Estimating Air‐Sea Carbon Flux Uncertainty Over the Tropical Pacific: Importance of Winds and Wind Analysis Uncertainty

Abstract

AbstractThe tropical Pacific is a major natural source of CO2 to the atmosphere and contributor to global air‐sea carbon flux variability. High time‐resolution wind and CO2 measurements from equatorial Pacific moorings reveal the primary factor controlling mooring‐observed flux variability to be near‐surface wind variability, above CO2 variability, in this region over the last 10 years. The analysis product winds used most widely in previous calculations of basin‐scale carbon flux are compared with mooring winds and found to exhibit significant differences in mean, variability, and trend. Earth system model calculations are in basic agreement with the mooring results and used to estimate effects of wind uncertainty on our knowledge of regional air‐sea carbon exchange. Results show that NCEP1 and NCEP2 winds contain biases large enough to obscure the interannual variability of CO2 flux (RMSE ≈ σ) and cause spurious 25‐year (1992–2016) trend components in equatorial Pacific carbon flux of 0.038–0.039 and 0.016–0.021 Pg C yr−1 per decade, respectively. These spurious trends act to reduce by up to 50% the 25‐year trend in equatorial Pacific carbon flux simulated by the Earth system model under increasing atmospheric CO2 concentration. The Cross‐Calibrated‐Multi‐Platform wind product tracks observed variability of equatorial Pacific wind better (interannual RMSE ≈ 0.4σ) than the NCEP reanalyses when site sampled at mooring locations yet still causes a spurious regional trend (0.03 Pg C yr−1 per decade) that masks 40% of the simulated 25‐year trend in carbon flux. The mooring observations are fundamental to identifying the limitations of current wind products to characterizing long‐term trends and understanding air‐sea carbon exchange.