Natural short-lived halogens exert an indirect cooling effect on climate

Observational evidence shows the ubiquitous presence of ocean-emitted short-lived halogens in the global atmosphere1–3. Natural emissions of these chemical compounds have been anthropogenically amplified since pre-industrial times4–6, while, in addition, anthropogenic short-lived halocarbons are currently being emitted to the atmosphere7,8. Despite their widespread distribution in the atmosphere, the combined impact of these species on Earth’s radiative balance remains unknown. Here we show that short-lived halogens exert a substantial indirect cooling effect at present (−0.13 ± 0.03 watts per square metre) that arises from halogen-mediated radiative perturbations of ozone (−0.24 ± 0.02 watts per square metre), compensated by those from methane (+0.09 ± 0.01 watts per square metre), aerosols (+0.03 ± 0.01 watts per square metre) and stratospheric water vapour (+0.011 ± 0.001 watts per square metre). Importantly, this substantial cooling effect has increased since 1750 by −0.05 ± 0.03 watts per square metre (61 per cent), driven by the anthropogenic amplification of natural halogen emissions, and is projected to change further (18–31 per cent by 2100) depending on climate warming projections and socioeconomic development. We conclude that the indirect radiative effect due to short-lived halogens should now be incorporated into climate models to provide a more realistic natural baseline of Earth’s climate system.

The climate significance of ocean-land-atmosphere gas exchange has primarily focused on the partitioning of greenhouse gases (for example, carbon dioxide (CO 2 ), methane (CH 4 ) and nitrous oxide (N 2 O)) and the release of biologically produced dimethyl sulfide (DMS), which forms aerosols through secondary oxidation reactions 9 . Less attention has been paid to natural sources of other reactive gases that, through altering the atmospheric oxidation capacity, have the potential to impact Earth's radiative balance and climate indirectly. One such group of reactive gases is the so-called short-lived halogen species (SLH; chlorine, bromine and iodine compounds with a lifetime of less than six months in the atmosphere). For the past two decades, observational evidence collected from around the world has shown the ubiquitous presence of SLH in the global atmosphere [10][11][12][13][14][15][16][17][18][19] . These species are naturally emitted from the oceans, polar ice and the biosphere 1,2 , presenting a variable spatio-temporal source strength that is expected to increase owing to climate change 20 . In addition, a recent rapid increase in anthropogenic emissions of chlorinated SLH has been identified in the atmosphere 7,8,21 .
The breakdown of SLH in the atmosphere yields highly reactive halogen radicals that play important roles in several atmospheric processes, including the depletion of ozone (O 3 ) through catalytic cycles, direct CH 4 chemical loss, alteration of the hydroxyl (OH) radical, and the hydrogen (HO 2 /OH) and nitrogen (NO 2 /NO) oxides balance 2 (see reactions R1-R17 in Supplementary Table 1). Halogens also oxidize oceanic DMS, influencing the formation of cloud condensation nuclei 22 , and in the case of iodine, higher iodine oxides and oxoacids have been shown to condense spontaneously to form ultrafine aerosols [23][24][25] . Combined, this large and growing body of research has demonstrated that natural plus anthropogenic SLH can exert a profound impact on the chemistry and composition of the atmosphere on a global scale. However, their effect on Earth's radiative balance remains almost unexplored.
Since their initial implementation in global chemistry-climate models 26,27 , the emissions and chemistry of SLH have revealed that they have the potential to substantially alter the oxidizing capacity of the atmosphere 28-31 , both over pristine and polluted environments. Atmospheric oxidation in turn determines the abundance of short-lived climate forcers (SLCF) such as CH 4 , tropospheric O 3 and aerosols, which are key contributors to climate warming 32-36 . In particular, SLH constitute a natural buffer to anthropogenic O 3 pollution owing to Article a negative feedback mechanism that regulates natural emissions of iodine 37,38 , as well as modulating CH 4 lifetime via direct and indirect chemical loss processes 39 . In addition, SLH affect the evolution of O 3 in the climate-relevant lower stratosphere 21,40 . So far, climate models used in international climate assessments, such as the Coupled Model Intercomparison Project Phase 6 (CMIP6) 41 and the different assessment reports of the Intergovernmental Panel on Climate Change (IPCC) 32,35 , have not included the sources and chemistry of SLH. Here we use a state-of-the-art Earth-system model to quantify the contribution of SLH to the global energy balance across pre-industrial, present-day and future climates. Our results show that natural SLH exert an indirect cooling effect on the climate system and that this natural cooling effect has been amplified since pre-industrial times owing to anthropogenic activity.

Net radiative effect
We use the Community Earth System Model (CESM; see 'CESM (CAM-Chem) model configuration and experiments design' in Methods) to quantify the influence of SLH on the global radiative balance (see Extended Data Table 1 for modelling cases). The radiative effect (RE) caused by halogen radicals on the main SLCF, namely O 3 , CH 4 , aerosols and stratospheric water vapour are computed for past, present and future climate scenarios ('The RRTMG radiation module in CESM' in Methods). Sources of SLH are grouped into three categories: natural (NAT), anthropogenically amplified natural emissions (AANE) and anthropogenic (ANT). For pre-industrial simulations (year 1750), we consider only natural sources, mainly emitted from the oceans (for example, hypoiodous acid (HOI) and bromoform (CHBr 3 )) and polar regions (for example, bromine monochloride (BrCl) and molecular chlorine (Cl 2 )) by biogenic and photochemical processes (NAT). For present-day (2020) and future (2100) simulations, anthropogenic pollution has important impacts on global SLH emissions, including (Extended Data Table 2): (1) the direct emissions of inorganic (for example, hydrogen chloride (HCl)) and organic (for example, dichloromethane (CH 2 Cl 2 )) SLH from anthropogenic activities (ANT; for example, industrial, coal burning, waste burning and so on); and (2) anthropogenic emissions of primary pollutants (for example, nitrogen oxides and volatile organic compounds from transport, shipping, industry, power plants and so on), which subsequently form secondary air pollutants (for example, O 3 and nitric acid) that drive the anthropogenic amplification of natural SLH emissions (AANE; Methods). Figure 1 shows that present-day natural and anthropogenic SLH exert a net (gas + aerosols) indirect cooling effect of −0.13 ± 0.03 W m −2 for all-sky conditions (see Extended Data Fig. 1 for results distinguishing clear-sky, clouds and aerosol-cloud contributions). This value is the result of the distinct halogen-mediated radiative impact on O 3 (−0.24 ± 0.02 W m −2 ), CH 4 (+0.09 ± 0.01 W m −2 ), aerosols (+0.03 ± 0.01 W m −2 ) and stratospheric water vapour (+0.011 ± 0.001 W m −2 ). A comprehensive analysis of model uncertainty and results dependence on SLH burden is provided in Supplementary Information and summarized in Extended Data Table 3. We now detail the influence of SLH on each of the main chemically active SLCF.

