Response to the Referees A negative feedback between anthropogenic ozone pollution and enhanced ocean emissions of iodine

We are grateful to Dr. Sander and to the anonymous Referee#1 for their constructive comments and for their appreciation of our work. Also, we would like to thank the interest on our work shown by the editor. Herein we address point-by-point the different suggestions (Referee CommentRC in bold letters, Authors CommentAC). Besides these responses, please note that in the new version of the manuscript the reference to the work of Saiz-Lopez et al. 2014 has been updated.

We are grateful to Dr. Sander and to the anonymous Referee#1 for their constructive comments and for their appreciation of our work. Also, we would like to thank the interest on our work shown by the editor. Herein we address point-by-point the different suggestions (Referee Comment-RC in bold letters, Authors Comment-AC). Besides these responses, please note that in the new version of the manuscript the reference to the work of Saiz-Lopez et al. 2014 has been updated. Attached to this document we also include the new version of the manuscript with changes marked in red.

Prados-Roman et al. investigate the feedback between anthropogenic ozone and marine iodine emissions. The study is very interesting and I recommend publication in ACP after considering my suggestions as described below.
My only major scientific concern is the question if surface iodide will remain constant in the future. If ozone levels continue to increase, and the oxidation of surface iodide by ozone also becomes faster, will this lead to a depletion of surface iodide? As far as I can see, the model calculates iodide as a function of temperature only. How fast is surface iodide replenished? How would the results change if the concentration of surface iodide decreases in the future?

AC:
We appreciate Dr. Sander's comments on our past vs. present modelling exercise. Indeed as the referee points out, predictions into the future are somehow complicated to address. Although future trends fall out of the scope of our manuscript, we understand that they are the logical progression from our work and they do deserve a whole separate study. Among changes in tropospheric ozone, there are indeed many parameters with an uncertain future trend such as the sea surface temperature (SST) or the wind speed and wind stress (IPCC) that may also affect the concentration of iodide. Further combined field and laboratory work and modelling studies would be needed to establish a future trend not only on those parameters but also on their geochemical coupling. Hopefully manuscripts like this encourage the scientific community to combine efforts towards that direction. Regarding the model's treatment of the iodide content in sea-water, as detailed in Sect. 2.2, it is indeed computed only as a function of SST. As the ocean component of CESM is not coupled with the atmosphere model CAM (identical SST and sea-ice conditions are used for Present-Day (PD) and Pre-Industrial (PI) times), we do not compute aqueous iodide concentrations nor depletion or replenishing rates on the oceans. As detailed in the manuscript, the assumed iodide concentration in the model is always obtained by means of the SST parameterised formulations given by MacDonald et al. (2014), yielding annual average concentrations that lie within the range of the sparse measured values reported in literature (Chance et al., 2014). Projecting scenarios of aqueous iodide concentration into the future (or for the past) only by means of a SST dependent function should be done with caution, as other quantities (such as acidity, replenishing rates, etc.) could also affect its temporal evolution. But until new findings are AC: We appreciate Dr. Sander's opinion. Nevertheless the establishment of the feedback mechanism between the anthropogenic increase of tropospheric ozone and the enhanced emission of ISG from the oceans is in fact one of the main results of our study. Therefore we consider the mechanism itself deserves a separate section as appears in the manuscript.

RC:
• In the acknowledgements, you mention that data supporting this article can be requested from the corresponding author. I think it would be much better if these data are included in the electronic supplement of the paper. I often had problems getting data for older papers because the authors could not be reached anymore. Putting the data into the supplement, however, they will be permanently archived together with the main article.
AC: For a global modeling exercise like this, we think it would be unmanageable to include tables with 3-D data from all variables across the global domain. Therefore, we maintain that specific data can be provided to the interested reader upon request. RC: • Fig. 6: I think that a multicolored pie-chart with a 3D effect is an overkill for presenting just 5 numbers. In my opinion, a small table would have been sufficient.

AC:
We are grateful to Referee #1 for his/her comments. We proceed to address the reviewer concerns point-by-point.

RC:
Specific points: 1. The manuscript would benefit from a more comprehensive assessment of the uncertainty in their interpretation in several areas, moreover, this uncertainty should be expressed in the values presented for increased ISG emissions, increased rates of ozone loss and the overall impact on radiative forcing. MacDonald et al. 2014 carried out some sensitivity analyses of their model parameterizations and a similar assessment of the robustness of the results of this model is needed.

