Statistics for Mapping Ocean Heat Content with Argo Floats: Modeling and Uncertainty Quantification

Argo floats measure seawater temperature and salinity in the upper 2000 meters of the ocean. These floats are uniquely capable of measuring the heat content of the global ocean, a quantity that is of central importance for understanding changes in the Earth's climate system. But providing detailed spatio-temporal estimates of the heat content is statistically challenging due to the complex structure and large size of the Argo dataset. We have previously demonstrated (Kuusela and Stein, 2018) that locally stationary Gaussian process regression leads to more accurate and computationally efficient interpolation of Argo data. Here we build upon those findings to produce improved Argo-based global ocean heat content estimates. We also study the sensitivity of these estimates to the underlying statistical modeling assumptions and present results indicating that the magnitude of the overall warming trend may depend on the detailed modeling of the climatological time trend in the mean field estimate. We also demonstrate the benefits of including time in the mapping and propose an uncertainty quantification method based on local conditional simulations that yields plausible uncertainties for the ocean heat content anomalies with appropriate spatial and temporal correlations.