Editing IPCC:AR6/WGI/Chapter-6 (section)

== 6.4 SLCF Radiative Forcing and Climate Effects ==

<div id="h1-5-siblings" class="h1-siblings"></div>

The radiative forcing on the climate system introduced by SLCFs is distinguished from that of long-lived greenhouse gases (LLGHGs) by the diversity of forcing mechanisms for SLCFs, and the challenges of constraining these mechanisms via observations and of inferring their global forcings from available data. [[IPCC:Wg1:Chapter:Chapter-7|Chapter 7]] assesses the global estimates of effective radiative forcing (ERF) due to SLCF abundance changes. This section assesses the characteristics (e.g., spatial patterns, temporal evolution) of forcings, emissions-based SLCF forcings, climate response and feedbacks due to SLCFs relying primarily on results from CMIP6 models. Additionally, the ERFs for several aerosol-based forms of solar –radiation modification (SRM) are discussed in Section 6.4.6.

Forcing and climate response due to changes in SLCFs are typically estimated from global models that vary in their representation of the various chemical, physical and radiative processes (Box 6.1) affecting the causal chain from SLCF emissions to climate response (Figure 6.1). The AR5 noted that the representation of aerosol processes varied greatly in CMIP5 models and that it remained unclear what level of sophistication is required to properly quantify aerosol effects on climate ( [[#Boucher--2013|Boucher et al., 2013]] ). Since the AR5, [[#Ekman--2014|Ekman (2014)]] found that the CMIP5 models with the most complex representations of aerosol impacts on cloud microphysics had the largest reduction in biases in surface temperature trends. CMIP6-generation CCMs that simulate aerosol and cloud-size distributions better represent the effect of a volcanic eruption on lower atmosphere clouds than a model with aerosol-mass only ( [[#Malavelle--2017|Malavelle et al., 2017]] ). This highlights the need for skilful simulation of conditions underlying aerosol–cloud interactions, such as the distribution, transport and properties of aerosol species, in addition to the interactions themselves (Chapter 7). In advance of CMIP6, representations of aerosol processes and aerosol–cloud interactions in ESMs have generally become more comprehensive ( Meehl et al. , 2020; Gliß et al. , 2021; Thornhill et al. , 2021b ; see also [[IPCC:Wg1:Chapter:Chapter-1#1.5|Section 1.5]] ), with enhanced links to aerosol emissions and gas-phase chemistry. Many CMIP6 models (Annex II: Table AII.5) now simulate aerosol number size distribution, in addition to mass distribution, which is a prerequisite for accurately simulating number concentrations of CCN ( [[#Bellouin--2013|Bellouin et al., 2013]] ), while some CMIP6 models use prescribed aerosol optical properties to constrain aerosol forcing (e.g., [[#Stevens--2017|Stevens et al., 2017]] ). Hence, the range of complexity in aerosol modeling noted in CMIP5 is still present in the CMIP6 ensemble. Although simulated CCN have been compared to surface ( [[#Fanourgakis--2019|Fanourgakis et al., 2019]] ) and aircraft ( [[#Reddington--2017|Reddington et al., 2017]] ) measurements, with mixed results, the lack of global coverage limits confidence in the evaluations. Evaluations of AOD have been more wide-ranging (Section 6.3.5; [[#Gliß--2021|Gliß et al., 2021]] ) but are less relevant to aerosol–cloud interactions as they do not allow the evaluation of vertical profiles, aerosol-cloud overlap regions, aerosol type or number. Nevertheless, biases in simulated patterns and trends in AOD, alongside biases in cloud fractions ( [[#Vignesh--2020|Vignesh et al., 2020]] ), likely affect quantifications of the aerosol–cloud interactions.

In summary, CMIP6 models generally represent more processes that drive aerosol–cloud interactions than the previous generation of climate models, but there is only ''medium confidence'' that those enhancements improve their fitness for the purpose of simulating radiative forcing due to aerosol–cloud interactions because only a few studies have identified the level of sophistication required to do so. In addition, the challenge of representing the small-scale processes involved in aerosol–cloud interactions, and a lack of relevant model-data comparisons, does not allow a quantitative assessment of the progress of the models from CMIP5 to CMIP6 in simulating the underlying conditions relevant for aerosol–cloud interactions at this time.

<div id="6.4.1" class="h2-container"></div>

<span id="historical-estimates-of-regional-short-lived-climate-forcing"></span>
=== 6.4.1 Historical Estimates of Regional Short-lived Climate Forcing ===

<div id="h2-19-siblings" class="h2-siblings"></div>

The highly heterogeneous distribution of SLCF abundances (Section 6.3) translates to strong heterogeneity in the spatial pattern and temporal evolution of forcing and climate responses due to SLCFs. This section assesses the spatial patterns of the current forcing due to aerosols and their historical evolution by region.

In AR5, the ''confidence'' in the spatial patterns of aerosol and ozone forcing was lower than that for the global mean because of the large spread in the regional distribution simulated by global models, and was assessed as ''medium'' . The AR5 assessment was based on aerosol and ozone RFs, and aerosol ERFs (with fixed SSTs) from ACCMIP and a small sample of CMIP5 models ( [[#Myhre--2013|Myhre et al., 2013]] ; [[#Shindell--2013|Shindell et al., 2013]] ). For this assessement, the spatial distribution of aerosol ERF due to human-induced changes in aerosol concentrations over 1850–2014 is quantified based on results from a seven-member ensemble of CMIP6 ESMs including interactive gas and aerosol chemistry analysed in AerChemMIP. There is insufficient information to estimate the spatial patterns of ozone ERF from CMIP6, however, the spatial patterns in SLCF ERF are dominated by that from aerosol ERF over most regions (e.g., [[#Shindell--2015|Shindell et al., 2015]] ). The aerosol ERF includes contributions from both direct aerosol–radiation (ERFari) and indirect aerosol–cloud interactions (ERFaci; [[IPCC:Wg1:Chapter:Chapter-7#7.3.3|Section 7.3.3]] ), and is computed as the difference between radiative fluxes from simulations with time-evolving aerosol and their precuror emissions, and identical simulations but with these emissions held at their 1850 levels ( [[#Collins--2017|Collins et al., 2017]] ). Both simulations are driven by time-evolving sea surface temperatures (SSTs) and sea ice from the respective coupled model historical simulation, and therefore, differ from ERFs computed using fixed pre-industrial SST and sea ice fields ( [[IPCC:Wg1:Chapter:Chapter-7#7.3.1|Section 7.3.1]] ), but the effect of this difference is generally small ( [[#Forster--2016|Forster et al., 2016]] ). A correction for land surface temperature change ( [[IPCC:Wg1:Chapter:Chapter-7#7.3.1|Section 7.3.1]] ) is not available from these data to explicitly quantify the contribution from adjustments. The ESMs included here used the CMIP6 anthropogenic and biomass-burning emissions for ozone and aerosol precursors but varied in their representation of the natural emissions, chemistry and climate characteristics contributing to spread in the simulated concentrations (Section 6.3) and resulting forcings, partly reflecting uncertainties in the successive processes ( [[#Thornhill--2021b|Thornhill et al., 2021b]] ).

The geographical distribution of the ensemble-mean aerosol ERF over the 1850–2014 period is highly heterogeneous (Figure 6.10a) in agreement with AR5. Negative ERF is greatest over and downwind of most industrialized regions in the Northern Hemisphere and to some extent over tropical biomass-burning regions, with robust signals. The largest negative forcing occurs over Eastern Asia and Southern Asia, followed by Europe and North America, reflecting the changes in anthropogenic aerosol emissions in recent decades (Section 6.2). Positive ERF <sub></sub> over high albedo areas, including cryosphere, deserts and clouds, also found in AR5 and attributed to absorbing aerosols, are not robust across the small CMIP6 ensemble applied here. Regionally aggregated shortwave (SW) and longwave (LW) components of the aerosol ERF <sub></sub> exhibit similar large variability across regions (Figure 6.10b). The SW flux changes come from aerosol–radiation and aerosol–cloud interactions while the small positive LW flux changes come from aerosol–cloud interactions (related to liquid-water path changes ( [[IPCC:Wg1:Chapter:Chapter-7#7.3.2.2|Section 7.3.2.2]] ). These spatial patterns in aerosol ERF are similar to the patterns reported in AR5.

