AUTHOR=Martellucci Riccardo , Dentico Carlotta , Coppola Laurent , Skjelvan Ingunn , Giani Michele , Pensieri Sara , Cantoni Carolina , Cardin Vanessa , Fourrier Marine , Bozzano Roberto , Paulsen Melf , Mauri Elena 

TITLE=Air-sea CO2 exchange in the Eastern Atlantic and the Mediterranean Sea based on autonomous surface measurements

JOURNAL=Frontiers in Marine Science

VOLUME=Volume 12 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/marine-science/articles/10.3389/fmars.2025.1633617

DOI=10.3389/fmars.2025.1633617

ISSN=2296-7745

ABSTRACT=IntroductionThe ATL2MED mission, conducted between October 2019 and July 2020, investigated the variability of air–sea CO2 exchange in the Eastern Atlantic and the Mediterranean Sea. The main objectives were to assess the spatial and temporal variability of the seawater partial pressure of CO2 (pCO2sw), identify its controlling physical and biogeochemical processes, estimate the CO2 fluxes across the sea–air interface, and evaluate the performance of neural network-based predictions (CANYON-MED) in contrasting oceanographic regions.MethodsHigh-resolution autonomous measurements were collected using Saildrone Unmanned Surface Vehicles (USVs), complemented by fixed ocean stations, gliders, and research vessels. Data quality was ensured through cross-validation among platforms, despite challenges such as sensor drift caused by biofouling.ResultsThe results reveal pronounced regional differences in pCO2sw and CO2 fluxes, driven by thermal effects, biological activity, and physical mixing. In the Eastern Atlantic, upwelling systems off northwest Africa induced strong outgassing, while the western Mediterranean acted as a CO2 sink during the spring bloom. Conversely, the Adriatic Sea exhibited episodic CO2 outgassing, particularly in its southern and northern basins, associated with thermal stratification, river plumes, and coastal upwelling.DiscussionSaildrone measurements successfully resolved sub-mesoscale processes typically missed by fixed platforms, demonstrating their potential to enhance ocean CO2 observations in under-sampled or logistically constrained regions. The good agreement with neural network-based estimates further supports the robustness of the dataset. Overall, these findings emphasize the value of high-resolution, multi-platform approaches for accurately quantifying CO2 fluxes and improving predictive capabilities in a changing ocean.