Full latitudinal marine atmospheric measurements of iodine monoxide

Iodine compounds destroy ozone (O₃) in the global troposphere and form new aerosols, thereby affecting the global radiative balance. However, few reports have described the latitudinal distribution of atmospheric iodine compounds. This work reports iodine monoxide (IO) measurements taken over unprecedented sampling areas from the Arctic to the Southern Hemisphere and spanning sea surface temperatures (SSTs) of approximately 0 to 31.5 ^∘C. The highest IO concentrations were observed over the Western Pacific warm pool (WPWP), where O₃ minima were also measured. There, a negative correlation was found between O₃ and IO mixing ratios at extremely low O₃ concentrations. This correlation is not explained readily by the O₃-dependent oceanic fluxes of photolabile inorganic iodine compounds, which is the dominant source in recent global-scale chemistry transport models representing iodine chemistry. Actually, the correlation rather implies that O₃-independent pathways can be similarly important in the WPWP. The O₃-independent fluxes result in a 15 % greater O₃ loss than that estimated for O₃-dependent processes alone. The daily O₃ loss rate related to iodine over the WPWP is as high as approximately 2 ppbv (parts per billion by volume) despite low O₃ concentrations of approximately 10 ppbv, with the loss being up to 100 % greater than that without iodine. This finding suggests that warming SST driven by climate change might affect the marine atmospheric chemical balance through iodine–ozone chemistry.

1 Introduction

Halogens play an important role in tropospheric and stratospheric chemistry through the catalytic destruction of ozone (O₃), which affects the atmosphere's oxidizing capacity and the radiative balance of the Earth (Alicke et al., 1999; Koenig et al., 2020; Read et al., 2008; Saiz-Lopez et al., 2012, 2014; Simpson et al., 2015). Iodine, particularly, is potentially important in tropospheric chemistry because of its rapid reactions, although its concentration in the troposphere is low compared to that of chlorine and bromine. Iodine also forms aerosol particles; it can, thereby, affect the global radiative balance (O'Dowd et al., 2002; Sipila et al., 2016; Gómez-Martín et al., 2020; Baccarini et al., 2020; Gómez-Martín et al., 2021; He et al., 2021).

Because of their low concentrations in the atmosphere, iodine compounds are difficult to quantify. Few reports have attempted to clarify their regional- to global-scale sources and roles in atmospheric chemistry (Großmann et al., 2013; Mahajan et al., 2012; Prados-Roman et al., 2015a; Dix et al., 2013; Volkamer et al., 2015). In the past, the primary source of iodine in the troposphere has long been regarded as organic compounds in coastal areas (Davis et al., 1996; Carpenter et al., 2012; Prados-Roman et al., 2015a). However, the results of recent studies suggest that iodine compounds over the open ocean are emitted from inorganic sources following O₃ deposition over the ocean surface (Carpenter et al., 2013; Macdonald et al., 2014; Prados-Roman et al., 2015b). The inorganic sources are now regarded as the dominant emission term over the oligotrophic oceans in the global-scale chemistry transport models representing iodine chemistry (e.g., Saiz-Lopez et al., 2014; Sekiya et al., 2020), although the emission process of inorganic iodine is still insufficiently clear in more recent studies (e.g., Inamdar et al., 2020).

This study specifically examines iodine monoxide (IO) in the marine boundary layer over the open ocean from the wide latitudinal bands. Specifically, we examine processes occurring over the tropical Western Pacific, where the global sea surface temperature (SST) reaches a maximum (warm pool) and where O₃ minima have been reported (Rex et al., 2014; Kanaya et al., 2019; Kley et al., 1996). Actually, IO observations in environments with SSTs of > 30 ^∘C are limited. Observations of IO have been made in the tropics but only for short time periods, with SST > 30 ^∘C, if any (Großmann et al., 2013; Dix et al., 2013; Prados-Roman et al., 2015a). Although the importance of halogen chemistry as a driver of O₃ losses in this region has been suggested (Großmann et al., 2013; Koenig et al., 2017), this point has yet to be examined in the context of full latitudinal distributions.