AC:
Although the model and the input parameters used are based on the state-of-the-art knowledge of the different compounds and processes involved in the ocean-atmosphere system, as the referee points out the modelling exercise we present-as any other modelling exercise-is linked to uncertainties. Note however that, although the magnitude of the changes in the tropospheric ozone budget or in the fluxes of ISG might be affected by those uncertainties, the establishment of the geochemical feedback describe in our study mechanism (i.e., the increase in tropospheric ozone since PI has yielded an increase emission of ISG from the oceans and this an acceleration of the ozone loss rate) is not. This paper intends to provide a hypothesis about a geochemical feedback that stills needs to be experimentally confirmed and its uncertainties further constrained by observations. Once this is said, we understand the referee's concerns regarding the degree of uncertainties and approximations assumed for the oceanic emissions of inorganic iodine. But as mentioned above, the parameterisation of the ISG fluxes based on the experimental work presented by Carpenter et al. (2014) and MacDonald et al. (2014) is, to our knowledge, the best proxy and therefore that is the one we have used in our work. Thus, we believe that the feedback mechanism and main implications presented in this work are fairly reliable. This is inherent for many other chemical, oceanic and meteorological components of chemistry-climate models which sometimes include very simplified expressions to represent extremely complex processes. Anyhow, it is worth implementing mathematical expressions that, even when approximated or simplified, allow us to represent a process "on-line" by means of other variables which are strongly constrained and coupled. This is the case for the implementation of the parameterised expression of the ISG flux (Eqs. (1)-(4)), which should be taken as a first order approximation to the real process in the atmosphere. In this way, shifting from a previous CAM-Chem setup imposing an additional ISG from a boundary condition file to this new version with an "on-line" estimation of the iodine sources, constitutes a step forward improving the knowledge upon possible geochemical-feedbacks required to understand the pastpresent-future evolution of the Earth-atmosphere system.  2014)). Hence we feel confident on the ISG flux levels and on their geographical distribution reproduced in our work.

AC:
As mentioned above, we believe it is worth to implement "on-line" formulations for the ISG fluxes in CAM-Chem even when only approximated parameterised expressions for the process are known. The only validation we can perform with these types of global models is to assess if the ISG fluxes obtained lie within the range of values reported in the literature. Indeed, the global modelled emissions of HOI/I 2 in the current version of the model account for ~1.9 Tg (I) yr −1 . This value is somewhat larger than the one ~1. (1)-(4), which overestimates the ISG fluxes for wind speed < 3m s -1 . MacDonald et al. mentioned "a factor of two at a wind speed of 0.5 m s -1 ". That sort of limitation is hence intrinsic to our work since we used their ISG flux parameterisation. Nevertheless, globally speaking, the wind speed over the oceans tends to be higher than the mentioned threshold. Only in some areas of the Equator the wind speed is consistently close to the threshold value. Note that, as shown in the figure below, only in the small region of the Halmahera Sea (offshore West Papua Province) the mean wind speed is fact below 3 m s -1 (never below 1 m s -1 in average). Hence, in general, we can assume that the ISG flux parameterisation based on wind speed used in our study is globally valid. As mentioned above, this is also confirmed when comparing our results to field campaign measurements (Chance et al., 2014). Noteworthy is also the fact that the major changes in tropospheric ozone, ISG fluxes and I y over time (Fig. 3b, Fig. 4 and Fig. 5c; respectively) are not particularly located over regions of low wind speed.

AC:
We understand the referee and, in fact, this issue is a matter of chosen reference. We would rather keep the Present-time as the reference time in these studies since most of the community is more familiar with the current tropospheric ozone levels. Also it is worth noting that, as the O 3 vmr in PI times were smaller than those for the PD, the percentage change when using PD in the denominator are smaller than those obtained if we would have used PI as the reference time.

AC:
The referred sentence in the introduction now reads as "laboratory studies have demonstrated the potential of the ocean to emit…" RC: 4. P21922, L17. This point is not entirely fair. This single factor might lower the estimates of ISG fluxes but the overall results comprise many other levels of uncertainty that could shift the balance between under or overestimation.

AC:
Indeed what we understand as a lower limit are the reproduced ISG fluxes. This has now made clearer in the new manuscript (end of Sect. 2.2) with "Hence the ISG fluxes modelled in this study should be regarded as lower limits". RC: 5. P21925, L3.3 I suggest altering the title to 'Iodine-mediated change in ozone radiative forcing. . ..'

AC:
The title of the Subsection is changed as suggested.
RC: 6. P21926, L25+. As stated above, a comprehensive explanation of the model uncertainties is needed in order to demonstrate how robust these conclusions are.

AC:
As mentioned above and also in the manuscript (Sec. 3.4), although the uncertainties in, e.g., the parameterisation of ISG fluxes (which appear to be rather small based on Chance et al., 2014) or in the change of SST over time (very uncertain based on IPCC), may propagate into the magnitude and geographical distribution of the changes in ISG flux, Iy and ozone loss rate shown in our work; the establishment of the proposed feedback mechanism is not linked to those uncertainties since it is a result of the human activities increasing the background ozone (as demonstrated in IPCC) and the experimentally proved emission of ISG as a result of the deposition of tropospheric ozone in the ocean (Garland et al. 1980). See also comments above.    (Fig. 3b). Consequently, the anthropogenic amplification of the natural oceanic 19 emission of iodine and, therefore the I y abundance in the MBL, also reflects a strong north 20 (NH) to south (SH) hemispheric gradient as shown in Fig. 5. 21

