A Synthetic Ensemble for Air-Sea CO2 Flux Observations

Holly Olivarez, University of Colorado Boulder, Environmental Studies Program, Boulder, CO, United States, Nicole S Lovenduski, University of Colorado, Department of Atmospheric and Oceanic Sciences, Boulder, CO, United States, Riley Xavier Brady, University of Colorado at Boulder, Department of Atmospheric and Oceanic Sciences, Boulder, CO, United States, Karen A McKinnon, University of California Los Angeles, Departments of Statistics, Institute of the Environment and Sustainability, Los Angeles, CA, United States, Peter Landschuetzer, Max Planck Institute for Meteorology, Hamburg, Germany and Galen A McKinley, Lamont-Doherty Earth Observatory of Columbia University, Palisades, NY, United States
Variations in the fluxes of carbon dioxide (CO2) between the ocean and the atmosphere are an important contributor to the atmospheric CO2 growth rate, the evolving global carbon budget, and the climate system. It is important to both quantify and understand the causes of these air-sea CO2 flux variations, so as to make better near-term predictions and long-term projections of the future climate system. The advent of intelligent gap-filling methods for surface ocean pCO2 has produced a number of observation-based estimates of air-sea CO2 flux at monthly, 1o x 1o resolution from as early as 1982 to the present day. These observation-based, gap-filled estimates suggest substantial past variations in the air-sea CO2 flux on interannual to decadal timescales. The pronounced decadal variability in the Southern Ocean carbon sink has gained particular attention. Here, we explore whether the observation-based variability in air-sea CO2 flux is of internal (i.e., associated with climate modes such as the El Niño Southern Oscillation (ENSO) or Southern Annular Mode (SAM)) or external (e.g., driven by anthropogenic/volcanic emissions) origin.

The external and internal components of variability in air-sea CO2 flux can be separated within large initial condition ensembles of Earth system model simulations. In these ensembles, the ensemble mean describes the externally forced variability, while the departures from the ensemble mean represent the internal component of variability. But how can one tease apart the contributions of internal and external processes to variability in real-world observations, for which we have only one realization or "ensemble member?"

Here, we statistically resample the observational record in order to generate a synthetic ensemble of air-sea CO2 flux. In the resulting observational large ensemble, each ensemble member has statistical properties that are similar to the observational record, but with a unique sequence of internal variability and an identical externally forced signal. We use this observational large ensemble to explore the origin of variability in the observation-based air-sea CO2 flux.