The initial production of atmospheric inorganic iodine species has not been fully examined in an environment where extremely low O₃ concentrations (< 10 ppb – parts per billion) are observed. Over the Atlantic Ocean (in Cabo Verde), long-term observations of iodine and ozone have been conducted, but they were in higher O₃ environments of approximately 20 ppbv (parts per billion by volume; Read et al., 2008). We, therefore, examined IO variations over the tropical Western Pacific and their potential contributions to regional O₃ losses, with an emphasis on SST as a potential key parameter controlling the initial iodine emissions. The global SST maximum is observed in the tropical Western Pacific, but observations reported from earlier studies were only taken in the regions surrounding the maximum (Großmann et al., 2013; Prados-Roman et al., 2015a). Investigation of iodine variations in the tropics is also important for elucidating the stratospheric chemical balance (Koenig et al., 2017) because transport from the troposphere to the stratosphere occurs through the tropical tropopause layer (Takashima et al., 2008; Saiz-Lopez et al., 2015; Koenig et al., 2017; Holton et al., 1995). In fact, it is particularly important over the tropical Western Pacific.

For this study, using the multi-axis differential optical absorption spectroscopy (MAX–DOAS) remote sensing technique, IO observations were made to quantify IO concentrations over the open ocean, covering the widest latitudinal range ever examined with a single instrument. The technique uses scattered solar radiation at several elevation angles to obtain atmospheric aerosol and gas profile concentrations (Hönninger et al., 2004; Wagner et al., 2004; Wittrock et al., 2004; Sinreich et al., 2005; Frieß et al., 2006; Kanaya et al., 2014). MAX–DOAS generally measures trace gas contents over a long light path (up to 10–20 km) at low elevation angles. The long light path enables the detection of low concentrations of species of interest at near-surface altitudes. MAX–DOAS is, therefore, useful for quantifying low-abundance tropospheric trace gases, such as IO, over the open ocean.

Multi-platform measurements by MAX–DOAS from aircraft (Koenig et al., 2017; Volkamer et al., 2009) and ships (Großmann et al., 2013; Takashima et al., 2012; Volkamer et al., 2009) have been developed in recent years. Earlier studies have retrieved IO concentrations (typically < 1 pptv – parts per trillion by volume) in the marine boundary layer over the open ocean from shipboard MAX–DOAS measurements (Großmann et al., 2013; Mahajan et al., 2012; Prados-Roman et al., 2015a; Inamdar et al., 2020). Since 2008, the Japan Agency for Marine–Earth Science and Technology (JAMSTEC) has undertaken an unprecedented set of MAX–DOAS measurements on board the research vessels (R/Vs) Kaiyo, Mirai, and Kaimei around the world (Takashima et al., 2012). This report presents IO and O₃ variations over the open ocean from the Arctic to the Southern Hemisphere as observed on R/V Mirai between 2014–2018.