Ozone
Halogen radicals efficiently destroy atmospheric O 3 through catalytic cycles 1 Table 4), leading to a total reduction in the O 3 RE of −0.16 ± 0.01 W m −2 (Extended Data Table 5). The corresponding changes in present-day tropospheric and stratospheric O 3 are −4.9 DU and −5.2 DU, respectively, which induce a net RE reduction of −0.24 ± 0.02 W m −2 . By the end of the century, projected O 3 RE is −0.19 ± 0.01 W m −2 (total O 3 loss of −8.5 DU) for the representative concentration pathway 6.0 (RCP6.0) scenario and −0.24 ± 0.02 W m −2 (−10.7 DU) for RCP8.5.

Methane
Tropospheric O 3 is the principal source of OH, the main atmospheric oxidant and the dominant chemical sink of CH 4 , which is the second-largest greenhouse gas after CO 2  an RE enhancement of +0.09 ± 0.01 W m −2 during both time periods. The greater burden and RE of CH 4 associated with SLH is the result of the indirect halogen-driven decrease in CH 4 oxidation by OH outweighing the direct increase in CH 4 loss by chlorine atoms 39 . By 2100, halogen-induced CH 4 RE is +0.10 ± 0.01 W m −2 for RCP6.0 and +0.11 ± 0.01 W m −2 for RCP8.5, resulting from burden increases of 464 Tg (11%) and 936 Tg (7%), respectively, compared with the corresponding future scenario omitting SLH (Extended Data Table 4).

Stratospheric water vapour
In the troposphere, water vapour is regulated by the local environment (for example, temperature, dew point and so on). However, in a predominantly dehydrated stratosphere, the chemical oxidation of CH 4 influences the stratospheric water vapour abundance. The chemistry of CH 4 in the lower stratosphere is similar to that in the troposphere, with OH radicals oxidizing CH 4 in the same manner (reaction R16 in Supplementary Table 1). As described above, SLH increase the CH 4 burden and thus stratospheric water vapour, leading to a warming RE in the stratosphere of +0.011 ± 0.001 W m −2 (Extended Data Table 5). The relative contribution of halogen-driven water vapour RE in the future stratosphere is similar to that at present (Fig. 1).

Aerosols
The aforementioned halogen impacts on atmospheric oxidants (OH radicals, O 3 , chlorine and so on) also lead to changes in the formation of secondary aerosols (aerosols formed following the oxidation of directly emitted gaseous precursors), including sulfate SO 4 2− , secondary organic aerosols (SOA) and ammonium nitrate (NH 4 NO 3 ; see reactions R18-R21 in Supplementary Table 1) 43 . It is noted that all these aerosol species present a dominant cooling effect in the troposphere owing to the reflection of solar incoming shortwave radiation, and the inclusion of SLH results in a reduction of this cooling effect by decreasing aerosol formation on the global scale. The estimated impact of halogens on aerosol RE reaches +0.03 ± 0.01 W m −2 for both pre-industrial and present-day conditions (see Methods for the contribution of individual aerosol species). Although a recent focus of research 43-45 , it is noted that large uncertainties still remain on the contribution of halogens to the global secondary aerosol loading.
In summary, natural changes in atmospheric composition mediated by SLH during pre-industrial times lead to a significant reduction in O 3 RE (−0.16 ± 0.01 W m −2 ), a relatively small increase in stratospheric water vapour RE (+0.011 ± 0.001 W m −2 ), a noticeable enhancement in the CH 4 RE (+0.09 ± 0.01 W m −2 ) and a slight increase in the RE from aerosols (+0.03 ± 0.01 W m −2 ; mostly due to sulfate reduction; Fig. 1 and Extended Data Table 5). The net pre-industrial RE is estimated to be −0.08 ± 0.02 W m −2 , with a dominant contribution from gaseous species (−0.11 ± 0.02 W m −2 ). In comparison, the SLH-driven reduction in net RE is stronger at present (−0.13 W m −2 versus −0.08 W m −2 ) because: (1) the inorganic halogen tropospheric burden is larger than in pre-industrial (147-187% for Cl y , 8-9% for Br y and 24-29% for I y ; Extended Data  . It is noted that the CH 4 RE reaches a maximum within the low latitudes resulting in net heating, whereas the O 3 radiative cooling is more prominent over the high latitudes.
The aerosol RE reaches a maximum over the Southern Ocean owing to the OH reduction caused by SLH, presenting spatial hotspots over industrialized regions such as Europe, North America and East Asia during the present day. The spatially resolved RE for the RCP6.0 and RCP8.5 scenarios is shown in Extended Data Fig. 2 and the radiative contribution for individual aerosol species is shown in Extended Data Fig. 4. All maps and elements were created by our research group using Matplotlib Basemap for Python.

Article
to anthropogenic activity (Extended Data Table 4). The present-day greater abundance of tropospheric reactive halogens largely responds to the anthropogenic amplification of natural emissions (AANE) over the oceans, which dominates the change in halogen sources and burden over the direct continental emissions of anthropogenic halogens (ANT; Extended Data Table 2).