<div id="_idContainer033" class="Basic-Text-Frame"></div>

[[File:631a947caeba367e8cec53a4eed8186a IPCC_AR6_WGI_Figure_6_10.png]]

'''Figure 6.10 |''' '''Multi-model mean effective radiative forcings (ERFs) over the recent-past (1995-2014) induced by aerosol changes since 1850.''' Panel '''(a)''' shows the spatial distribution of the net ERF with area-weighted global mean ERF shown at the lower right corner. Uncertainty is represented using the advanced approach: no overlay indicates regions with robust signal, where ≥66% of models show change greater than variability threshold and ≥80% of all models agree on sign of change; diagonal lines indicate regions with no change or no robust signal, where &lt;66% of models show a change greater than the variability threshold; crossed lines indicate regions with conflicting signal, where ≥66% of models show change greater than variability threshold and &lt;80% of all models agree on sign of change. For more information on the advanced approach, please refer to the Cross-Chapter Box Atlas.1. Panel '''(b)''' shows the mean shortwave and longwave ERF for each of the 14 regions defined in the Atlas. Violins in panel (b) show the distribution of values over regions where ERFs are significant. ERFs are derived from the difference between top of the atmosphere (TOA) radiative fluxes for Aerosol Chemistry Model Intercomparison Project (AerChemMIP) experiments ''histSST'' and ''histSST-piAer'' ( [[#Collins--2017|Collins et al., 2017]] ) averaged over 1995–2014 (Box 1.4, Chapter 1). The results come from seven Earth system models: MIROC6, MPI-I-ESM-1-2-HAM, MRI-ESM2-0, GFDL-ESM4, GISS-E2-1-G, NorESM2-LM and UKESM-0-LL. These data can be seen in the Interactive Atlas. Further details on data sources and processing are available in the chapter data table (Table 6.SM.3).

Time evolution of 20-year means of regional net aerosol ERF shows that the regions are divided into two groups depending on whether the mean ERF attains its negative peak value in the 1970s–1980s (e.g., Europe, North America) or in the late 1990s–2000s (e.g., Asia, South America; Figure 6.11). Qualitatively, this shift in the distribution of ERF trends is consistent with the regional long-term trends in aerosol precursor emissions (Section 6.2; Figures 6.18 and 6.19) and their abundances (Section 6.3). However, at finer regional scales, there are regions where sulphate aerosols are still following an upward trend (e.g., Southern Asia; Section 6.3.5) implying that the trends in ERF may not have shifted for these regions. The continental-scale ERF <sub></sub> trends are also in line with the satellite-observed AOD trends assessed in [[IPCC:Wg1:Chapter:Chapter-2#2.2.6|Section 2.2.6]] . Global mean ERF reaches maximum negative values in the mid-1970s and its magnitude gradually decreases thereafter. This weakening of the negative forcing since 1990 agrees with findings that attribute this to a reduction in global mean SO <sub>2</sub> emissions combined with an increase in global BC ( [[#Myhre--2017|Myhre et al., 2017]] ). Uncertainties in model-simulated aerosol ERF distribution and trends can result from inter-model variations in the representation of aerosol–cloud interactions and aerosol microphysical processes as also demonstrated by [[#Bauer--2020|Bauer et al. (2020)]] .

<div id="_idContainer035" class="Basic-Text-Frame"></div>

[[File:02d6b81f8579b92f37bfd2b1b4716097 IPCC_AR6_WGI_Figure_6_11.png]]

'''Figure 6.11 |''' '''Time evolution of 20-year multi-model mean averages of the annual area-weighted mean regional net effective radiative forcings (ERFs) due to aerosols''' '''for each of the 14 major regions in the Atlas, and global mean, using the models and model experiments as in Figure 6.''' '''10.''' Further details on data sources and processing are available in the chapter data table (Table 6.SM.3).

In summary, the spatial and temporal distribution of the net aerosol ERF from 1850–2014 is highly heterogeneous ( ''high confidence'' ). Globally, there has been a shift from increase to decrease of the negative net aerosol ERF driven by trends in aerosol and their precursor emissions ( ''high confidence'' ). However, the timing of this shift varies by continental-scale region and has not occurred for some finer regional scales.

<div id="6.4.2" class="h2-container"></div>

<span id="emissions-based-radiative-forcing-and-effect-on-global-surface-air-temperature-gsat"></span>
=== 6.4.2 Emissions-based Radiative Forcing and Effect on Global Surface Air Temperature (GSAT) ===

<div id="h2-20-siblings" class="h2-siblings"></div>

The ERFs attributable to emissions versus concentrations for several SLCFs including ozone and methane are different. A concentration change, used to assess the abundance - based ERF, results from The changes in emissions of multiple species and subsequent chemical reactions. The corollary is that the perturbation of a single emitted compound can induce subsequent chemical reactions and affect the concentrations of several climate forcers (chemical adjustments); this is what is accounted for in emissions-based ERF. Due to non-linear chemistry (Section 6.3) and non-linear aerosol–cloud interactions ( [[IPCC:Wg1:Chapter:Chapter-7#7.3.3.2|Section 7.3.3.2]] ), the ERF attributed to the individual species cannot be precisely defined and can only be estimated through model simulations. For example, the ERF attributed to methane emissions, which includes indirect effects through ozone formation and oxidation capacity with feedbacks on the methane lifetime, depend non-linearly on the concentrations of NO <sub>x</sub> , CO and NMVOCs. This means that the results from the model simulations depend to some extent on the chosen methodology. In AR5 (based on [[#Shindell--2009|Shindell et al., 2009]] ; [[#Stevenson--2013|Stevenson et al., 2013]] ) the attribution was done by removing the anthropogenic emissions of individual species one by one from a control simulation for present-day conditions. Further, only the radiative forcings, and not the ERF (mainly including the effect of aerosol–cloud interactions) were attributed to the emitted compounds.

Since AR5, the emissions estimates have been revised and extended for CMIP6 ( [[#Hoesly--2018|Hoesly et al., 2018]] ), the models have been further developed, the period has been extended (1750–2019, versus 1750–2011 in AR5) and the experimental setup for the model simulations has changed ( [[#Collins--2017|Collins et al., 2017]] ), making a direct comparison of results difficult. Figure 6.12a shows the global and annual mean ERF attributed to emitted compounds over the period 1750–2019 based on AerChemMIP simulations (Thornhill et al., 2021b) where anthropogenic emissions or concentrations of individual species were perturbed from 1850 to 2014 levels (methodology described in Supplementary Material 6.SM.1).