2 Methodology

2.1 Iodine monoxide observations from ship-based MAX–DOAS measurements

The shipboard MAX–DOAS apparatus used for this study comprised two main components, i.e., an outdoor telescope and an indoor UV–Vis spectrometer (SP-2358 coupled to a PIXIS 400B back-illuminated CCD (charge-coupled device) detector; Acton Research Corporation and Teledyne Princeton Instruments, respectively). These were connected using a 10–14 m long fiber-optic cable (100 µm radius; 60 core or 40 core). The telescope unit was developed jointly by the Japan Agency for Marine–Earth Science and Technology (JAMSTEC) and Prede Co., Ltd. (Tokyo, Japan). The movable prism of the telescope unit rotates for elevation angles (ELs) of 3, 5, 10, 30, and 90^∘. The EL is changed every minute to observe scattered sunlight. The target EL is attained by adjusting the angle of the prism actively and by considering the angle of the ship's roll (Takashima et al., 2016). The telescope line of sight was off the starboard side of the ship, with a field of view of approximately 1.0^∘. The spectrometer was housed in an adiabatic plastic box, with the temperature held constant at 35 ^∘C ± 0.1 ^∘C using a temperature controller (KT4; Panasonic Inc., Japan). The CCD was cooled to −70 ^∘C. The spectrometer was equipped with a 600 line per millimeter grating at 300 nm. The slit width was 100 µm. The typical exposure time was 0.1–0.2 s. Spectral data were selected for analysis when the EL was within ±0.5^∘ of the target. Data were analyzed using the DOAS method (Platt and Stutz, 2008). A nonlinear least squares spectral fitting procedure was used to derive differential slant column densities (DSCDs) of the oxygen collision complex (O₂–O₂ or O₄) and IO using the QDOAS software package (Danckaert et al., 2017), for which the absorption cross section data presented in Table 1 were used. For O₄ and IO retrievals, 425–490 and 415–438 nm fitting windows were applied, respectively. Examples of fitting results and the time series of DSCDs are presented, respectively, in Figs. 1 and 2. The typical fitting error of the IO DSCDs was approximately 1 × 10¹² molec. cm⁻², with a detection limit of approximately 4 × 10¹² molec. cm⁻² (2σ).

Figure 1 Nonlinear least squares spectral fitting results for IO concentrations observed on 2 December 2014. Panels (a) and (b) show the fitting for IO and H₂O. Black lines represent the cross section scaled to the spectrum (red) determined by differential optical absorption spectroscopy. Panels (c) and (d) show the Ring effect contribution and the residual spectrum.

Figure 2 Time series of IO and H₂O differential slant column densities (DSCDs) for elevation angles of 3, 5, 10, 20, 30, and 90^∘, the root mean square residual, and the solar zenith angle observed on 1–2 December 2014 over the tropical Western Pacific.

Table 1 Cross sections of iodine monoxide (IO) and O₄ differential slant column densities used for this study.

^* Correction factors from Lampel et al. (2015) were applied.

The Mexican MAX-DOAS Fit (MMF) retrieval algorithm (Friedrich et al., 2019) was used for the retrieval of IO profiles and vertical column densities. The version of MMF used in this study is the same as used in Frieß et al. (2019) and Tirpitz et al. (2021) but with adjusted a priori and variance–covariance matrix settings to fit for IO retrieval. MMF applies the optimal estimation method and uses a two-step approach in which the aerosol profile is first retrieved from O₄ DSCDs. Then, the IO profile is retrieved from the IO DSCDs, using the earlier retrieved aerosol profile in the forward model. We used VLIDORT (v.2.7; Spurr, 2006) as the forward model in a pseudo-spherical multiple-scattering setting. Only intensity information and its analytically calculated Jacobians were used. No other Stokes parameter was used. MMF was used in logarithmic retrieval space on a retrieval grid of up to 4 km with a 200 m layer height.

Both a priori profiles were constructed as constant below 500 m, with an exponentially decreasing profile above 500 m for aerosol and IO profiles, to examine near-surface areas specifically. The a priori aerosol optical depth was set as 0.18. The a priori IO vertical column density (VCD) was set to 2.5 × 10¹² molec. cm⁻². The a priori covariance matrix S_a for both aerosol and IO retrieval was constructed using the square of 100 % of the a priori profile on the diagonal and a correlation length of 200 m. For the aerosols, the only retrieved quantity was the partial aerosol optical depth per layer. Therefore, in the forward model, a constant single scattering albedo of 0.95 was used for both retrievals, i.e., aerosol and IO. The phase function moments were constructed using the Henyey–Greenstein phase function (Henyey and Greenstein, 1941) with a constant asymmetry factor of 0.72. The surface albedo in the forward models was set as 0.06. Here, the single scattering albedo, asymmetry factor, and surface albedo were used, similar to the work presented by Großmann et al. (2013). The degrees of freedom (DOF) for the IO retrieval for MR14-06 (leg1) were 1–1.4. Typical averaging kernels for IO are presented in Fig. S1 in the Supplement. It is also noteworthy that the observed IO contents might be a little low compared to those from earlier studies conducted over the open ocean because of inaccuracy of the water–vapor cross section used in earlier retrievals (Lampel et al., 2015).