Spatial distribution of radiative effect
SLH are emitted from various sources around the globe with large spatial heterogeneity. The dominant natural sources arise from the ocean whereas the main anthropogenic sources are located over continental regions (Supplementary Information). Figure 2 shows that the SLH-mediated RE during the pre-industrial and the present day is most noticeable over the open ocean and polar regions where natural halogens are emitted by seawater, sea-salt aerosols, first-year sea-ice and blowing snow (see Extended Data Fig. 2 for future scenarios). The large SLH-driven RE within high latitudes is mainly due to both tropospheric and stratospheric O 3 changes, with a much smaller contribution from CH 4 ; whereas over the low latitudes, opposite contributions from O 3 and CH 4 almost cancel out (Fig. 3). Indeed, SLH have been shown to increase the depth and size of the stratospheric ozone hole over Antarctica during austral spring 40,46 , which further enhances the cooling effect of halogens in the lower stratosphere over the Southern Hemisphere during the present day compared with the pre-industrial scenario (see 'Additional aspects of SLH influence on SLCF' in Methods). Hence, the cooling effect of SLH peaks at high latitudes for all climate scenarios, that is, within the Earth regions that are predicted to be most affected by global warming 47,48 .
SLH also lead to a reduction in aerosol formation and a subsequent warming on the global scale, mostly driven by the reduction in tropospheric OH abundance caused by halogens (Extended Data    Fig. 4). This includes highly localized cooling effects over China (Extended Data Fig. 4b), which are consistent with the SLH-driven enhancement in aerosol haze pollution 43 (see 'Additional aspects of SLH influence on SLCF' in Methods).

Change relative to pre-industrial times
We now quantify the present and future changes in the RE of active SLCF relative to the pre-industrial climate (ΔRE; see 'The RRTMG radiation module in CESM' in Methods), evaluating the contribution and time evolution of AANE compared with ANT. Figure 4 shows that the combined ΔRE due to SLH is −0.05 ± 0.03 W m −2 at present, of which about 30% is due to ANT and about 70% is due to AANE. Changes in ΔRE are more uncertain towards the future: the SLH-mediated ΔRE by 2100 is −0.01 ± 0.03 W m −2 for RCP6.0 (about 51% for ANT and about 49% for AANE) and −0.02 ± 0.03 W m −2 for RCP8.5 (about 17% for ANT and about 83% for AANE). It is noted that the SLH-mediated radiative changes from the pre-industrial to present and future are mostly driven by AANE, that is, natural halogen emissions that are amplified by anthropogenic perturbations (Extended Data Fig. 1). Thus, the increase in the SLH indirect cooling effect since pre-industrial times is not a result of direct anthropogenic emissions but instead indirectly arises from the amplification of natural halogen emissions owing to human activities, and the subsequent effects of these emissions on various SLCF. The main driver of AANE is the anthropogenic increase in O 3 pollution and its subsequent deposition onto the ocean surface 38,49 that has amplified, by a factor of two to three, the oceanic emission of iodine since the mid-twentieth century, as evidenced by measurements in Arctic and Alpine ice cores and tree rings [4][5][6] . The presence of anthropogenic air pollutants (for example, strong acids) also affects the partitioning of reactive halogen species and their heterogeneous recycling on sea-salt

Fig. 5 | Conceptual representation of the SLH influence on atmospheric composition and radiative feedbacks within the climate system.
Halogens influence the climate system through direct changes in O 3 and OH radical chemical cycling, which in turn regulate the abundance of radiatively active SLCF such as CH 4 , aerosols and stratospheric water vapour (H 2 O). The widening (thinning) of the semi-circular arrows within the chemical process layer represents an enhancement (reduction) of the efficiency of the direct SLH-driven (light blue) and indirect OH-driven (dark blue) chemical recycling of CH 4 , H 2 O and O 3 . The green, grey and black upwards arrows within the precursor's layer are the direct emissions of natural SLH, anthropogenic SLH and anthropogenic air pollutants, respectively. The U-shaped arrows show natural atmospheric cycling processes of halogenated (greenish tail) and anthropogenic (greyish tail) chemical reservoirs, respectively, both of which have been anthropogenically amplified (orange head) and altered the baseline state of the climate system. The length variation of the curly yellow and pink arrows on the uppermost SLCF layer represents the effect induced by SLH on Earth's radiative balance. The individual warming and cooling effect of each individual SLCF, as well as the net SLH-driven cooling RE, are synthesized as coloured thermometers. Figure 5 was created by NorArte Visual Science (https://www.norarte.es/en/) upon request.
Article aerosols and blowing snow, which perturbs the release of gaseous bromine and chlorine to the atmosphere 2 .
The breakdown of ΔRE shows that the relative contributions of individual SLCF are of opposite signs and compensate each other to result in a net cooling effect ( Fig. 4 and Extended Data Fig. 5). For instance, AANE dominates ΔRE for CH 4 at the end of the century regardless of the scenario considered, which responds to the anthropogenic amplification of iodine and bromine emissions from the global oceans 38 , which significantly reduce the levels of OH radicals and in turn CH 4 oxidation 39 . In contrast, during the present day, the relative contributions of AANE and ANT to CH 4 ΔRE are comparable (Fig. 4). It is noted that most of the ΔREdriven by CH 4 occurs in the lower troposphere, whereas for O 3 , a significant ΔRE contribution also occurs in the lower stratosphere, where in addition to the natural SLH changes, the rapid increase in the anthropogenic emissions of short-lived chlorocarbons also contributes to O 3 depletion 21,28 (Extended Data Table 4). Thus, ΔRE for O 3 from the pre-industrial to the present has a significant contribution from pure anthropogenic sources (about 26% ANT compared with about 74% AANE), whereas it is projected to reduce to about 17% ANT (about 83% AANE) for RCP6.0 and about 6% ANT (about 94% AANE) for RCP8.5 by the end of the century. Similarly, the present-day ΔRE for CH 4 relative to the pre-industrial period is attributed to about 42% ANT and about 58% AANE, whereas by the end of the century AANE dominates the signal (about 87% AANE under RCP6.0 and about 96% AANE for RCP8.5). The changing radiative effect of SLH across pre-industrial, present-day and future climates highlights the complex nonlinear chemical interaction between SLH and the abundance of key chemically active SLCF.

