policies.

Preview

Text (Open Access)
Copyright Author(s). Distributed under license by the University Corporation for Atmospheric Research (UCAR). This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
CI2019_paper_33.pdf - Published Version
Available under License Creative Commons Attribution 4.0.
Download (608kB) | Preview

Official URL: http://dx.doi.org/10.5065/y82j-f154

Abstract/Summary

Abstract—We use an unsupervised classification technique (i.e. Gaussian mixture modeling or GMM) to identify ocean regions with similar balances between processes that determine the surface budget of dissolved inorganic carbon. GMM objectively locates sub-populations in the distribution of carbon budget terms. We use a simple four-class description and find regimes that are broadly consistent with classical theoretical frameworks. Class 1 covers 24% of ocean surface area and corresponds to highly productive areas with strong vertical mixing, wind-driven open ocean upwelling, and absorption of atmospheric carbon dioxide. Class 2 covers 8% of ocean surface area and corresponds to regions of especially weak productivity. Class 3 covers 16% of ocean surface area and corresponds to wind-driven coastal and equatorial upwelling. Finally, class 4 covers the remaining 52% of ocean surface area and corresponds to the relatively unproductive subtropical gyres, which are typically characterized by downwelling and low surface nutrient concentrations. We argue that GMM may be a useful method for comparing biogeochemical regimes between climate models.