Ocean Heat Transport Convergence Drives Regional Energy Imbalance in the Sunlit Ocean Layer

Abstract. The sunlit ocean layer (SOL, 0–200 m) is where ocean heat transport (OHT) interacts directly with radiative forcing, where temperature changes feed back on the atmosphere, and where impacts of global warming on marine life are concentrated. Below 200 m, except for geothermal heating in the abyss, energy imbalance is entirely governed by OHT processes. For the SOL, we introduce the concept of SOL energy imbalance (SOL-EI), defined as a closed heat budget that separates air-sea heat uptake from OHT convergence on a 10-year timescale. SOL-EI is estimated globally and regionally over 1980–2022 with observational uncertainty quantification. Results reveal that recent changes in SOL warming patterns are generally controlled by OHT convergence rather than local air-sea heat flux, which instead tends to act as a regional feedback mechanism. Global SOL-EI accumulation is estimated as the convergence between global air-sea heat uptake and heat export through 200 m depth. This global index reveals interdecadal variability, with near-zero values linked to volcanic and internal climate forcing, and unabated accumulation of 30–60 ZJ/decade over the past decade. A leading interdecadal oscillation pattern (IOP) is furthermore identified directly from SOL-EI maps, and then decomposed into regional OHT convergence and air-sea flux terms. Here again results show that OHT convergence acts mainly as the driving process, and air-sea flux as a regional feedback. Because our framework defines SOL-EI as a time-evolving equation for 10-year trends, it provides a unified lens for interpreting global change, interdecadal variability, and regional patterns in terms of known processes. In addition, the mapping of Argo data within OCCA2 operates on residuals computed relative to a time-evolving state estimate that provides closed heat budgets. This methodological innovation enhances interpretability as compared to other gridded Argo datasets. Uncertainty quantification within the OCCA2 family of estimates is readily extensible to other Argo mapping methods and to the pre-Argo period, via alternative computations of the residual term, as demonstrated here using the IAPV4 gridded data product.