<div id="_idContainer037" class="Basic-Text-Frame"></div>

[[File:7ab95919189fd0eeeb59371a85fa3019 IPCC_AR6_WGI_Figure_6_12.png]]

Figure 6.12 | '''Contribution to effective radiative forcing (ERF) (a) and global mean surface air temperature (GSAT) change (b) from component emissions between 1750 to 2019 based on CMIP6 models (Thornhill et al.''' ''', 2021b).''' ERFs for the direct effect of well-mixed greenhouse gases (WMGHGs) are from the analytical formulae in section 7.3.2, H <sub>2</sub> O (strat) is from Table 7.8. ERFs for other components are multi-model means from [[#Thornhill--2021b|Thornhill et al. (2021b)]] and are based on ESM simulations in which emissions of one species at a time are increased from 1850 to 2014 levels. The derived emissions-based ERFs are rescaled to match the concentration-based ERFs in Figure 7.6. Error bars are 5–95% and for the ERF account for uncertainty in radiative efficiencies and multi-model error in the means. ERFs due to aerosol–radiation (ERFari) and cloud effects are calculated from separate radiation calls for clear-sky and aerosol-free conditions ( [[#Ghan--2013|Ghan, 2013]] ; [[#Thornhill--2021b|Thornhill et al., 2021b]] ). ‘Cloud’ includes cloud adjustments (semi-direct effect) and ERF from indirect aerosol-cloud to –0.22 W m <sup>–2</sup> for ERFari and –0.84 W m <sup>–2</sup> interactions (ERFaci). The aerosol components (SO <sub>2</sub> , organic carbon and black carbon) are scaled to sum to –0.22 W m <sup>–2</sup> for ERFari and –0.84 W m <sup>–2</sup> for ‘cloud’ ( [[IPCC:Wg1:Chapter:Chapter-7#7.3.3|Section 7.3.3]] ). For GSAT estimates, time series (1750–2019) for the ERFs have been estimated by scaling with concentrations for WMGHGs and with historical emissions for SLCFs. The time variation of ERFaci for aerosols is from Chapter 7. The global mean temperature response is calculated from the ERF time series using an impulse response function (Cross-Chapter Box 7.1) with a climate feedback parameter of –1.31 W m <sup>–2</sup> °C <sup>–1</sup> . Contributions to ERF and GSAT change from contrails and light-absorbing particles on snow and ice are not represented, but their estimates can be seen on Figure 7.6 and 7.7, respectively. Further details on data sources and processing are available in the chapter data table (Table 6.SM.3).

The ERF based on primary CO <sub>2</sub> emissions is slightly lower than the abundance-based estimate ( [[IPCC:Wg1:Chapter:Chapter-7#7.3.2.1|Section 7.3.2.1]] ) because the abundance-based ERF combines the effect of primary CO <sub>2</sub> emissions and a small additional secondary contribution from atmospheric oxidation of methane, CO, and NMVOCs (4%) of fossil origin, consistent with AR5 findings.

Ozone-depleting substances, such as N <sub>2</sub> O and halocarbons, cause a reduction in stratospheric ozone, which affects ozone and OH production in the troposphere through ultraviolet radiation changes (and thus affect methane). They also have indirect effects on aerosols and clouds ( [[#Karset--2018|Karset et al., 2018]] ), since changes in oxidants induce changes in the oxidation of aerosol precursors.

The net ERF from N <sub>2</sub> O emissions is estimated to be 0.24 [0.13 to 0.34] W m <sup>–2</sup> , which is very close to the abundance-based estimate of 0.21 W m <sup>–2</sup> [[IPCC:Wg1:Chapter:Chapter-7#7.3.2.3|Section 7.3.2.3]] ). The indirect contributions from N <sub>2</sub> O are relatively minor with negative (methane-lifetime) and positive (ozone-and-cloud) effects nearly compensating each other. Emissions of halogenated compounds, including CFCs and HCFCs, were assessed as very likely causing a net-positive ERF in the AR5. However, recent studies ( [[#Morgenstern--2020|Morgenstern et al., 2020]] ; [[#O’Connor--2021|O’Connor et al., 2021]] ; [[#Thornhill--2021b|Thornhill et al., 2021b]] ) find strong adjustments in Southern Hemisphere aerosols and clouds, such that the ''very likely'' range in the emission-based ERF for CFC + HCFCs + HFCs now also include negative values.

For methane emissions, in addition to their direct effect, there are indirect positive ERFs from methane enhancing its own lifetime, causing ozone production, enhancing stratospheric water vapour, and influencing aerosols and the lifetimes of HCFCs and HFCs ( [[#Myhre--2013|Myhre et al., 2013]] ; [[#O’Connor--2021|O’Connor et al., 2021]] ). The ERF from methane emissions is considerably higher than the ERF estimate resulting from its abundance change. The central estimate with the ''very likely'' range is 1.19 [0.81 to 1.58] W m <sup>–2</sup> for the emissions-based estimate <sup></sup> compared with 0.54 W m <sup>–2</sup> for the abundance-based estimate ( [[IPCC:Wg1:Chapter:Chapter-7#7.3.5|Section 7.3.5]] ). The abundance-based ERF estimate for methane results from contributions of its own emissions and the effects of several other compounds, some decreasing methane lifetime, notably NO <sub>x</sub> , which importantly reduce the methane abundance-based ERF. Emissions of CO and NMVOCs both indirectly contribute to a positive ERF through enhancing ozone production in the troposphere and increasing the methane lifetime. For CO and NMVOCs of fossil origin there is also a 0.07 W m <sup>–2</sup> contribution to CO <sub>2</sub> from their oxidation. The ''very likely'' total ERF of CO and NMVOCs emissions is estimated to be 0.44 [0.22 to 0.67] W m <sup>–2</sup> .

NO <sub>x</sub> causes a positive ERF through enhanced tropospheric ozone production and a negative ERF through enhanced OH concentrations that reduce the methane lifetime. There is also a small negative ERF contribution through the formation of nitrate aerosols, although only three of the AerChemMIP models include nitrate aerosols. The best estimate of the net ERF from changes in anthropogenic NO <sub>x</sub> emissions is –0.27 [–0.55 to 0.01] W m <sup>–2</sup> . The magnitude is somewhat greater than the AR5 estimate (–0.15 [–0.34 to +0.02] W m <sup>–2</sup> ) but with a similar level of uncertainty. The difference between AR6 and AR5 estimates is possibly due to the different modeling protocols (see Supplementary Material: 6.SM.1).

Anthropogenic emissions of SO <sub>2</sub> lead to the formation of sulphate aerosols and a negative ERF through aerosol–radiation and aerosol–cloud interactions. The emissions-based ERFaci, which was not previously considered in AR5, is now included. The estimated ERF is thus considerably more negative than the AR5 estimate with a radiative forcing of –0.4 W m <sup>–2</sup> , despite the decline of ERF due to aerosols since 2011 ( [[IPCC:Wg1:Chapter:Chapter-7#7.3.3.1.3|Section 7.3.3.1.3]] , Figure 6.12a). SO <sub>2</sub> emissions are estimated to contribute to a negative ERF of –0.94 [–1.63 to –0.25] W m <sup>–2</sup> , with –0.23 W m <sup>–2</sup> from aerosol–radiation interactions and –0.70 W m <sup>–2</sup> from aerosol–cloud interactions. Emissions of NH <sub>3</sub> lead to formation of ammonium-nitrate aerosols with an estimated ERF of –0.03 W m <sup>–2</sup> .

The best estimate for the ERF due to emissions of BC is reduced from the AR5, and is now estimated to be 0.11 [–0.20 to 0.42] W m <sup>–2</sup> with an uncertainty also including negative values. As discussed in [[IPCC:Wg1:Chapter:Chapter-7#7.3.3.1.2|Section 7.3.3.1.2]] , a significant portion of the positive BC forcing from aerosol–radiation interactions is offset by negative atmospheric adjustments due to cloud changes, as well as lapse rate and atmospheric water vapour changes, resulting in a smaller positive net ERF for BC compared with AR5. The large range in the forcing estimate stems from variation in the magnitude and sign of atmospheric adjustments across models and is related to the differences in the model treatment of different processes affecting BC (e.g., ageing, mixing) and its interactions with clouds and cryosphere ( [[IPCC:Wg1:Chapter:Chapter-7#7.3.3|Section 7.3.3]] ; [[#Thornhill--2021b|Thornhill et al., 2021b]] ). The emissions-based ERF for organic carbon aerosols is –0.21 [–0.44 to +0.02] W m <sup>–2</sup> , a weaker estimate compared with AR5 attributed to stronger absorption by OC ( [[IPCC:Wg1:Chapter:Chapter-7#7.3.3.1.2|Section 7.3.3.1.2]] ).