Radiative influence of SLH on climate
SLH are naturally emitted from the oceans, ice and aerosol surfaces, as well as from the biosphere and anthropogenic activities. Their natural emissions are strongly linked to climate (for example, sea surface temperature, primary productivity, lifting of sea-salt aerosols by winds and sea-ice extent) and to anthropogenic pollution (O 3 deposition to the ocean and atmospheric acidification) 37,39 . In addition, anthropogenic SLH not controlled by the Montreal Protocol have shown a rapid increase over East Asia and other developing regions during the past decade 7,8,21 . This changing role of SLH in controlling the oxidizing capacity of the troposphere and, consequently, in regulating the abundance of radiative-active SLCF, together with the anthropogenic amplification of the natural SLH emissions (AANE), affects the baseline radiative budget of the atmosphere in different ways (Fig. 5). Therefore, past and future changes in halogen emissions, and their indirect effect on Earth's radiative balance through altering the oxidative capacity, are determined by a combination of natural and anthropogenic emissions, climate variability and atmospheric chemistry.
The addition of present-day anthropogenic halogen emissions on top of AANE induces a slight change in global net RE (−0.11 ± 0.03 W m −2 for only AANE and −0.13 ± 0.03 W m −2 for AANE + ANT). Indeed, the relative contribution of AANE to the total halogen effect further increases in the future, regardless of considering a mid-or high-emissions scenario ( Fig. 4 and Extended Data Table 5). This highlights that amplified natural halogen sources (AANE), which cannot be directly controlled by environmental agreements but whose emissions depend on the emissions of anthropogenic pollutants that can be regulated, dominate the global SLH effect on the climate system. The analysis demonstrates that the SLH-driven RE is a persistent and significant signal during all time periods, with variable uncertainties dominated by the predicted levels of tropospheric halogens within each scenario.
The halogen impacts on RE have a marked geographical distribution. Noticeably, given the larger RE influence of SLH at high latitudes (Fig. 3), the inclusion of SLH is expected to alter the atmospheric heat redistribution from the equatorial regions to the high latitudes, that is, maximizing the cooling effect of halogens over the polar regions, which are expected to suffer the largest temperature enhancements owing to global warming 47,48 .
Finally, we highlight that the net indirect cooling effect caused by SLH is the result of a trade-off between the spatially variable effects of halogens mainly on O 3 (both tropospheric and stratospheric) and CH 4 , with a minor contribution from aerosols and stratospheric water vapour. This so far unrecognized interplay between natural SLH and Earth's radiative balance is nonlinear across pre-industrial, present-day and future climates. Models that do not include this indirect RE may overestimate the warming induced by SLCF since pre-industrial times. Furthermore, our results show that the net cooling effect of halogens has been amplified since pre-industrial times owing to the linkage between halogen emissions and atmospheric pollutants, and this complex interplay is expected to further change depending on future climate projections. The forcing caused by SLH over the industrial era (−0.05 W m −2 ) is similar to that produced by the increase of dust emissions (−0.07 W m −2 ) 50 and of equivalent magnitude but opposite sign as the combined contrail and contrail-induced cirrus forcing (0.06 W m −2 ) 32 . We conclude that SLH are a key component of the natural climate system as they exert an indirect cooling effect currently not accounted for in climate model assessments and, therefore, we suggest the need to include a complete representation of natural and anthropogenic SLH in climate models to reduce uncertainties in the contribution of SLCF to the evolution of Earth's radiative balance from pre-industrial to future climates.

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-023-06119-z. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

