A new remote sensing-based approach to monitor ocean heat content (OHC) anomalies is proposed to overcome challenges with observing OHC over the entire ocean.
The output of an ocean state estimate - using the Estimating the Circulation & Climate of the Ocean (ECCO) framework - is assumed to be perfect observational data and used to identify prospective variables that could be calculated from remotely monitored characteristics of the ocean. The depth-integrated electrical conductivity (potentially derived from magnetometry) is shown to be highly predictive of OHC in poorly observed regions - such as those covered by sea ice - so it is used together with sea surface heights (from altimetry) and ocean bottom pressures (from gravimetry) to estimate OHC. The seafloor depth, sea surface height anomalies, ocean bottom pressure, and depth-integrated electrical conductivity explain virtually all of the variance in OHC.
To demonstrate the feasibility of a method that uses these ocean characteristics - inferable from global satellite coverage - to monitor OHC, the output of ECCO is sampled along historical hydrographic transects, a machine learning algorithm - called a Generalized Additive Model or GAM - is trained on these samples, and OHC is estimated everywhere. This remote monitoring method can estimate global OHC within 0.15% spatial root-mean-square error (RMSE) on a bi-decadal time scale. This RMSE is sensitive to the spatial variance in OHC that gets sampled by hydrographic transects, the variables included in the GAM, and their measurement errors when inferred from satellite data - in particular the noise levels of depth-integrated electrical conductivity and ocean bottom pressure. OHC could be remotely monitored over sufficiently long time scales when enough spatial variance in OHC is explained in the training data over those time scales. This method could potentially supplement existing methods to monitor OHC.