The emissions-based contributions to GSAT change (Figure 6.12b) were not assessed in AR5, but with the ERF from aerosol–cloud interactions attributed to the emitted compounds there is now a better foundation for this assessment. The contribution to emissions-based ERF at 2019 (Figure 6.12a) is scaled by the historical emissions (over the period 1750–2019) of each compound to reconstruct the historical time series of ERF. An impulse response function (Cross-Chapter Box 7.1, Supplementary Material 7.SM5.2) is then applied to obtain the contribution of SLCF emissions to the GSAT response. Due to the non-linear chemical and physical processes described above relating emissions to ERF, and the additional non-linear relations between ERF and GSAT, these emissions-based estimates of GSAT responses strongly depend on the methodology applied to estimate ERF and GSAT (Supplementary Material 6.SM.2). Therefore, the relative contribution of each compound through its primary emissions versus secondary formation or destruction (e.g., for methane emissions its ozone versus methane contributions), by construction (omitting the non-linear processes), will be equal for ERF and GSAT. Uncertainties in the GSAT response are estimated using the assessed range of the equilibrium climate sensitivity (ECS) from [[IPCC:Wg1:Chapter:Chapter-7|Chapter 7]] of this report. For most of the emitted compounds the uncertainty in the GSAT response is dominated by the uncertainty in the relationship between emissions and the ERF.

The contributions from the emitted compounds to GSAT broadly follow their contributions to the ERF, mainly because their evolution over the past decades have been relatively similar and slow enough compared to their lifetimes to be reflected similarly in their ERF and GSAT despite the delay of the GSAT response to ERF changes (Section 6.6.1). However, for some SLCFs (e.g., SO <sub>2</sub> ) that have been reduced globally, their contribution to GSAT change is slightly higher compared with that of CO <sub>2</sub> than their relative contribution to ERF because the peak in their ERF change has already occurred (Section 6.4.1) whereas the peak of their GSAT effect started to decline recently (Figure 7.9). This is due to the inertia of the climate system delaying the full response of GSAT to a change in forcing (Figure 6.15).

In summary, emissions of SLCFs, especially methane, NO <sub>x</sub> and SO <sub>2</sub> , have substantial effects on effective radiative forcing (ERF) ( ''high confidence'' ). The net global emissions-based ERF of NO <sub>x</sub> is negative and that of NMVOCs is positive, in agreement with the AR5 assessment ( ''high confidence'' ). For methane, the emissions-based ERF is twice as high as the abundance-based ERF ( ''high confidence'' ). SO <sub>2</sub> emissions make the dominant contribution to the ERF associated with the aerosol–cloud interaction ( ''high'' ''confidence'' ). The contributions from the emitted compounds to GSAT broadly follow their contributions to the ERF ( ''high confidence'' ). However, due to the inertia of the climate system delaying the full GSAT response to a change in forcing, the contribution to GSAT change due to SO <sub>2</sub> emissions is slightly higher compared with that due to CO <sub>2</sub> emissions (than their relative contributions to ERF) because the peak in emission-induced SO <sub>2</sub> eRF has already occurred.

<div id="6.4.3" class="h2-container"></div>

<span id="climate-responses-to-slcfs"></span>
=== 6.4.3 Climate Responses to SLCFs ===

<div id="h2-21-siblings" class="h2-siblings"></div>

This section briefly discusses the climate response to SLCFs, in particular to changes in aerosols, and gathers complementary information and assessments from Chapters 3, 7, 8 and 10.

Warming or cooling atmospheric aerosols, such as BC and sulphate, can affect temperature and precipitation in distinct ways by modifying the shortwave and longwave radiation, the lapse rate of the troposphere, and influencing cloud microphysical properties ( [[IPCC:Wg1:Chapter:Chapter-10#10.1.4.1.4|Section 10.1.4]] , Box 8.1). An important distinction between scattering and absorbing aerosols is the opposing nature of their influences on circulation, clouds and precipitation, besides surface temperature as evident from the contrasting regional climate responses to regional aerosol emissions (e.g., [[#Lewinschal--2019|Lewinschal et al., 2019]] ; [[#Sand--2020|Sand et al., 2020]] ; also see Chapters 8 and 10).

On the global scale, as assessed in Chapter 3, anthropogenic aerosols have ''likely'' cooled GSAT since 1850–1900 driven by the negative aerosol forcing, while it is ''extremely likely'' that human-induced stratospheric ozone depletion has primarily driven stratospheric cooling between 1979 and the mid-1990s. Multiple modelling studies support the understanding that present-day emissions of SO <sub>2</sub> , a precursor for sulphate aerosols, are the dominant driver of near- surface air temperature responses in comparison to BC or OC even though, for some regions, BC forcing plays a key role (Baker et al. , 2015; Samset et al. , 2016; Stjern et al. , 2017; Zanis et al. , 2020) '''.''' Further, there is ''high confidence'' that the aerosol-driven cooling has led to detectable large-scale water-cycle changes since at least the mid-20th <sup></sup> century as assessed in Chapter 8. The overall effect of surface cooling from anthropogenic aerosols is to reduce global precipitation and alter large-scale atmospheric circulation patterns ( ''high confidence'' ), primarily driven by the cooling effects of sulphate aerosols ( [[IPCC:Wg1:Chapter:Chapter-8#8.2.1|Section 8.2.1]] ). In addition, there is ''high confidence'' that darkening of snow through the deposition of black carbon and other light-absorbing particles enhances snowmelt ( [[IPCC:Wg1:Chapter:Chapter-7#7.3.4.3|Section 7.3.4.3]] ; SROCC Chapter 3).

In AR5, there was ''low confidence'' in the overall understanding of climate response to spatially varying patterns of forcing, though there was ''medium'' to ''high confidence'' in some regional climate responses, such as the damped warming of the NH and shifting of the ITCZ from aerosols, and positive feedbacks enhancing the local response from high-latitude snow and ice albedo changes. Since AR5, the relationship between inhomogeneous forcing and climate response is better understood, providing further evidence of the climate influence of SLCFs (aerosols and ozone in particular) on global to regional scales ( [[#Collins--2013|Collins et al., 2013]] ; [[#Shindell--2015|Shindell et al., 2015]] ; [[#Aamaas--2017|Aamaas et al., 2017]] ; [[#Kasoar--2018|Kasoar et al., 2018]] ; [[#Persad--2018|Persad and Caldeira, 2018]] ; [[#Wilcox--2019|Wilcox et al., 2019]] ) which differ from the relatively homogeneous spatial influence from LLGHGs.

Large geographical variations in aerosol ERFs (Section 6.4.1) affect global and regional temperature responses ( [[#Myhre--2013|Myhre et al., 2013]] ; [[#Shindell--2015|Shindell et al., 2015]] ). Multi-model CMIP6 ensemble-mean results (Figure 6.13) show cooling over almost all areas of the globe in response to increases of aerosol and their precursor emissions from 1850 to the recent past (1995–2014). While the ERF has hotspots, the temperature response is more evenly distributed in line with the results of CMIP5 models including the temperature response to ozone changes ( [[#Shindell--2015|Shindell et al., 2015]] ). The ensemble-mean global mean surface temperature decreases by 0.66°C ± 0.51°C while decreasing by 0.97°C ± 0.54°C for the Northern Hemisphere and 0.34°C ± 0.2°C for the Southern Hemisphere. The zonal-mean temperature response is negative at all latitudes ( ''high confidence'' ) and becomes more negative with increasing latitude, with a maximum ensemble-mean decrease of around 2.7°C at northern polar latitudes. The zonal-mean response is not directly proportional to the zonal-mean forcing, especially in the Arctic where the temperature response is cooling while the local ERF is positive (Figure 6.10). This is consistent with prior studies showing that the Arctic, in particular, is highly sensitive to forcing at NH mid-latitudes (e.g., [[#Shindell--2009|Shindell and Faluvegi, 2009]] ; [[#Sand--2013a|Sand et al., 2013a]] ) and with results from CMIP5 models (more on the Arctic below; [[#Shindell--2015|Shindell et al., 2015]] ). Thus, there is ''high confidence'' that the temperature response to aerosols is more asymmetric than the response to WMGHGs and negative at all latitudes.