2.2 Zero-dimensional photochemical box model with iodine chemistry

A zero-dimensional photochemical box model (Kanaya et al., 2007a, b), based on the Regional Atmospheric Chemistry Mechanism (RACM; Stockwell et al., 1997) and custom iodine chemistry, was updated to include 91 chemical species and 275 reactions (reactions of iodine chemistry added to RACM are presented in Table 3). It was used to simulate the time evolution of mixing ratios of O₃ (initially 18 ppbv) and iodinated species in the boundary layer with an assumed height of 500 m over the equatorial Pacific region, where the maximum concentrations of IO and minimum concentrations of O₃ were observed. For O₃, dry deposition at a velocity of 0.04 cm s⁻¹ was considered (Pound et al., 2020). An entrainment flux of 1.2 × 10⁸ molec. cm⁻² s⁻¹ was assumed for NO₂, for which the initial mixing ratio was assumed to be 15 pptv. Fluxes of hypoiodous acid (HOI) and I₂ from the ocean surface were estimated, respectively (Carpenter et al., 2013), as 8.4 × 10⁷ and 2.6 × 10⁶ molec. cm⁻² s⁻¹ at 10 ppbv of O₃, for an aqueous I⁻ concentration of 74 nM and wind speed of 5 m s⁻¹ (8.9 × 10⁷ molec. cm⁻² s⁻¹ as total HOI$/$I₂ (= HOI + 2I₂) flux). The I⁻ concentration was referred from the nearest observation data at 12^∘ N and 158^∘ E (Tsunogai and Henmi, 1971). The assumed wind speed was from observations made during MR14-06 cruise over the region. For Case 1a, the fluxes were assumed to be linearly dependent on O₃, which is consistent with Carpenter et al. (2013). For Case 1b, a 25 % reduction in the flux was assumed, potentially because of the presence of a sea surface microlayer or dissolved organic matter (Hayase et al., 2010, 2012; Shaw and Carpenter, 2013; Tinel et al., 2020). The blue band of Fig. 3 represents the range of Cases 1a and 1b, representing the case with O₃-dependent fluxes. In Cases 2a and 2b, the O₃-dependent flux in Case 1a was reduced to half and compensated by O₃-independent inorganic iodine fluxes of 3.3 (or 6.6) × 10⁷ molec. cm⁻² s⁻¹ (red band of Fig. 3, which represents the quasi-O₃-dependent case). As a reference, a hypothetical case (Cases 3a and 3b) with a purely O₃-independent flux of the magnitude of 9.9 (or 13) × 10⁷ molec. cm⁻² s⁻¹ was also tested (orange band of Fig. 3, which represents the purely O₃-independent case). The time-dependent simulations continued for 5 d, with the evaluation of the mixing ratio of O₃ and its relation with IO involving daytime averages (06:00–18:00 ship local time) over the first to fourth days. Dry deposition velocities of iodine species (I, IO, HI, HOI, OIO, I₂O₂, INO, INO₂, IONO₂, and I₂) were assumed to be 1 cm s⁻¹.

Table 2 Reactions of iodine chemistry added to RACM.

Table 3 Research cruises of the R/V Mirai that generated the data used for this study.

Figure 3 Daily median IO mixing ratio for 0–200 m (pptv) observed by MAX–DOAS versus daily median in situ ozone mixing ratio (ppbv). Results of box model simulations with O₃-dependent (Case 1), quasi-O₃-dependent (Case 2), and pure O₃ independent (Case 3) emission fluxes of iodine compounds are superimposed respectively as blue, red, and orange shaded areas.

2.3 Backward trajectory calculation

The origins of air masses over the tropical Western Pacific were investigated using 5 d backward trajectory calculations (Takashima et al., 2011) based on meteorological analysis data of the European Centre for Medium-Range Weather Forecasts (ECMWF).