CESM (CAM-Chem) model configuration and experiments design
The Community Earth System Model (CESM) version 1. 1.1 (ref. 51), including the Community Atmospheric Model with interactive chemistry (CAM-Chem) version 4.0 (ref. 52), was used to quantify the overall impact of SLH on Earth's energy balance from pre-industrial times to the end of the twenty-first century. The model was configured with a horizontal resolution of 1.9° latitude × 2.5° longitude (96 × 144 grid points, respectively) and 26 vertical levels that extend from the surface to approximately 40 km (3.5 hPa in the upper stratosphere), following a hybrid sigma pressure coordinate 53 .
The standard chemical scheme implemented in CAM-Chem includes 169 species with both gas-phase and heterogeneous reactions coupled to the radiation module 54 . Updates for the chemical processing of SLH include a state-of-the-art chemical mechanism for halogens in the troposphere and the stratosphere, which has been described in detail in previous studies. Briefly, ref. 26 presented the implementation of reactive halogen species sources and chemistry in CAM-Chem, including a comprehensive validation of halocarbon source gases using available observations. References 30,38,55,56 then further updated the halogen CAM-Chem set-up to include a more detailed representation of chlorine, bromine and iodine gas-and heterogeneous-phase chemistry, which allowed to quantify the influence of SLH on stratospheric O 3 (ref. 40). A polar module, including inorganic halogen sea-ice emissions from the Arctic and Antarctica 57 , as well as the impact of halogens on CH 4 lifetime and burden 39 , have also been implemented into the current SLH version of CAM-Chem. Furthermore, here we have implemented and improved a few additional model developments: (1) the OH/O 3 /NO 3 -initiated SOA production yield was updated 58 ; (2) chlorine-and bromine-induced formation of SOA was considered 43 ; we added (3) the HOBr + S reaction to form sulfate aerosol following refs. 59,60, as well as (4) the heterogeneous recycling of bromine species on anthropogenic aerosols 43,61 ; and we also included (5) a consistent representation of iodine-containing particle formation from higher iodine oxides 25,62 and (6) the injection of gas-phase and particulate iodine to the stratosphere 13,46 . The main reactions of relevance for this work are summarized in Supplementary Table 1; for a full set of halogen reactions implemented in CAM-Chem, see the supplementary material in ref. 63.
Natural SLH sources in CESM (CAM-Chem) include both biogenic and abiotic pathways (Fig. 5). Biogenic sources comprise nine halocarbons (CHBr 3 , CH 2 Br 2 , CH 2 BrCl, CHBr 2 Cl, CHBrCl 2 , CH 3 I, CH 2 I 2 , CH 2 IBr and CH 2 ICl), which are the result of micro-and macro-algae as well as phytoplankton metabolism coupled to photochemistry at the ocean's surface 26 . The evolution of these SLH biogenic emissions is treated in a consistent framework in which they are coupled to physical and biogeochemical changes (for example, sea surface temperature, marine primary production and so on) related to climate and atmospheric composition 37 . Abiotic sources have distinct routes for chlorine and bromine compared with iodine. Chlorine and bromine are released from sea-salt aerosols following acid displacement (for example, induced by HNO 3 ) as well as heterogeneous reactions of nitrogenated (for example, N 2 O 5 ), halogenated (for example, HOBr, HOCl and HOI) and halo-nitrogenated (for example, BrONO 2 , ClONO 2 and IONO 2 ) reservoirs, constituting the dominant sources of reactive bromine and chlorine in the lower troposphere 55,64-66 . Inorganic iodine (HOI and I 2 ), however, is directly emitted from the ocean surface following O 3 deposition on seawater and its reactions with aqueous iodide 38,49,67 . Emissions of bimolecular inter-halogen species (that is, Cl 2 , Br 2 and I 2 , as well as BrCl, IBr and ICl) from the sea-ice surface within the Arctic and Antarctica are also computed online 57 (Fig. 5).
Anthropogenic SLH sources are included following an emission inventory of the two dominant organic chlorine species (CH 2 Cl 2 and C 2 Cl 4 ) (ref. 68), complemented by lower boundary conditions of other anthropogenic chlorinated substances (CHCl 3 , C 2 H 4 Cl 2 and C 2 HCl 3 ) (refs. 39, 64,69). In this study, we further implemented an anthropogenic global emission inventory of reactive inorganic halogen species for the year 2014 (applied to present-day conditions), including inorganic chlorine (HCl and fine particle chloride) from coal burning, biomass burning and waste burning, as well as inorganic bromine (HBr and Br 2 ) and iodine (HI and I 2 ) from coal burning (see further details in 'Emission inventory of global anthropogenic inorganic halogens'). Extended Data Table 2 and Supplementary Figs. 1-3 show the contribution of natural and anthropogenic emissions to the atmospheric halogen budget, and Extended Data Table 3 summarizes the surface mixing ratios and tropospheric burden for total inorganic chlorine (Cl y ), bromine (Br y ) and iodine (I y ) for the natural (AANE) and full (AANE + ANT) simulations during the pre-industrial, the present-day and the end of the century. Supplementary Figs. 5-7 show the geographical and vertical distributions of Cl y , Br y and I y .
The standard CESM (CAM-Chem) anthropogenic pollutant emissions developed for the Chemistry-Climate Intercomparison Project (CCMI) 70  The model set-up is based on specified dynamic time-slice simulations considering three distinct periods: pre-industrial times, representative of the year 1750; present-day conditions for the year 2020; and future conditions at the end of the century (year 2100) for two different projected scenarios (see below). Time-slice simulations for each period comprise 15-year integrations driven by nudging every 3 h a varying meteorology (temperature, winds and surface pressure) from a previous simulation that omitted the contribution of SLH 30, 55,56 . It is noted that even though the meteorology was obtained considering mean climatological boundary conditions representative of 2000-2020, pre-industrial, present-day and future sensitivities considered sea surface temperature and sea-ice conditions representative of each time period 37 . All experiments were initialized from a previous simulation after allowing 40 years of spin-up to ensure all chemical species, particularly CH 4 , were stabilized. Beyond historical periods, future projections are based on the mid-and high-end RCP (RCP6.0 and RCP8.5, respectively) emissions scenarios 74,75 for both long-lived species and short-lived precursor emissions. Long-lived halogen-containing species (CH 3  Benchmark model simulations for all time periods are split into three categories with distinctive treatment of SLH (Extended Data Table 1): (1) NoSLH: standard chemical scheme without SLH sources and chemistry; (2) NAT/AANE: only natural SLH emissions scenario (NAT for pre-industrial; AANE for present-day and future scenarios where the online computation of natural SLH emissions have been anthropogenically amplified); and (3) AANE + ANT: anthropogenically amplified natural emissions plus anthropogenic SLH sources for the present day and future. It is noted that anthropogenic and biogenic emissions other than SLH are identical within the NoSLH, AANE and AANE + ANT scenarios. Given that no anthropogenic SLH emissions are considered for pre-industrial runs, NAT represents pristine background halogen conditions, whereas AANE represents perturbed halogen conditions owing to anthropogenic air pollutants affecting the SLH natural source strength, particularly via the abiotic route of O 3 deposition on the ocean surface, the acid enhancement of sea-salt recycling and the biotic route of SLH emissions due to changes in climate ( Fig. 5 and Extended Data Table 2) 38,39,77 . The difference in the radiation budget between AANE and NoSLH represents the RE driven by the natural amplification of the halogen burden owing to the background levels of pollutants during a fixed period of time; whereas the difference between AANE + ANT and NoSLH represents the RE of all reactive halogen species.