<div id="_idContainer039" class="_idGenObjectStyleOverride-1"></div>

[[File:06b25de61a738d01818e9556b9bc6b0b IPCC_AR6_WGI_Figure_6_13.png]]

'''Figure 6.13 |''' '''Multi-model mean surface air temperature response over the recent past (1995–2014) induced by aerosol changes since 1850.''' Calculation is based on the difference between CMIP6 ‘historical’ and AerChemMIP ‘hist-piAer’ experiments averaged over 1995–2014, where '''(a)''' is the spatial pattern of the annual mean surface air temperature response, and '''(b)''' is the mean zonally averaged response. Model means are derived from the years 1995–2014. Uncertainty is represented using the advanced approach: No overlay indicates regions with robust signal, where ≥66% of models show change greater than variability threshold and ≥80% of all models agree on sign of change; diagonal lines indicate regions with no change or no robust signal, where &lt;66% of models show a change greater than the variability threshold; crossed lines indicate regions with conflicting signal, where ≥66% of models show change greater than variability threshold and &lt;80% of all models agree on sign of change. For more information on the advanced approach, please refer to the Cross-Chapter Box Atlas.1. AerChemMIP models MIROC6, MRI-ESM2-0, NorESM2-LM, GFDL-ESM4, GISS-E2-1-G and UKESM1-0-LL are used in the analysis. Further details on data sources and processing are available in the chapter data table (Table 6.SM.3).

The asymmetric aerosol and greenhouse gas forcing on regional-scale climate responses have also been assessed to lead to contrasting effects on precipitation in Chapter 8. The asymmetric historical radiative forcing due to aerosols led to a southward shift in the tropical rain belt ( ''high confidence'' ) and contributed to the Sahel drought from the 1970s to the 1980s ( ''high confidence'' ). Furthermore, the asymmetry of the forcing led to contrasting effects in monsoon precipitation changes over West Africa, Southern Asia and Eastern Asia over much of the mid-20th <sup></sup> century due to GHG-induced precipitation increases counteracted by anthropogenic aerosol-induced decreases ( ''high confidence'' ) (see [[IPCC:Wg1:Chapter:Chapter-8#8.3|Section 8.3]] and Box 8.1).

The Arctic region is warming considerably faster than the rest of the globe ( [[IPCC:Wg1:Chapter:Atlas|Atlas]] 11.2.2) and, generally, studies indicate that this amplification of the temperature response toward the Arctic has an important contribution from local and remote aerosol forcing ( [[#Stjern--2017|Stjern et al., 2017]] ; [[#Westervelt--2018|Westervelt et al., 2018]] ). Several studies indicate that changes in long-range transport of sulphate and BC from northern mid-latitudes can potentially explain a significant fraction of Arctic warming since the 1980s (e.g., Navarro et al. , 2016; Breider et al. , 2017; Ren et al. , 2020) . Modelling studies show that changes in mid-latitude aerosols have influenced Arctic climate by changing the radiative balance through aerosol–radiation and aerosol–cloud interactions, and enhancing poleward heat transport ( [[#Navarro--2016|Navarro et al., 2016]] ; [[#Ren--2020|Ren et al., 2020]] ). Idealized aerosol-perturbation studies have shed further light on the sensitivity of Arctic temperature response to individual aerosol species. Studies show relatively large responses in the Arctic to BC perturbations and reveal the importance of remote BC forcing by rapid adjustments (Sand et al. , 2013b; Stjern et al. , 2017; L. Liu et al. , 2018; Yang et al. , 2019b) . Perturbations in SO <sub>2</sub> emissions over major emitting regions in the Northern Hemisphere have been shown to produce the largest Arctic temperature responses ( [[#Kasoar--2018|Kasoar et al., 2018]] ; [[#Lewinschal--2019|Lewinschal et al., 2019]] ).

The effects of changes in aerosols on local and remote changes in temperature, circulation and precipitation are sensitive to a number of model uncertainties affecting aerosol sources, transformation and resulting radiative efficacy. Therefore, regional climate effects in global model studies must be interpreted with caution. When investigating the climate response to regional aerosol emissions, such uncertainties are likely to be confounded even further by the variability between models in regional climate and circulation patterns, leading to greater inter-model spread at regional scales than at a global scale ( [[#Baker--2015|Baker et al., 2015]] ; [[#Kasoar--2016|Kasoar et al., 2016]] ).

In summary, over the historical period, changes in aerosols and their ERF have primarily contributed to cooling, partly masking the human-induced warming ( ''high confidence'' ). Radiative forcings induced by aerosol changes lead to both local and remote changes in temperature ( ''high confidence'' ). The temperature response preserves hemispheric asymmetry of the ERF but is more latitudinally uniform with strong amplification of the temperature response towards the Arctic ( ''medium confidence'' ).

<div id="6.4.4" class="h2-container"></div>

<span id="indirect-radiative-forcing-through-effects-of-slcfs-on-the-carbon-cycle"></span>
=== 6.4.4 Indirect Radiative Forcing Through Effects of SLCFs on the Carbon Cycle ===

<div id="h2-22-siblings" class="h2-siblings"></div>

Deposition of reactive nitrogen (Nr; i.e., NH <sub>3</sub> and NO <sub>x</sub> ) increases the plant productivity and carbon sequestration in N-limited forests and grasslands, and also in open and coastal waters and open ocean. Such inadvertent fertilization of the biosphere can lead to eutrophication and reduction in biodiversity in terrestrial and aquatic ecosystems. The AR5 assessed that it is ''likely'' that Nr deposition over land currently increases natural CO <sub>2</sub> sinks, in particular in forests, but the magnitude of this effect varies between regions ( [[#Ciais--2013|Ciais et al., 2013]] ). Increasing Nr deposition or the synergy between increasing Nr deposition and atmospheric CO <sub>2</sub> concentration could have contributed to the increasing global-net land CO <sub>2</sub> sink ( [[IPCC:Wg1:Chapter:Chapter-5#5.2.1.4.1|Section 5.2.1.4.1]] ).

Ozone uptake itself damages photosynthesis and reduces plant growth with consequences for the carbon and water cycles ( [[#Ainsworth--2012|Ainsworth et al., 2012]] ; [[#Emberson--2018|Emberson et al., 2018]] ). The AR5 concluded there was robust evidence of the effect of ozone on plant physiology and subsequent alteration of the carbon storage, but considered insufficient quantification of and a lack of systematic incorporation of the ozone effect in carbon-cycle models as a limitation to assess the terrestrial carbon balance ( [[#Ciais--2013|Ciais et al., 2013]] ). Since AR5 several more ESMs have incorporated interactive ozone-vegetation damage resulting in an increase in evidence to support the influence of ozone on the land carbon cycle. The new modelling studies tend to focus on ozone effects on plant productivity rather than the land carbon storage and agree that ozone-induced gross-primary productivity (GPP) losses are largest today in the eastern USA, Europe and eastern China, ranging from 5–20% on the regional scale ( ''low confidence'' ) (Yue and [[#Unger--2014|Unger, 2014]] ; Lombardozzi et al. , 2015; Yue et al. , 2017; Oliver et al. , 2018). There is ''medium evidence'' and ''high agreement'' based on observational studies and models that ozone-vegetation interactions further influence the climate system, including water and carbon cycles by affecting stomatal control over plant transpiration of water vapour between the leaf surface and atmosphere (Wittig et al. , 2007; Sun et al. , 2012; VanLoocke et al. , 2012; Lombardozzi et al. , 2013; Hoshika et al. , 2015; Arnold et al. , 2018). While some modelling studies suggest that the unintended Nr deposition fertilization effect in forests may potentially offset the ozone-induced carbon losses ( [[#Felzer--2007|Felzer et al., 2007]] ; [[#de%20Vries--2017|de Vries et al., 2017]] ), complex interactions have been observed between ozone and Nr deposition to ecosystems that have not yet been included in ESMs. For some plants, the effects of increasing ozone on root biomass become more pronounced as Nr deposition increased, and the beneficial effects of Nr on root development were lost at higher ozone treatments ( [[#Mills--2016|Mills et al., 2016]] ). Reducing uncertainties in ozone vegetation damage effects on the carbon cycle requires improved information on the sensitivity of different plant species to ozone, and measurements of ozone dose-response relationships for tropical plants, which are currently lacking. Surface ozone effect on the land carbon sink and indirect CO <sub>2</sub> forcing, therefore, remains uncertain. Collins et al. (2010) showed that adding in the effects of surface ozone on vegetation damage and reduced uptake of CO <sub>2</sub> added about 10% to the methane emissions metrics and could change the sign of the NO <sub>x</sub> metrics. However, this estimate has to be considered as an upper limit due to limitation of the paramaterization used by Sitch et al. (2007) considering more recent knowledge and is thus not included in the current metrics ( [[IPCC:Wg1:Chapter:Chapter-7#7.6.1.3|Section 7.6.1.3]] ).

