Estimation of autumn oxygen emissions from the Pacific Ocean from atmospheric observations

In recent years, development of precise methods for measuring oxygen (O₂) concentrations have allowed us to detect change in the atmospheric O₂ concentrations. Global atmospheric O₂ observation programs have revealed that the spatiotemporal variations in O₂ concentrations can provide information on various processes on the Earth’s surface including the global carbon cycles. One of such information that the atmospheric O₂ observations can provide us is the gas exchange between the atmosphere and the ocean. In spring and summer, O₂ produced during photosynthesis (primary production) by phytoplankton at the ocean surface is released into the atmosphere. In autumn and winter, the surface cooling enhances the vertical ocean mixing, which brings the deeper water with depleted O₂ to the surface, resulting in the absorption of atmospheric O₂ into the ocean. The seasonal change in the sea surface temperature, causing the O₂ solubility change, enhances the above seasonal change in the ocean O₂ flux. The seasonal variation in atmospheric O₂ concentrations reflects the changes in the O₂ flux from the terrestrial biosphere and the above-mentioned O₂ flux from the ocean, and shows seasonal variations with a maximum in summer and a minimum in winter. Additionally, it has been pointed out that O₂ may be released from the ocean in autumn. There are two tentative mechanisms for such autumn O₂ emissions. One mechanism is outgassing from ocean subsurface layer, rich in dissolved O₂, associated with the increasing mixed layer depth. The other one is O₂ outgassing from phytoplankton bloom in autumn. The possibility of O₂ release in autumn due to these mechanisms has long been pointed out, but it has not been clear which ocean areas and to what extent it actually occurs.

A research team led by Yasunori Tohjima of the Earth System Division of the National Institute for Environmental Studies and Kazuhiro Tsuboi of the Meteorological Research Institute of the Japan Meteorological Agency (hereinafter referred to as "our research team") has been conducting wide-area observations of atmospheric O₂ concentrations in the Asia-Pacific region. Therefore, in this study, we aimed to clarify the spatial distribution of seasonal variations in atmospheric O₂ concentrations in the Pacific Ocean and to explore the possibility of O₂ emission from the ocean in autumn through a detailed comparison of the atmospheric O₂ seasonal cycles between the observation and the model simulation based on previously reported ocean O₂ fluxes and an atmospheric transport model.

Our research team developed an original method to precisely measure the O₂ concentration in air samples using a gas chromatograph and a thermal conductivity detector, and has been observing atmospheric O₂ concentrations since the late 1990s^{Note 1}. In addition, a non-dispersive infrared analyzer (NDIR) was used to determine the atmospheric carbon dioxide (CO₂) concentration. In this study, to conduct an analysis focusing on the air-sea gas exchange, we used atmospheric potential oxygen (APO), which is defined as the sum of atmospheric O₂ and CO₂ ^{Note 2}. APO, being defined to be unchanged by the exchange of O₂ and CO₂ in the terrestrial biosphere, is a tracer that mainly reflects the O₂ exchange between the atmosphere and the ocean.

In order to clarify the spatiotemporal variations of atmospheric O₂ concentrations over a wide area of the Asia-Pacific region, our research team conducted flask sampling of air samples using cargo ships operating between Japan and North America, Oceania, and Southeast Asia, as well as ground stations at Hateruma Island, Ochiishi Cape, and Minamitorishima. Figure 1 shows the distribution of flask sampling points. Since sampling was conducted on the route, air samples were collected in the Western Pacific and North Pacific. In this study, to obtain detailed spatial distribution of the average seasonal variation of APO, the data from the cargo ships was grouped into 10º latitude × 10º longitude bins, and the spatial distribution of the average seasonal variation in the western Pacific and North Pacific was obtained. The 41 pink-shaded bins in Figure 1 are the locations where the seasonal variation could actually be obtained.

In this study, APO was simulated using the ocean O₂ and CO₂ fluxes^{Note 3} and an atmospheric transport model (NIES-TM), and the simulated seasonal variation was compared in detail with the observed ones. The climatological monthly ocean O₂ fluxes used for the simulation were taken from an empirical product that relates O₂ flux to ocean heat flux (Garcia & Keeling, 2001; hereafter “GK2001 O₂ fluxes”). Previous studies have shown that the GK2001 O₂ fluxes reproduce the seasonal variation of APO well, but the autumn flux has not been explicitly considered.

A function expressed as the sum of a quadratic equation corresponding to the long-term trend and two harmonics with periods of 1 year and 1/2 year was fitted to the observed and simulated APO by using a least squares method. The average seasonal cycle was obtained as the sum of the two harmonics.

Figure 2 shows typical examples of APO seasonal cycle derived from air samples collected onboard cargo ships in the two 10º x 10º bins. The simulated seasonal APO cycles based on the GK2001 flux (orange line) reproduces the observed seasonal cycles (blue line) well. However, detailed comparison shows some differences in amplitude and phase. Furthermore, focusing on the decrease in APO from summer to winter, we found that the observed results tend to decrease more slowly than the simulation results in the mid- and high-latitude zones of the Northern Hemisphere.