The RRTMG radiation module in CESM
In this work, we distinguish between the terms radiative effect (RE) and the change in radiative effect (ΔRE): RE (Fig. 1) is the change in the radiative balance between a simulation considering SLH with respect to a baseline simulation omitting SLH, both during the same time period; whereas ΔRE ( Fig. 4) is the change in RE between different time periods (for example, between present day and pre-industrial times). RE and ΔRE were computed using the Rapid Radiative Transfer Model for Global circulation models (RRTMG) package 78 , which is currently the default radiative transfer scheme included in CESM v2 (ref. 79). The RRTMG radiation module provides an online diagnostic tool to quantify and distinguish the downwards and upwards as well as shortwave and longwave radiation at various layers, including the surface and top of the model 80,81 . In particular, RRTMG allows splitting the individual radiative contribution for independent radiatively active constituents, which can be added or subtracted one by one to or from the complete radiative components list (for example, considering the single-addition and single-subtraction contribution of each species to the total radiative budget; see ref. 80). Radiative magnitudes shown in this work were obtained considering the 15-year global mean for each individual benchmark configurations and the 5-year mean for the complete set of sensitivities described in Supplementary Information. The RE uncertainty associated with each independent simulation represents the interannual variability computed as two times the standard deviation (2σ) of the multi-year global average.
Here we use the RRTMG diagnosis variables FSNT (net solar flux at top of model) and FLNT (net longwave flux at top of model) for all-sky conditions, as well as their equivalent streaming for clear-sky conditions (FSNTC and FLNTC, respectively). Individual values of all magnitudes were obtained for the following list of radiatively active climate forcers (CESM name-list variables included in parenthesis): (1) gases: water vapour (H 2 O), carbon dioxide (CO 2 ), nitrous oxide (N 2 O), ozone (O 3 ), methane (CH 4 ) and chlorofluorocarbons (CFC12 and CFC-11STAR, which includes the contribution from CFC11 plus other minor CFCs and HCFCs); and (2) aerosols: sulfate (SO 4 ), dust (DST01-04), black carbon (CB1 and CB2), organic carbon (OC1 and OC2), secondary organic aerosols (SOAM, SOAI, SOAT, SOAB and SOAX), ammonium nitrate (NH 4 NO 3 ) sea salt (SSLT01-04) and iodine particles (IOP). The RE for individual gas-and aerosol-phase species, as well as that resulting from the sum of all gases, aerosols and the net (gas + aerosols) effect of each radiatively active species (represented by S) for each emission sensitivity case (C; NAT, AANE, AANE + ANT) and period of time (T; pre-industrial, present day, and future RCP6.0 and RCP8.5), were computed as follows: (1) The change in the RE for a given time period with respect to preindustrial times is computed as follows (see equations (2) and (3) below). First, the RE for the AANE and AANE + ANT scenarios for each period of time (for example, respectively defined as S RE( ) PD AANE and S RE( ) PD AANE+ANT for present-day (PD) conditions) is computed relative to the NoSLH scenario. For the case of pre-industrial (PI), only the natural RE S (RE( ) ) PI NAT is considered. Second, we compute the change in RE during the present day S (∆RE( ) ) C PD−PI , always with respect to the RE obtained for the pre-industrial, and split the natural (AANE) with respect to the anthropogenic (ANT) contributions as follows: C RCP8.5−PI scenarios relative to pre-industrial times. It is noted that owing to the superposition of absorption bands of the different radiatively active species, the sum of the individual RE contribution of each species slightly differs from the net RE of all species combined 82 . Indeed, this difference depends on the consideration of a single-addition or single-subtraction analysis in the radiative computation, and can result in a non-zero RE contribution from non-reactive gases such as CO 2 . To minimize these overlapping differences, we computed the normalized RE for all species considering the 0.428 (single addition) and 0.572 (single subtraction) weighting factors provided in ref. 80, whereas for aerosols only, single-subtraction magnitudes were considered. The small nonlinearity on the radiative assignation of the net RE to individual SLCF (that is, −0.03 W m −2 ) is attributed to neglecting the contribution of rapid adjustments (that is, radiation-driven changes in land surface and tropospheric temperatures) as well as to the different longwave absorption of overlapping bands when individual species are added to or subtracted from the radiation name-list. For the particular case of stratospheric water vapour, we computed its RE as 12.5% of the CH 4 RE (that is, in the middle of the various estimates compiled in the IPCC Sixth Assessment Report) 32 .

Emissions inventory of global anthropogenic inorganic halogens
In this study, we further develop a global emissions inventory of reactive inorganic halogen species for the year 2014 (applied to present-day conditions), including inorganic chlorine (HCl and fine particle chloride) from coal burning, biomass burning and waste burning, as well as inorganic bromine (HBr and Br 2 ) and iodine (HI and I 2 ) from coal burning. Source strength estimates of these inorganic halogen sources are zeroed for pre-industrial conditions and scaled into the future based on the RCP6.0 or RCP8.5 evolution of anthropogenic sulfur dioxide and carbon monoxide from biomass burning 39 .
Within our global inorganic halogen inventory, country-level emissions are calculated using the emissions factor method, following the methodology used in previous studies 43, 83 . Briefly, for activity data, country-level coal consumption from power plants, industry and residential burning are obtained from the International Energy Agency (www.iea.org) database. Dry matter burned from forest, grassland, peat and agriculture waste are derived from the Global Fire Emission Database (www.globalfiredata.org). Waste burned in incineration plants or by open burning are obtained from official statistics or calculated based on ref. 84. For China, detailed local and county-level activity data are used. Emissions factors of gas-phase halogen species arising from coal burning are calculated based on halogen content in coal and removal efficiencies of air pollution control devices. The halogen content in coal is obtained from our previous studies 43, 83 , the United States Geological Survey database and other measurements 85,86 . The installation rates of different air pollution control devices are from the Tsinghua emissions database [87][88][89] for China and from the PKU-FUEL database for other regions (inventory.pku.edu.cn). Other parameters, such as release rates, removal efficiencies and other emissions factors, are described in detail in our previous studies 43, 83 . The proportions of emitted inorganic halogen species were set as 70% and 30% for HBr and Br 2 , and 95% and 5% for HI and I 2 , respectively 90,91 .
Supplementary Figs. 1-3 present the spatial distribution of anthropogenic halogen emissions in comparison with the oceanic natural SLH emissions implemented in CAM-Chem 37 . Hotspots for continental chlorine emissions are located in China, India, Southeast Asia and Africa, with a peak emission intensity larger than 1.0 × 10 −13 kg m −2 s −1 .
China and India are also the major emitters of anthropogenic bromine and iodine, with emission fluxes higher than 1.0 × 10 −13 kg m −2 s −1 in polluted areas. The global mean source strength for natural (NAT), anthropogenic (ANT) and AANE is compared in Extended Data Table 2.