Tropospheric aerosols influence the land and ocean ecosystem productivity and the carbon cycle through changing physical climate and meteorology ( [[#Jones--2003|Jones, 2003]] ; [[#Cox--2008|Cox et al., 2008]] ; [[#Mahowald--2011|Mahowald, 2011]] ; [[#Unger--2017|Unger et al., 2017]] ) and through changing deposition of nutrients including nitrogen, sulphur, iron and phosphorous ( [[#Mahowald--2017|Mahowald et al., 2017]] ; [[#Kanakidou--2018|Kanakidou et al., 2018]] ). There is ''robust evidence'' and ''high agreement'' from field (Oliveira et al. , 2007; Cirino et al. , 2014; Rap et al. , 2015; X. Wang et al. , 2018) and modelling ( [[#Mercado--2009|Mercado et al., 2009]] ; [[#Strada--2016|Strada and Unger, 2016]] ; [[#Lu--2017|Lu et al., 2017]] ; [[#Yue--2017|Yue et al., 2017]] ) studies that aerosols affect plant productivity through increasing the diffuse fraction of downward shortwave radiation, although the magnitude and importance to the global land carbon sink is controversial. At large scales the dominant effect of aerosols on the carbon cycle is ''likely'' a global cooling effect of the climate ( ''medium confidence'' ) ( [[#Jones--2003|Jones, 2003]] ; [[#Mahowald--2011|Mahowald, 2011]] ; [[#Unger--2017|Unger et al., 2017]] ). We assess that these interactions between aerosols and the carbon cycle are currently too uncertain to constrain quantitatively the indirect CO <sub>2</sub> forcing.

In summary, reactive nitrogen, ozone and aerosols affect terrestrial vegetation and the carbon cycle through deposition and effects on large-scale radiation ( ''high confidence'' ) but the magnitude of these effects on the land carbon sink, ecosystem productivity and indirect CO <sub>2</sub> forcing remain uncertain due to the difficulty in disentangling the complex interactions between the effects. As such, we assess the effects to be of second order in comparison to the direct CO <sub>2</sub> forcing ( ''high confidence'' ) but, at least for ozone, it could add a substantial (positive) forcing compared with its direct forcing ( ''low confidence'' ).

<div id="6.4.5" class="h2-container"></div>

<span id="non-co-2-biogeochemical-feedbacks"></span>
=== 6.4.5 Non-CO <sub>2</sub> biogeochemical Feedbacks ===

<div id="h2-23-siblings" class="h2-siblings"></div>

Climate change-induced changes in atmospheric composition and forcing due to perturbations in natural processes constitute an Earth system feedback amplifying (positive feedback) or diminishing (negative feedback) the initial climate perturbation ( [[#Ciais--2013|Ciais et al., 2013]] ; [[#Heinze--2019|Heinze et al., 2019]] ). Quantification of these biogeochemical feedbacks is important to allow for a better estimate of the expected effects of emissions reduction policies for mitigating climate change and the effect on the allowable global carbon budget ( [[#Lowe--2018|Lowe and Bernie, 2018]] ). Biogeochemical feedbacks due to changes in the carbon cycle are assessed in [[IPCC:Wg1:Chapter:Chapter-5#5.4.5|Section 5.4.5]] , while physical and biogeophysical climate feedbacks are assessed in [[IPCC:Wg1:Chapter:Chapter-7#7.4.2|Section 7.4.2]] . Additionally, non-CO <sub>2</sub> biogeochemical feedbacks due to climate-driven changes in methane sources and N <sub>2</sub> O sources and sinks are assessed in [[IPCC:Wg1:Chapter:Chapter-5#5.4.7|Section 5.4.7]] . The goal of this section is to estimate the feedback parameter ( α as defined in section 7.4.1.1) from climate-induced changes in atmospheric abundances or lifetimes of SLCFs mediated by natural processes or atmospheric chemistry. These non-CO <sub>2</sub> biogeochemical feedbacks act on time scales of years to decades and have important implications for climate sensitivity and emissions-abatement policies. The feedback parameter is quantified entirely from ESMs that expand the complexity of CCMs by coupling the physical climate and atmospheric chemistry to land and ocean biogeochemistry. In AR5, α for non-CO <sub>2</sub> biogeochemical feedbacks was estimated from an extremely limited set of modelling studies with much less confidence associated with the estimate. Since AR5, ESMs have advanced to include more feedback processes, facilitating a relatively more robust assessment of α . CMIP6 ESMs participating in AerChemMIP performed coordinated sets of experiments ( [[#Collins--2017|Collins et al., 2017]] ) facilitating the consistent estimation of α ( [[#Thornhill--2021a|Thornhill et al., 2021a]] ) and we rely on this multi-model analysis for the best estimates (Table 6.8). Considering the consistent methodology, the assessed central values and 5–95% ranges for α are based on the AerChemMIP estimates. The full range of model uncertainty is not captured in AerChemMIP because of the relatively small ensemble size, therefore estimates from studies using other models or with different protocols are discussed to reinforce or critique these values.

<div id="_idContainer042" class="Basic-Text-Frame"></div>

'''Table 6.8 |''' '''Assessed central estimates and the''' very likely '''ranges (5–95%) of non-CO''' <sub>2</sub> '''biogeochemical feedback parameter''' ( α <sub>x</sub> ''') based on the AerChemMIP ensemble estimates (Thornhill''' '''et al.''' ''', 2021a).''' As in [[IPCC:Wg1:Chapter:Chapter-7#7.4.1.1|Section 7.4.1.1]] , α <sub>x</sub> (W m <sup>−2</sup> °C <sup>−1</sup> ) for a feedback variable ''x'' is defined as   where   is the change in TOA energy balance in response to a change in ''x'' induced by a change in surface temperature ( ''T'' ). The 5–95% range is calculated as mean ± standard deviation × 1.645 for each feedback. The level of confidence in these estimates is ''low'' owing to the large intermodel spread. Published estimates of α <sub>x</sub> are also shown for comparison.

{| class="wikitable"
|-
| '''Non-CO''' <sub>2</sub> '''Biogeochemical Climate Feedback''' ( x ''')'''

| '''Number of AerChemMIP Models'''

| '''Assessed Central Estimate and Very Likely Range of Feedback Parameter''' ( α <sub>x</sub> ''')'''