The differences in amplitude and phase may reflect those of the GK2001 O₂ fluxes used in the simulation. Since the primary objective of this study is to investigate the difference in the seasonal pattern between the observation and simulation, differences in the amplitude and phase of the seasonal cycle must be minimized. Therefore, the simulated seasonal cycle is scaled by a single factor and shifted by the phase difference to match the observations. The results are shown in the light blue dashed lines in Figure 2, which reproduce better the observed seasonal cycles. The red curve shows the difference (observation – adjusted simulation) between the observation (red curve) and the adjusted simulation (light blue dashed line). The difference shows almost zero except during autumn-winter period, when a peak-like increase (apparent autumn peak) is observed. These results suggest the existence of autumn O₂ flux that is not expressed in the GK2001 flux.

Fig. 3 shows the spatial distribution of the amplitude of the apparent autumn peaks. There are prominent enhancements of the apparent autumn peaks except the equatorial region (20S-20N), especially north of 40N. A time series of vertical profiles of dissolved O₂ concentrations in the Pacific Ocean at 20-40ºS and 20-40ºN showed that a subsurface shallow oxygen maximum appeared at a depth of about 60m in summer and disappeared in autumn and winter. Meanwhile, a study based on satellite observations of chlorophyll a concentration (an index of phytoplankton abundance) suggested that the 40-60ºN in the Pacific Ocean is where autumn blooms occur. These facts suggested the possibility of O₂ emission from the subsurface oxygen maximum in the 20-40ºS and 20-40ºN latitudes, and from the autumn bloom in the 40-60ºN latitudes.

In this study, we were unable to quantitatively estimate the intensity and distribution of ocean O₂ fluxes in autumn. However, in the future, such quantitative estimation will be conducted by inverse analyses using an atmospheric transport model. Since ocean O₂ flux is closely related to marine biogeochemical processes and circulation, deepening our knowledge about air-sea O₂ exchange is expected to contribute to the analysis of variations in atmospheric O₂ concentration and the development of marine biogeochemical models.

1 Changes in the atmospheric O₂ concentration are expressed as relative deviations of the O₂ to nitrogen (N₂) ratio in the atmosphere from an arbitrary reference air according to the following equation:

　　　δ(O₂/N₂)={(O₂/N₂)_sam/(O₂/N₂)_ref-1}×10⁶　（１）

where the subscripts “sam” and “ref” represent “sample” and “reference”, respectively. The unit of “per meg” is used to express the δ(O₂/N₂) value. A change of 4.77 per meg corresponds to a change of 1 ppm in the trace component.
2 Atmospheric potential oxygen (APO) is defined by the following equation:

　　　APO=O₂+α_B×CO₂　（２）

where α_B represents the exchange ratio of O₂ and CO₂ in the terrestrial biosphere. Since O₂/N₂ ratio is used to express the O₂ concentration change as is described in Note 1, APO is calculated as follows:

　　　δAPO=δ(O₂ /N₂)+α_BX_CO2 /X_O2)-1850　（３）

3 Since the APO measurements are calculated as the sum of δ(O₂/N₂) and CO₂ concentrations as shown in equation (3), the ocean N₂ fluxes are also used in addition to the ocean O₂ and CO₂ fluxes to simulate APO. Furthermore, since the APO is also affected by fossil fuel combustion, the CO₂ flux from fossil fuels is also used in the calculation.

This study was supported by funds provided by a Grant-in-Aid for Scientific Research (S) (KAKENHI) of the Japan Society for the Promotion of Science (Grant Number 22H05006) and the Global Environmental Research Coordinate System from the Ministry of the Environment, Japan (grant no. E1951).

【Article title】 Observed APO Seasonal Cycle in the Pacific: Estimation of Autumn O₂ Oceanic Emissions 【Authors】 Yasunori Tohjima, Tomoko Shirai, Misa Ishizawa, Hitoshi Mukai, Toshinobu Machida, Motoki Sasakawa, Yukio Terao, Kazuhiro Tsuboi, Shintaro Takao, Shin-ichiro Nakaoka 【Journal】Global Biogeochemical Cycles 【URL】https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2024GB008230 【DOI】10.1029/2024GB008230

National Institute for Environmental Studies
　Earth System Division
　Biogeochemical Cycle Observation Section
　Specially Appointed Researcher　 Yasunori Tohjima
　　Guest Researcher　　　　　　 Hitoshi Mukai
　　Senior Researche　　　　　　 Yukio Terao

　Center for Global Environmental Research
　Office for Global Environmental Data Integration and Analytics
　　Head　　　　　　　　　　　　Tomoko Shirai
　　Guest Researcher　　　　　　 Misa Ishizawa¹⁾

　Center for Global Environmental Research
　Office for Atmospheric and Oceanic Monitoring
　　Head　　　　　　　　　　　　Toshinobu Machida
　　Chief Senior Researcher　　　 Motoki Sasakawa
　　Senior Researcher　　　　　　Shin-ichiro Nakaoka
　　Senior Researcher　　　　　　Shintaro Takao

Meteorological Research Institute
　Department of Climate and Geochemistry Research
　　Head　　　　　　　　　　　　Kazuhiro Tsuboi²⁾

1)　Environment and Climate Change Canada
2)　Now at Japan Meteorological Agency, Atmosphere and Ocean Department

National Institute for Environmental Studies
Earth System Division
Yasunori Tohjima

National Institute for Environmental Studies
Planning Division, Public Relations Office
Email: kouhou0 (please append @nies.go.jp to this address)