Change in iodine-mediated ozone loss rate since pre-industrial times 22
Considering all the ozone depleting families (i.e., odd oxygen, hydrogen, nitrogen, iodine, 23 bromine, chlorine) (Brasseur and Solomon, 2005, see also Saiz-Lopez et al., 2014), we 24 calculate that the industrialisation process has on average increased the rate of the total ozone 25 chemical loss in the global MBL from 1.89 nmol mol -1 d -1 to 3.19 nmol mol -1 d -1 , mainly 26 driven by changes in the abundance of odd oxygen, hydrogen and iodine. On a global annual 27 average, 25% of that enhanced ozone loss rate results from the human-driven boosting of 28 inorganic iodine emissions that has accelerated iodine-mediated ozone destruction from 0.54 29 nmol mol -1 d -1 in pre-industrial times, to 0.89 nmol mol -1 d -1 in the present-day. Figure 6  30 depicts the ozone loss rates by the different chemical families in the present-day scheme. As 1 shown in Fig. 7, the ozone-driven increase in iodine emissions since PI times has resulted in a 2 remarkable acceleration of ozone loss in the global MBL with a strong hemispheric gradient. 3 We calculate that since 1850 the total surface O 3 loss rate has increased by 2.1 and 0.6 nmol 4 mol -1 d -1 in NH and SH, respectively. When only the contribution of the iodine cycle is 5 considered, hemispheric annual changes in the O 3 loss rate are 0.5 and 0.2 nmol mol -1 d -1 in 6 NH and SH, respectively (Fig. 7). Notably, iodine was and still is the second strongest ozone 7 depleting family in the MBL, being responsible for about 30% of the total ozone loss in that 8 region of the atmosphere (Fig. 6). Integrating the tropospheric column, the rate of iodine-9 catalyzed ozone destruction has increased by 90 Tg y -1 since the pre-industrialisation era, 10 yielding a total present day tropospheric ozone removal by iodine of 280 Tg y -1 . 11 In general, marine regions surrounding northern developed and developing countries, and 12 areas connecting them, have undergone the strongest amplification of the natural cycle of 13 inorganic iodine emissions as a result of the enhanced deposition of ozone to those regions of 14 the ocean (Figs. 4-7). Remarkably, the current anthropogenic influence maximizes in highly 15 polluted coastal regions such as the East-South China Sea, the South Bay of Bengal, the Gulf 16 of Mexico and California's offshore waters (Fig. 3) where we calculate an increase of up to 17 70% in atmospheric iodine since PI times (Fig. 5). In these regions of continental ozone-rich 18 outflow the iodine-mediated ozone loss rate in recent times has accelerated by a similar factor, 19 i.e. about 6 times more (up to 2 nmol mol -1 d -1 ) than the global average of 0.35 nmol mol -1 d -1 20 ( Figs. 6 and 7). in the marine troposphere would mitigate the warming long-wave radiative effect of 4 tropospheric ozone by up 20% globally, and up to 40% in the NH. 5 As compared to 1850, we estimate that the abovementioned 45% increase in I y loading has 6 yielded a significant decrease in the RF associated with tropospheric ozone, reinforcing the 7 need of a better process-level understanding of the uncertainties in atmospheric iodine 8 chemistry in order to assess the impact of iodine on the tropospheric ozone RF and its future 9 trends. 10

Geochemical feedback mechanism 11
In this study we suggest that the human-driven increase of tropospheric ozone has led to an 12 amplification of the natural cycle of iodine emissions that has consequently decreased the 13 lifetime of ozone in the marine atmosphere, thus closing a negative feedback loop as 14 conceptually illustrated in Fig. 8 atmosphere, as compared to pre-industrial times, represents a mechanism by which 26 anthropogenic activities have increased the overall reactivity of the atmosphere and have 27 amplified the natural cycle of iodine. The human-mediated boosting of the ISG emissions has 28 on average increased by 25% the rate of present day ozone chemical loss in the global marine 29 environment, with regions where this increase can be up to 70%, compared with the pre-30 industrial era. The subsequent negative radiative forcing induced by the enhanced iodine-31 mediated ozone depletion currently mitigates up to 20-40% the effect of tropospheric ozone 1 as a GHG in the northern hemisphere. The human-driven enhanced iodine emissions may also 2 have two important side implications. First, it has likely led to a larger accumulation of the 3 iodine fraction (iodate and iodide) on marine aerosols (Baker, 2004). Second, it may have 4 increased the input of iodine, as an essential dietary element for mammals (Whitehead, 1984) 5 that is transported from its oceanic source to the continents. 6 The negative feedback mechanism described in this work represents a natural buffer of ozone-7 related pollution and its radiative forcing in the marine environment. This feedback represents 8 a potentially important new link between climate change and tropospheric ozone since the 9 oceanic emissions of iodine are not only linked to surface ozone, but also to SST and wind 10 speed (both parameters with a high uncertainty regarding future trends, Rhein et al., 2013), 11 and might also be linked to climatically driven changes in the state of the World´s oceans 12 (e.g., upwelling, acidity). All of this highlights the importance of a better understanding of 13 background natural oceanic biogeochemical processes in currently changing environments. has consequently decreased the lifetime of ozone in the marine atmosphere and its associated 5 RF, thus closing a negative feedback loop and presenting the ocean emissions of iodine as a 6 natural mitigating factor for anthropogenic RF in the marine environment (PD cycle in red). 7