Additional aspects of SLH influence on SLCF
Given the current uncertainties on the SLH-aerosol interaction over both polluted and pristine environments 43 , the net aerosol RE induced by SLH sources and chemistry presents the largest relative errors of all the SLCF considered in this work (Supplementary Information). Sulfate dominates the net RE of aerosols, reaching +0.036 ± 0.005 W m −2 for the pre-industrial and +0.030 ± 0.006 W m −2 for present-day conditions. Even though the SLH-induced NH 4 NO 3 RE is small at present (+0.004 ± 0.001 W m −2 ), this species showed a pre-industrial to present-day burden enhancement that is two times larger than that for sulfate (Extended Data Table 4). This means that during pre-industrial times, the larger halogen-driven changes in atmospheric oxidants affected mostly sulfate, which has a significant natural precursor. In contrast, during present-day and future scenarios, the SLH influence on NH 4 NO 3 is larger because of its dominant anthropogenic precursors. Consequently, the cooling effect of both sulfate and NH 4 NO 3 is weaker when natural halogens are considered. Regarding secondary organic aerosols, AANE drives a global reduction of their formation (owing to a less oxidative atmosphere), while localized ANT emissions of inorganic halogens over industrial regions can enhance secondary aerosol formation during haze pollution events 43 (Extended Data Fig. 4).
In the end-of-the-century future projections, the net RE induced by SLH is weaker than in the present time regardless of the emissions scenario considered (RE = −0.09 ± 0.03 W m −2 for RCP6.0 and RE = −0.10 ± 0.03 W m −2 for RCP8.5; Fig. 1 and Extended Data Table 5). However, the independent contributions of the individual gases altering the net radiative balance differ: under RCP6.0 and owing to the more stringent restriction on air pollutant emissions, the global tropospheric O 3 burden is reduced by the end of the century 37 , and consequently the SLH influence on O 3 RE is significantly weaker (RE = −0.19 ± 0.01 W m −2 ) compared with the present. In contrast, owing to the future increase in global CH 4 burden, its warming RE owing to SLH slightly increases with respect to the present (RE = +0.10 ± 0.01 W m −2 ). Under RCP8.5, the enhancement in tropospheric O 3 burden results in a similar net RE as in the present day (RE = −0.24 ± 0.02 W m −2 ), which is offset by the larger increase in CH 4 emissions projected under RCP8.5, resulting in a halogen-driven RE warming of +0.11 ± 0.01 W m −2 for CH 4 . This dichotomy in the opposite contribution of halogen-mediated O 3 and CH 4 REs under different climate scenarios (Fig. 4) highlights the nonlinear chemical interaction between SLH and SLCF 33 .
Regarding the spatial heterogeneity of the RE, it is noted that the SLH-mediated changes in atmospheric composition depend significantly on the chlorine, bromine and iodine distribution over both oceanic and continental domains ( Supplementary Figs. 5-7), which in turn shift the nonlinear atmospheric chemistry response in different ways (Extended Data Fig. 3). For instance, emissions of SLH in clean environments (for example, oceanic and polar) tend to reduce tropospheric O 3 , thereby leading to a reduction in atmospheric oxidation capacity (see reactions R3-R7 in Supplementary Table 1); whereas in polluted environments (for example, urban and industrial) SLH emissions can result in tropospheric O 3 formation 92 , which in turn enhances the oxidizing capacity (that is, increase in OH) on regional scales (see reactions R9-R15 in Supplementary Table 1). Consequently, during present-day and future simulations, where SLH coexist with high levels of air pollutants, the change in RE due to SLH over continental regions is more pronounced compared with pre-industrial times (Extended Data Fig. 4). Therefore, future research focused on the spatial and seasonal variability of the SLH-mediated RE is needed to improve our understanding of the evolution of the baseline Earth's radiative budget.
The SLH influence on O 3 is the highest in the lowermost stratosphere, presenting a pronounced latitudinal dependence 93 that increases towards the high latitudes, altering the O 3 budget exactly in the region where surface temperature and climate are most sensitive to O 3 perturbations 94 . Despite the well known influence of SLH on the Antarctic ozone hole 76 , approximately half of the additional stratospheric O 3 destruction driven by short-lived bromine over Antarctica during the present time corresponds to a baseline O 3 destruction on the global stratosphere 40 , a feature observed in our simulations during all time periods. This background additional O 3 destruction owing to SLH is larger at high latitudes compared with the low latitudes (Fig. 3) and is also observed for sensitivity simulations where polar halogen sea-ice emissions are turned off (Supplementary Information). In addition, as the efficiency of the natural bromine and iodine background on stratospheric O 3 depletion peaks during late spring and summer 40,46 , significant SLH-driven stratospheric O 3 cooling is observed also by the end of the century over the Arctic and Antarctica regardless of the continuous reduction of anthropogenic long-lived O 3 -depleting substances ( Fig. 2 and Extended Data Fig. 2). It is noted that our model configuration does not consider the dynamical feedbacks of stratospheric O 3 that have been shown to influence surface temperature and precipitation over the southern tip of South America 95 . Both chemical (production and mostly loss) and transport (stratosphere-to-troposphere exchange) processes are altered when SLH are included 63 , although a distinction of each independent contribution is outside the scope of this work. Finally, it is noted that the radiative changes driven by stratospheric water vapour are only due to the chemical contribution from CH 4 photochemistry in the stratosphere but, as all simulations were forced with the same meteorology, the results presented here do not account for the changes in tropopause temperature and/or dynamical features affecting the climate evolution of stratosphere-troposphere exchange.

