Effect of seasonal change in gas transfer coefficient on air–sea CO₂ flux in the western North Pacific

Abstract

We used an eddy-permitting three-dimensional ocean ecosystem model and applied it in the western North Pacific to understand the seasonal variations and horizontal distributions of the air–sea CO₂ flux and difference in the partial pressure between sea water and the atmosphere (∆pCO₂). The high-resolution model reproduced the observed zonal belt of strong CO₂ uptake in the mid-latitude (30–45°N) western North Pacific including the Kuroshio extension and mixed water regions, which was difficult to show in previous coarse-resolution models. The East Asian winter monsoon, an important phenomenon in the western North Pacific, affects the seasonal CO₂ air–sea gas exchange with a high (low) gas transfer coefficient in winter (summer). In the subtropical region, ∆pCO₂ is negative in winter and positive in summer as a result of the temperature effect. Combination of seasonal change in gas transfer coefficient with ∆pCO₂ suppresses CO₂ release in the subtropical region, and vice versa in the subarctic region (i.e., suppresses CO₂ uptake). That is, the East Asian winter monsoon in the western North Pacific contributes to the reduction of the annual CO₂ flux through the seasonal change in the gas transfer coefficient, leading to an overall annual CO₂ uptake in the subtropical region and CO₂ release in the subarctic region.

Appendices

Appendix 1: a sensitivity experiment by using JRA 25 wind data

We obtained results by using wind data from the Japanese 25-year Reanalysis (JRA-25) (http://jra.kishou.go.jp/JRA-25/index_en.html) instead of MIROC wind data. The JRA-25 is compiled by the Japan Meteorological Agency (JMA) and the Central Research Institute of Electric Power Industry (CRIEPI). The 6-hourly surface (10 m) zonal and meridional wind data from 1980 to 1989 are taken to the calculate CO₂ gas transfer coefficient. Wind data from the JRA-25 show a similar tendency of daily change with MIROC wind (Fig. 9). Two wind products are also similar to the probability statistical distribution of the number of occurrences. The frequency of high wind speed (e.g., greater than 10 m/s) in the MIROC wind is slightly greater than that in the JRA-25 wind both at the Kuroshio extension site and at the subarctic site, whereas it is less than in the JRA-25 at the subtropical site (Fig. 10).

Fig. 9

Comparison of absolute wind speeds of model wind (solid line) with JRA-25 wind (dashed line) by using 1-year data at the a subtropical site, b Kuroshio extension site, and c the subarctic site

Fig. 10

Frequency distributions of 1 m/s interval of absolute wind speed at the a subtropical site, b Kuroshio extension site, and c subarctic site by using daily data over 10 years

Although the CO₂ gas transfer coefficient calculated by using JRA-25 wind is slightly lower than that calculated by using MIROC wind, the results of the annually averaged ∆pCO₂ and air–sea CO₂ flux calculated by using JRA-25 wind are similar to the results calculated by using MIROC wind. Maxima of the CO₂ gas transfer coefficients calculated by using JRA-25 wind are about 4.0 × 10⁻⁹ mol m⁻² s⁻¹ μatm⁻¹ in monthly average in February and 2.75 × 10⁻⁹ mol m⁻² s⁻¹ μatm⁻¹ in annual average, respectively (Fig. 11), while there are areas in which CO₂ gas transfer coefficients are greater than 4.5 × 10⁻⁹ mol m⁻² s⁻¹ μatm⁻¹ in February and 3.0 × 10⁻⁹ mol m⁻² s⁻¹ μatm⁻¹ in annual average calculated by using MIROC wind (Figs. 4c, 5c). Annually averaged ∆pCO₂ and air–sea CO₂ flux calculated by using JRA-25 winds show similar distributions to standard experimental results except that positive areas of annually averaged ∆pCO₂ are contracted compared with the standard experimental results (Figs. 5, 12). Both results calculated by using JRA-25 wind and MIROC wind show that $\overline{{\varepsilon^{'} \Delta p{\text{CO}}_{2}^{'} }}$ are negative in the subtropical region and positive in the arctic region (Figs. 12d, 6b).

Fig. 11

Distributions of the monthly averaged CO₂ gas transfer coefficients in a February and b August. c Distribution of annually averaged CO₂ gas transfer coefficient

Fig. 12

Annually averaged a ∆pCO₂, b air–sea CO₂ flux, c is the same as b except for shaded areas showing the areas in which signs of the annually averaged ∆pCO₂ and the air–sea CO₂ flux are different from each other. d Difference between results of $\overline{F}$ and $\overline{{F^{*} }}$, i.e., $\overline{{\varepsilon^{'} \Delta p{\text{CO}}_{2}^{'} }}$, which is affected by combination of seasonal changes in ε with ∆pCO₂. e Difference between monthly averaged $\overline{{F_{\text{monthly}} }}$ and $\overline{{F_{\text{monthly}}^{*} }}$ in February, and f in August

Appendix 2: a sensitivity experiment by using a coefficient of 0.26

The gas transfer piston velocity κ in Eq. (3) in this study is 0.31, based on Wanninkhof et al. (1992). Sweeney et al. (2007), Takahashi et al. (2009), and Wanninkhof et al. (2013) suggested using the coefficient of 0.27, 0.26, and 0.251 to calculate the gas transfer piston velocity, respectively. The coefficient of 0.26 was used to estimate the air–sea CO₂ flux and the results are shown in Fig. 13.

Fig. 13

a–f Are the same as Fig. 12. a–f except for calculating by using model wind data and using gas transfer piston velocity κ calculated with a coefficient of 0.26

The distributions of the air–sea CO₂ flux are similar to the results calculated by using 0.31 (Figs. 5b, 13b; Figs. 6b, 13d). The western North Pacific is a strong sink for atmospheric CO₂ except for the areas off the Kuril Islands and south Bering Sea. A belt of strong CO₂ sink zone is located in mid-latitude (30–45°N), and the greatest absolute value of the air–sea CO₂ flux is more than −3.2 mol m⁻² year⁻¹.