Data-based sea-air CO2 flux estimation -The Surface Ocean pCO2 Mapping intercomparison (SOCOM)

Abstract

Measurements of surface-ocean carbon content, such as the pCO2 observations collected in the Surface Ocean CO2 Atlas (SOCAT), contain spatially and temporally explicit information about the sea-air CO2 exchange. In order to map the point data into continuous spatio-temporal fields, a variety of methods has been developed by different groups. As the chosen approaches span a wide range of complexities from statistical interpolation, regression, neural networks, and data assimilation into process models, they comprise interesting complementarity, for example in the different weights given to individual input data streams, or in whether they use model-based or generic relationships between carbon and driving variables. The SOCOM initiative aims (1) to exploit this complementarity to learn more about the ocean biogeochemical signals and how they can be retrieved from different information sources, and (2) ultimately to provide a best guess CO2 flux estimate from pCO2, DIC, or other data, including an assessment of its robustness/limits. Thecontribution will present the ensemble in terms of their spatial patterns, biome-specific temporal variations, and comparison to independent data. Robust features will be identified.ta into continuous spatio-temporal fields, a variety of methods has been developed by different groups. As the chosen approaches span a wide range of complexities from statistical interpolation, regression, neural networks, and data assimilation into process models, they comprise interesting complementarity, for example in the different weights given to individual input data streams, or in whether they use model-based or generic relationships between carbon and driving variables. The SOCOM initiative aims (1) to exploit this complementarity to learn more about the ocean biogeochemical signals and how they can be retrieved from different information sources, and (2) ultimately to provide a best guess CO2 flux estimate from pCO2, DIC, or other data, including an assessment of its robustness/limits. Thecontribution will present the ensemble in terms of their spatial patterns, biome-specific temporal variations, and comparison to independent data. Robust features will be identified.