'''W m''' <sup>−2</sup> '''°C''' <sup>−1</sup>

| '''Published Estimates of''' α <sub>x</sub>

'''W m''' <sup>−2</sup> '''°C''' <sup>−1</sup>

|-
| Sea salt

| 6

| –0.049 [–0.13 to +0.03]

| –0.08 ( [[#Paulot--2020|Paulot et al., 2020]] )

|-
| DMS

| 3

| 0.005 [0.0 to 0.01]

| –0.02 ( [[#Ciais--2013|Ciais et al., 2013]] )

|-
| Dust

| 6

| –0.004 [–0.02 to +0.01]

| –0.04 to +0.02 ( [[#Kok--2018|Kok et al., 2018]] )

|-
| Ozone

| 4

| –0.064 [–0.08 to -0.04]

| –0.015 ( [[#Dietmüller--2014|Dietmüller et al., 2014]] ), –0.06 ( [[#Muthers--2014|Muthers et al., 2014]] , stratospheric ozone changes only), –0.01 ( [[#Marsh--2016|Marsh et al., 2016]] , stratospheric ozone changes only), –0.13 ( [[#Nowack--2015|Nowack et al., 2015]] , stratospheric ozone and water vapour changes), –0.007 ± 0.009 ( [[#Heinze--2019|Heinze et al., 2019]] , tropospheric ozone changes only)

|-
| BVOC

| 4

| –0.05 [–0.22 to +0.12]

| –0.06 ( [[#Scott--2017|Scott et al., 2017]] , aerosol effects only), –0.01 ( [[#Paasonen--2013|Paasonen et al., 2013]] ; indirect aerosol effects only), 0–0.06 ( [[#Ciais--2013|Ciais et al., 2013]] )

|-
| Lightning

| 4

| –0.010 [–0.04 to +0.02]

|
|-
| Methane lifetime

| 4

| –0.030 [–0.12 to +0.06]

| –0.30 ± 0.01 ( [[#Heinze--2019|Heinze et al., 2019]] )

|-
| Total non-CO <sub>2</sub> Biogeochemical feedbacks assessed in this chapter

|
| –0.200 [–0.41 to +0.01]

| 0.0 ± 0.15 ( [[#Sherwood--2020|Sherwood et al., 2020]] )

|}

'''Climate–sea-spray feedback:''' Sea-spray emissions from ocean surfaces influence climate directly or indirectly through the formation of CCN as discused in Section 6.2.1.2. They are sensitive to SST and sea ice extent, as well as to wind speed, and are therefore expected to feedback on climate ( [[#Struthers--2013|Struthers et al., 2013]] ). However, there are large uncertainties in the strength of climate feedback from sea-spray aerosols because of the diversity in the model representation of emissions (many represent sea-salt emissions only) and their functional dependence on environmental factors noted above, in situ atmospheric chemical and physical processes affecting the sea-spray lifetime, and aerosol–cloud interactions ( [[#Struthers--2013|Struthers et al., 2013]] ; [[#Soares--2016|Soares et al., 2016]] ; [[#Nazarenko--2017|Nazarenko et al., 2017]] ). Additional work is needed to identify how sea-spray and POA emissions respond to shifts in ocean biology and chemistry in response to warming, ocean acidification and changes in circulation patterns (Cochran et al. , 2017) , and affect CCN and INP formation ( [[#DeMott--2016|DeMott et al., 2016]] ). AerChemMIP models, representing only the sea-salt emissions, agree that the sea-salt-climate feedback is negative, however there is a large range in the feedback parameter indicating large uncertainties (Table 6.8).

'''Climate–DMS feedback:''' Dimethyl sulphide (DMS) is produced by marine phytoplankton and is emitted to the atmosphere where it can lead to the subsequent formation of sulphate aerosol and CCN (Section 6.2.2.5). Changes in DMS emissions from ocean could feedback on climate through their response to changes in temperature, solar radiation, ocean mixed-layer depth, sea ice extent, wind speed, nutrient recycling or shifts in marine ecosystems due to ocean acidification and climate change, or atmospheric processing of DMS into CCN ( [[#Heinze--2019|Heinze et al., 2019]] ). Models with varying degrees of representation of the relevant biogeochemical processes and effects on DMS fluxes produce diverging estimates of changes in DMS emissions strength under climate change resulting in large uncertainties in the DMS–sulphate–cloud albedo feedback ( [[#Bopp--2004|Bopp et al., 2004]] ; [[#Kloster--2007|Kloster et al., 2007]] ; [[#Gabric--2013|Gabric et al., 2013]] ). In AR5, the climate-DMS feedback parameter was estimated to be –0.02 W m <sup>–2</sup> °C <sup>–1</sup> based on a single model. Since AR5, new modelling studies using empirical relationships between pH and total DMS production find that global DMS emissions decrease due to combined ocean acidification and climate change, leading to a strong positive climate feedback ( [[#Six--2013|Six et al., 2013]] ; [[#Schwinger--2017|Schwinger et al., 2017]] ). However, another study argues for a much weaker positive feedback globally due to complex and compensating regional changes in marine ecosystems (S. [[#Wang--2018|]] [[#Wang--2018|]] [[#Wang--2018|Wang et al., 2018]] ). The AerChemMIP multi-model analysis suggests small positive feedback (Table 6.8), consistent with these recent studies, but with large uncertainties in the magnitude of α .

'''Climate–dust feedback:''' Mineral dust is the most abundant aerosol type in the atmosphere, when considering aerosol mass, and affects the climate system by interacting with both longwave and shortwave radiation as well as contributing to the formation of CCN and INP. Because dust emissions are sensitive to climate variability (e.g., through changes in the extent of arid land; Section 6.2.2.4), it has been hypothesized that the climate-dust feedback could be an important feedback loop in the climate system. Since AR5, an improved understanding of the shortwave absorption properties of dust as well as a consensus that dust particles are larger than previously thought has led to a revised understanding that the magnitude of radiative forcing due to mineral dust is small ( [[#Kok--2017|Kok et al., 2017]] ; [[#Ryder--2018|Ryder et al., 2018]] ). A recent study notes that global models underestimate the amount of coarse dust in the atmosphere and accounting for this limitation raises the possibility that dust emissions warm the climate system ( [[#Adebiyi--2020|Adebiyi and Kok, 2020]] ). Model predictions of dust emissions in response to future climate change range from an increase ( [[#Woodward--2005|Woodward et al., 2005]] ) to a decrease ( [[#Mahowald--2003|Mahowald and Luo, 2003]] ), thus leading to high uncertainties on the sign of the climate-dust feedback. Since AR5, [[#Kok--2018|Kok et al. (2018)]] estimated the direct dust-climate feedback parameter, from changes in the dust direct radiative effect only, to be in the range –0.04 to +0.02 W m <sup>–2</sup> °C <sup>–1</sup> . The assessed central value and the 5–95% range of the climate-dust feedback parameter based on AerChemMIP ensemble (Table 6.8) is within the range of the published estimate, however both the magnitude and sign of α are model-dependent.

'''Climate–ozone feedback:''' Changes in ozone concentrations in response to projected climate change have been shown to lead to a potential climate-atmospheric chemistry feedback. Chemistry–climate models consistently project a decrease in lower tropical stratospheric ozone levels due to enhanced upwelling of ozone-poor tropospheric air associated with surface warming-driven strengthening of the Brewer-Dobson circulation ( [[#Bunzel--2013|Bunzel and Schmidt, 2013]] ). Further, models project an increase in middle and extratropical stratospheric ozone due to increased downwelling through the strengthened Brewer-Dobson circulation ( [[#Bekki--2013|Bekki et al., 2013]] ; [[#Dietmüller--2014|Dietmüller et al., 2014]] ). These stratospheric ozone changes induce a net-negative global mean ozone radiative feedback ( [[#Dietmüller--2014|Dietmüller et al., 2014]] ). Tropospheric ozone shows a range of responses to climate with models generally agreeing that warmer climate will lead to decreases in the tropical lower troposphere owing to increased water vapour, and increases in the subtropical to mid-latitude upper troposphere due to increases in lightning and stratosphere-to-troposphere transport ( [[#Stevenson--2013|Stevenson et al., 2013]] ). A small positive feedback is estimated from climate-induced changes in global mean tropospheric ozone ( [[#Dietmüller--2014|Dietmüller et al., 2014]] ) while a small negative feedback is estimated by [[#Heinze--2019|Heinze et al. (2019)]] based on the model results of [[#Stevenson--2013|Stevenson et al. (2013)]] . Additionally, these ozone feedbacks induce a change in stratospheric water vapour amplifying the feedback due to stratospheric ozone ( [[#Stuber--2001|Stuber et al., 2001]] ). Since AR5, several modelling studies have estimated the intensity of meteorology-driven ozone feedbacks on climate from either combined tropospheric and stratospheric ozone changes or separately with contrasting results. One study suggests no change ( [[#Marsh--2016|Marsh et al., 2016]] ), while other studies report reductions of ECS ranging from 7–8% ( [[#Dietmüller--2014|Dietmüller et al., 2014]] ; [[#Muthers--2014|Muthers et al., 2014]] ) to 20% ( [[#Nowack--2015|Nowack et al., 2015]] ). The estimate of this climate-ozone feedback parameter is very strongly model-dependent with values ranging from –0.13 to –0.01 W m <sup>–2</sup> °C <sup>–1</sup> though there is agreement that it is negative. The assessed central value and the 5–95% range of climate-ozone feedback parameter based on AerChemMIP ensemble is within the range of these published estimates, but closer to the lower bound. This climate-ozone feedback factor does not include the feedback on ozone from lightning changes which is discussed separately below.