Evaluation of CESM (CAM-Chem) performance
We have conducted our simulations on the basis of previous studies (see ' allowed reproducing aircraft observations in the tropical upper tropopause 56 , suggesting the occurrence of iodine-driven stratospheric O 3 depletion, which was later confirmed by ref. 13 Here we provide further evaluation of CESM results for CH 4 and O 3 in pre-industrial and present-day simulations. The previous reported level of CH 4 in the pre-industrial era is about 722 ppbv (parts per billion by volume) 35,100 , ranging from 697 ppbv over Antarctica to 759 ppbv over the Arctic based on ice-core observations 101 . Our simulated global average CH 4 for the pre-industrial NAT case is at similar levels (722 ppbv global, 703 ppbv for Antarctica and 745 ppbv for the Arctic), suggesting that our model set-up properly represents the pre-industrial CH 4 abundance. Reports on pre-industrial O 3 are sparse. Supplementary Table 10 summarizes the available observation reports of O 3 in the pre-industrial periods 102 . The average O 3 mixing ratio at various sites is about 10 ppbv, ranging from 6.2 ppbv to 14.4 ppbv. Our modelling results at the same locations as observations for pre-industrial conditions averaged to be about 20 ppbv in the NoSLH case (between 16.2 ppbv and 24.1 ppbv) and about 15 ppbv in the NAT case (between 11.5 ppbv and 18.9 ppbv). Although the NAT case still overestimates the uncertain pre-industrial data, a robust feature of our simulations is that the model bias is significantly reduced compared with the NoSLH case (Supplementary  Table 10), which supports and highlights the importance of considering SLH in climate models to improve the representation of pre-industrial O 3 abundance. It is worth noting that most current climate models tend to overestimate the low surface O 3 concentrations compared with these rather uncertain semi-quantitative observations performed during the late nineteenth century 102 .
For the present-day CH 4 evaluation, we used the surface monthly average CH 4 mixing ratio observations for the period 2000-2019 from the National Oceanic and Atmospheric Administration (NOAA) network 103 , which show a global average mixing ratio of 1,848 ppbv. Our CESM present-day results for both NoSLH and AANE + ANT represent reasonably well the global CH 4 , with global mean surface mixing ratios of 1,683 ppbv and 1,836 ppbv, respectively, considering the same grid points where the observations were made ( Supplementary  Fig. 10). Noticeably, SLH increased the simulated CH 4 surface mixing ratio by 153 ppbv (or about 9%), compared with the NoSLH case within these sampling sites, highlighting that the inclusion of SLH brings the simulated CH 4 levels closer to the NOAA observations. We used the monthly average surface O 3 data from 2000 to 2015 in the Tropospheric Ozone Assessment Report (TOAR) dataset 104 (https://toar-data. org/) to evaluate our present-day O 3 CESM (CAM-Chem) results for the different model configurations. Supplementary Fig. 11 shows that both the NoSLH and AANE + ANT cases reproduce the global mean and range of observed surface O 3 concentrations, although with some overestimation. The global O 3 average of the TOAR dataset reaches 27.7 ppbv, whereas that for NoSLH is 41.1 ppbv and that for AANE + ANT is 33.8 ppbv, suggesting that the inclusion of SLH results in a more realistic representation of global surface O 3 . It is noted that even though the NoSLH configuration of CESM (CAM-Chem) tends to overestimate present-day surface O 3 , our modelled tropospheric O 3 burden is on the lower edge of the group of chemistry-climate models participating in the CMIP6 activity 41, 63 . However, our model configuration including SLH sources and chemistry reduces the high model bias of CMIP6 models 41 , and therefore results in a closer agreement with satellite-and ozonesonde-derived products relevant to global atmospheric chemistry model evaluation 63,105 .
To determine a robust estimation of the main uncertainties of the radiative and climatic influence of SLH, we performed a comprehensive sensitivity analysis where the emission strength, recycling efficiency and/or chemical reactivity of chlorine, bromine and iodine, as well as the background abundance of SLCF during each period of time, were varied within the range of values based on the most recent literature. This 'standalone' sensitivity analysis is described in Supplementary  Information and relies on refs. 107-113. The analysis demonstrates that SLH induce a persistent and significant cooling signal during all time periods, with variable uncertainties dominated by the predicted levels of tropospheric halogens within each scenario. This SLH-driven RE is a robust signal for all scenarios considered, surpassing the estimated uncertainties related to the variable levels of tropospheric halogens and abundance of climate forcers for the different configurations. The uncertainty range for RE and ΔRE is computed considering the complete set of model sensitivities as described in Supplementary Information and summarized in Extended Data Table 5. Further research using other models and projected scenarios is required to shed light on the remaining unknowns related to the coupling between anthropogenic pollutant emissions and the SLH influence on the evolution of Earth's radiative balance.

Data availability
The data supporting this article, including the SLH chemical mechanism, model configuration files and post-processing scripts, are available at Mendeley Data (https://doi.org/10.17632/gb7695c4vy.2). The complete dataset and routines used in this study are available from the corresponding author on reasonable request.

Code availability
The benchmark CESM code is available from https://www.cesm.ucar. edu/models/releases. Administration of model updates related to SLH chemistry are maintained by NCAR engineers and will become available with the next CESM release. Fig. 1 | Radiative effect of SLH on short-lived climate forcers at the top of model. RE for the only AANE and AANE+ANT configurations are shown by empty and black-striped colored bars, respectively, for all-sky (a) and clear-sky (b) conditions. Results for the pre-industrial period are on the left and consider only natural halogen emissions, while the RE in year 2100 for RCP6.0 (light-grey shading) and RCP8.5 (heavy-grey shading) climate scenarios are shown on the right. The RE due to clouds and aerosol-cloud (Aer-Cld) interaction is shown on top of the net (gas+aerosol) effect on b. Comparison of only AANE and AANE+ANT results indicates that most of the RE due to SLH arise from the contribution of natural sources that have been anthropogenically amplified during present-day and end-of-the-century conditions. The uncertainty range for each species is computed as half of the difference between the maximum and minimum RE obtained for the complete set of model sensitivities for each individual time period (mean ± range/2) as described in the Supplementary Information (see Extended Data Table 5).