2.4 In situ gas measurements

For measurements of O₃ and CO, ambient air was sampled using approximately 20 m of Teflon tubing (6.35 mm outer diameter) from the bow (Kanaya et al., 2019). To avoid contamination from ship exhaust, 1 min data that deviated more than 1σ from the hourly discrete average were deleted. The typical magnitude of 1σ over the remote ocean was approximately 0.1–0.5 ppbv.The O₃ and CO concentrations were measured, respectively, using UV and infrared absorptions with O₃ and CO monitors (49C and 48C; Thermo Scientific, USA). The O₃ instrument was calibrated twice per year in the laboratory, before and after deployment, using a primary standard O₃ generator. The CO instrument was calibrated on board twice per year, on embarking and disembarking of the instrument, using a premixed standard gas. The reproducibility of the calibration was to within 1 % for O₃ and 3 % for CO (Kanaya et al., 2019). The O₃ concentrations observed from R/V Mirai cruises, presented in Table 3, are shown in Fig. S2.

3 Results and discussion

The IO contents (differential slant column densities (DSCDs) for an elevation angle of 3^∘) observed from R/V Mirai during seven research cruises during 2014–2018 are presented in Fig. 4. The cruises are presented in Table 3. Although observations were limited to some seasons and years (e.g., Arctic measurements were limited to the Northern Hemisphere summer), whole latitudinal bands were covered from 74^∘ N to 67^∘ S, and strong latitudinal variations in IO concentrations were observed, with a maximum detected clearly in the tropics (10^∘ S–10^∘ N) but not at higher latitudes in either hemisphere. Over Southeast Asia (near Indonesia), high IO concentrations were sometimes observed near coastal areas. The highest values of up to approximately 2 × 10¹³ molec. cm⁻² (DSCD) were also observed in the tropical Western Pacific, with wide variations in global SST maxima (> 30 ^∘C). From similar earlier studies (Gómez-Martín et al., 2013; Großmann et al., 2013; Mahajan et al., 2012), no data obtained under very high SST conditions over a long period were reported. Therefore, our IO observations at SST maxima (up to 31.5 ^∘C), and during more than 2 weeks, represent the most comprehensive measurements of reactive iodine over the tropical Western Pacific warm pool (WPWP).

Figure 4 Daily median IO content (DSCDs for an elevation angle of 3^∘; molec. cm⁻²) observed from the R/V Mirai during 2014–2018. Color contours represent the optimum interpolated SST averaged for 2014–2018.

Specifically regarding IO variations over the tropical Western Pacific, we found IO VCDs of approximately 0.7–1.8 × 10¹² molec. cm⁻² (Fig. 5). The 5 d backward trajectories indicate that air masses in this region originated from the open ocean (Fig. 5). The carbon monoxide (CO) content was constantly low (60 ppbv; Fig. 6), which is also consistent with an air mass originating from the open ocean. In addition, the chlorophyll content, based on satellite MODIS measurements (MODIS-Aqua, MODISA, Level 3 version 2018) in the source region, was also low (Fig. 5), implying that any organic source of iodine can be expected to be negligible (although we also must consider abiotic organic source and mesotrophic conditions; Jones et al., 2010). The IO data collected over the tropical Western Pacific are consistent with I⁻ variations reported in earlier studies (Chance et al., 2014, 2019; Sherwen et al., 2016), indicating an increase in I⁻ concentration with SST.

Figure 5 Daily median tropospheric IO vertical column densities (VCDs; molec. cm⁻²) observed from R/V Mirai during 16 November to 2 December 2014 and chlorophyll a concentrations observed via satellite (MODIS). Blue crosses and lines represent 5 d backward trajectories.

For the time series of IO concentrations near the ocean surface (0–200 m height; Fig. 7), the values were approximately 0.3–0.8 pptv, with wide variations over a timescale of a few days. The IO concentration near the surface depends on the shape of the a priori profile used for the retrieval, but day-to-day variations near the surface were unaffected by the choice of profile. Insufficient data were retrieved to document diurnal IO variations accurately. At times, the O₃ concentrations were generally low (< 20 ppbv) and extremely low (< 10 ppbv; Fig. 7). One unique finding was that, even under low-O₃ conditions, a negative correlation was found between IO and O₃ concentrations in the daily dataset (Figs. 3 and 7). Laboratory studies indicate that high O₃ concentrations can cause emissions of iodine from ocean to atmosphere (Carpenter et al., 2013; Macdonald et al., 2014; Sakamoto et al., 2009). This O₃-dependent iodine release has been regarded as being more dominant than other O₃-independent types of emissions, including photo-labile iodocarbons such as CH₂I₂, CH₂ICl, and their subsequent photolysis over the open oceans in every global-scale chemistry transport model representing iodine chemistry (Saiz-Lopez et al., 2014; Sekiya et al., 2020; Sherwen et al., 2016). However, with an O₃-dependent HOI/I₂ flux of approximately 9 × 10⁷ molec. cm⁻² s⁻¹ (Sect. 2.2), as expected under an O₃ mixing ratio of approximately 10 ppbv, a zero-dimensional box model was not able to reproduce the negative correlation found between IO and O₃. Because the initial HOI/I₂ release flux limited by O₃ in the < 12 ppbv mixing ratio range cannot drive the strong O₃ reduction, the scenario produced only a positive correlation (Case 1; Fig. 3). In contrast, another case, in which the O₃-independent flux was added to compensate for the O₃-dependent term weakened by a factor of 2 (Case 2; Fig. 3), better reproduced the observed trend. The weakened flux might be explained by dissolved organic carbon (Shaw and Carpenter, 2013) or the presence of a sea surface microlayer (Tinel et al., 2020) impeding iodine vaporization. The added O₃-independent flux is not explainable solely by flux from the photolysis of iodocarbons within the marine boundary layer (approximately 10⁷ molec. cm⁻² s⁻¹) generally assumed in the three-dimensional models (Saiz-Lopez et al., 2014; Sekiya et al., 2020; Sherwen et al., 2016). While indirectly considering the global total fluxes of CH₂IX (X = I, Br, and Cl), as described by Ordóñez et al. (2012) in these model simulations, the chl-a-based parameterization reduced the fluxes to too a low level over this oceanic region. It, therefore, necessitates a survey of missing sources. We might not need a brand new flux mechanism but, rather, a good parameterization of the traditional organoiodine fluxes (including their photolysis) over the region.

Figure 6 Time series of the CO mixing ratio (ppbv), color index (CI; defined as the ratio of the measured intensities at the two wavelengths of 500 and 380 nm; Takashima et al., 2009), wind speed (m s⁻¹), and SST (^∘C).

Table 4 Net and process-specific O₃ loss rates in three cases at an O₃ concentration of 10 ppbv, as calculated using the box model.

Figure 7 Time series of IO mixing ratio for 0–200 m (blue; pptv) observed by MAX–DOAS, and in situ O₃ mixing ratio (red; ppbv). Circles and horizontal bars, respectively, represent the daily median and 1 standard deviation.

The third case, with only an O₃-independent flux (Case 3; Fig. 3) might explain the negative correlation more easily, whereas the total change of the flux type is simply not being supported. We, therefore, hypothesize that O₃-independent processes are more important than has been represented by recent models. Indeed, a larger magnitude of organic iodine flux (approximately 7 × 10⁷ molec. cm⁻² s⁻¹) was reported in the low-latitude Pacific (Großmann et al., 2013) and would, therefore, be the most likely cause of the negative correlation. However, that study (Großmann et al., 2013) relied on assumption of an even larger inorganic iodine emission flux to explain the observed IO concentrations. Therefore, our analysis is the first to suggest that the O₃-independent flux can be comparably important to the O₃-dependent flux in this region. Other O₃-independent iodine release mechanisms, such as photooxidation of aqueous I⁻ (Watanabe et al., 2019), might also be worth exploring. The modeled net O₃ loss rate attributable to iodine in Case 2 increased by up to 100 % over that without iodine. The O₃ loss rate in the iodine cycle in Case 2 increased by approximately 15 % over that in Case 1 (Table 4).

The expectation that a positive correlation between O₃ and IO would occur with O₃-dependent processes over a low O₃ concentration range was also confirmed using three-dimensional global chemistry transport models, including halogen chemistry (Sekiya et al., 2020; Saiz-Lopez et al., 2014), over the tropical Western Pacific (Figs. S3, S4). An alternative explanation for the observed negative correlation would be the mixing of air masses with different degrees of iodine chemistry. If so, such a negative correlation could appear in the chemistry transport model results. However, this feature was not found. Therefore, we propose an O₃-independent flux. Over the Atlantic, the O₃ mixing ratio rarely reaches these low levels (10 ppbv or fewer). Therefore, such process analyses have not been undertaken there. Under the influence of O₃-independent sources, even lower O₃ concentrations would be attainable. Radiative forcing of O₃, as estimated recently with halogen chemistry (Sherwen et al., 2017; Iglesias-Suarez et al., 2020; Saiz-Lopez et al., 2012; Hossaini et al., 2015), might be influenced by the dependence of iodine flux on O₃ concentration, which might play a major role in estimating past and future concentrations of O₃.

The time series of meteorological parameters including wind speed and SST was also investigated, but a clear correlation, such as that shown by O₃ and IO, was not observed in the correlation with O₃ or IO concentrations on a timescale of a few days (Fig. 6). The correlation coefficient between SST and IO was found to be 0.39, that between SST and O₃ was −0.51, and that between wind speed and IO was −0.45. In addition, that between wind speed and O₃ was 0.59. It is noteworthy that the correlation coefficient between IO and O₃ was −0.75, which is much higher than others and, therefore, is the dominant feature. An earlier study (Kanaya et al., 2019) investigating the diurnal variation of O₃ in this area based on a comparison of observational data and a chemical transport model indicated that an as-yet-unidentified O₃ loss might occur over the tropical Western Pacific. Our results imply that iodine chemistry plays an important role in O₃ loss in the area of SST maxima, which is regarded as an entry point from the troposphere to stratosphere. Moreover, these results provide insights into the manner by which increasing SST associated with climate change might modify the marine atmospheric chemical balance, which warrants further investigation. Results of recent studies indicate a roughly threefold increase in iodine since the 1950s, with at least 50 % attributed to anthropogenic O₃ (Cuevas et al., 2018; Legrand et al., 2018; Zhao et al., 2019). If half of the inorganic flux were O₃-independent, as suggested by Case 2, then either some other cause should be sought or the change in O₃-dependent fluxes to produce the observed change is even more dramatic than previously thought. Further investigation of these points is necessary.

4 Summary

In this study, shipboard multi-axis differential optical absorption spectroscopy (MAX–DOAS), a remote sensing technique, was used during seven research cruises covering the widest latitudinal bands, from the Arctic to the Southern Hemisphere, ever made with a single instrument, spanning SSTs of approximately 0 to 31.5 ^∘C and allowing the investigation of the variation of IO concentrations. It was particularly abundant over the tropical Western Pacific (warm pool), appearing as an iodine fountain, where SST maxima (> 30 ^∘C) and O₃ minima are observed. This report describes negative correlation between IO and O₃ concentrations over the IO maximum, even under extremely low O₃ conditions, which few earlier studies have demonstrated. This correlation is not explained easily by the O₃-dependent oceanic fluxes of photolabile inorganic iodine compounds adopted for recent simulation studies. Our findings rather imply that O₃-independent pathways which release iodine compounds from the ocean are also important. Iodine input to the atmosphere from the ocean surface is greater in areas of higher SST, leading to an iodine fountain in the Western Pacific warm pool because the I⁻ concentration in the ocean surface is likely to be higher in these areas. This higher concentration might contribute to more pronounced O₃ destruction over the Western Pacific warm pool than estimated earlier. Warming SSTs associated with climate change can change the atmospheric chemical balance through halogen chemistry, warranting further quantitative investigation.