'''Climate–BVOC feedback:''' BVOCs, such as isoprene and terpenes, are produced by land vegetation and marine plankton (Sections 6.2.2.3 and 6.2.2.5). Once in the atmosphere, BVOCs and their oxidation products lead to the formation of secondary organic aerosols (SOA) exerting a negative forcing, and increased ozone concentrations and methane lifetime exerting a positive forcing. BVOC emissions are suggested to lead to a climate feedback in part because of their strong temperature dependence observed under present-day conditions ( [[#Kulmala--2004|Kulmala et al., 2004]] ; [[#Arneth--2010a|Arneth et al., 2010a]] ). Their response to future changes in climate and CO <sub>2</sub> levels remains uncertain (Section 6.2.2.3). Estimates of the climate-BVOC feedback parameter are typically based on global models which vary in their level of complexity of emissions parametrization, BVOC speciation, the mechanism of SOA formation and the interaction with ozone chemistry ( [[#Thornhill--2021a|Thornhill et al., 2021a]] ). Since AR5, observational studies ( [[#Paasonen--2013|Paasonen et al., 2013]] ) and models ( [[#Scott--2018|Scott et al., 2018]] ) estimate the feedback due to biogenic SOA (via changes in BVOC emissions) to be in the range of about –0.06 to –0.01 W m <sup>–2</sup> °C <sup>–1</sup> . The assessed central estimate of the climate-BVOC feedback parameter based on the AerChemMIP ensemble suggests that climate-induced increases in SOA from BVOCs will lead to a strong cooling effect that will outweigh the warming from increased ozone and methane lifetime, however the uncertainty is large ( [[#Thornhill--2021a|Thornhill et al., 2021a]] ).

'''Climate–lightning NO''' <sub>x</sub> '''feedback:''' As discussed in Section 6.2.2.1, climate change influences lightning NO <sub>x</sub> emissions. Increases in lightning NO <sub>x</sub> emissions will not only increase tropospheric ozone and decrease methane lifetime but also increase the formation of sulphate and nitrate aerosols, via oxidant changes, offsetting the positive forcing from ozone. The response of lightning NO <sub>x</sub> to climate change remains uncertain and is highly dependent on the parametrization of lightning in ESMs (Section 6.2.1.2; [[#Finney--2016b|Finney et al., 2016b]] ; [[#Clark--2017|Clark et al., 2017]] ). AerChemMIP multi-model ensemble mean estimate a net-negative climate feedback from increases in lightning NO <sub>x</sub> in a warming world ( [[#Thornhill--2021a|Thornhill et al., 2021a]] ). All AerChemMIP models use a cloud-top height lightning parametrization that predicts increases in lightning with warming. However, a positive climate-lightning NO <sub>x</sub> feedback cannot be ruled out because of the dependence of the response to lightning parametrizations as discussed in Section 6.2.2.1.

'''Climate–methane lifetime feedback:''' Warmer and wetter climate will lead to increases in OH and oxidation rates leading to reduced atmospheric methane lifetime – a negative feedback ( [[#Naik--2013|Naik et al., 2013]] ; [[#Voulgarakis--2013|Voulgarakis et al., 2013]] ). Furthermore, since OH is in turn removed by methane, the climate-methane lifetime feedback will be amplified (Section 6.3.1; [[#Prather--1996|Prather, 1996]] ) . Based on the multi-model results of [[#Voulgarakis--2013|Voulgarakis et al. (2013)]] , α for climate-methane lifetime is estimated to be –0.030 ± 0.01 W m <sup>−2</sup> °C <sup>−1</sup> by [[#Heinze--2019|Heinze et al. (2019)]] . The assessed central value of α based on the AerChemMIP ensemble is within the range of this estimate but with greater uncertainty ( [[#Thornhill--2021a|Thornhill et al., 2021a]] ).

'''Climate–fire feedback:''' Wildfires are a major source of SLCF emissions (Section 6.2.2.6). Climate change has the potential to enhance fire activity (Sections 12.4 and 5.4.3.2) thereby enhancing SLCF emissions leading to feedbacks. Climate-driven increases in fire could potentially lead to offsetting feedback from increased ozone and decreased methane lifetime (due to increases in OH) leaving the feedback from aerosols to dominate with an uncertain net effect (e.g., [[#Landry--2015|Landry et al., 2015]] ). The AR5 assessment of climate-fire feedbacks included a value of α due to fire aerosols to be in the range of –0.03 to + 0.06 W m <sup>−2</sup> °C <sup>−1</sup> based on [[#Arneth--2010a|Arneth et al. (2010a)]] . A recent study estimates climate feedback due to fire aerosols to be greater than that due to BVOCs, with a value of α equal to –0.15 (–0.24 to –0.05) W m <sup>−2</sup> °C <sup>−1</sup> ( [[#Scott--2018|Scott et al., 2018]] ). Clearly, the assessment of fire-related non-CO <sub>2</sub> biogeochemical feedbacks is very uncertain because of limitations in the process understanding of the interactions between climate, vegetation and fire dynamics, and atmospheric chemistry and their representation in the current generation ESMs. Some AerChemMIP ESMs include the representation of fire dynamics but do not activate their interaction with atmospheric chemistry. Given the large uncertainty and lack of information from AerChemMIP ESMs, we do not include a quantitative assessment of climate-fire feedback for AR6.

In summary, climate-driven changes in emissions, atmospheric abundances or lifetimes of SLCFs are assessed to have an overall cooling effect, that is, a negative feedback parameter of –0.20 [–0.41 to + 0.01] W m <sup>−2</sup> °C <sup>−1</sup> , thereby reducing climate sensitivity ( [[IPCC:Wg1:Chapter:Chapter-7#7.4.2.5.1|Section 7.4.2.5.1]] ). This net feedback parameter is obtained by summing the assessed estimates for the individual feedback given in Table 6.8. ''confidence'' in the magnitude and the sign of most of the individual as well as the total non-CO <sub>2</sub> biogeochemical feedbacks remains ''low'' as evident from the large range in the value of α . This large uncertainty is attributed to the diversity in model representation of the relevant chemical and biogeochemical processes based on limited process-level understanding.

<div id="6.4.6" class="h2-container"></div>

<span id="erf-by-aerosols-in-proposed-solar-radiation-modification"></span>
=== 6.4.6 ERF by Aerosols in Proposed Solar Radiation Modification ===

<div id="h2-24-siblings" class="h2-siblings"></div>

Solar radiation modification (SRM; Sections 4.6.3.3 and 8.6.3) has the potential to exert a significant ERF on the climate, mainly by affecting the SW component of the radiation budget (e.g., [[#Caldeira--2013|Caldeira et al., 2013]] ; [[#NRC--2015|NRC, 2015]] ; [[#Lawrence--2018|Lawrence et al., 2018]] ). The possible ways and the extent to which the most commonly discussed options may affect radiative forcing is addressed in this section. Side effects of SRM on stratospheric ozone and changes in atmospheric transport due to radiative heating of the lower stratosphere are discussed in [[IPCC:Wg1:Chapter:Chapter-4#4.6.3.3|Section 4.6.3.3]] .

Stratospheric aerosol injections (SAI) have the potential to achieve a high negative global ERF, with maximum ERFs ranging from –5 to –2 W m <sup>–2</sup> ( [[#Niemeier--2015|Niemeier and Timmreck, 2015]] ; [[#Weisenstein--2015|Weisenstein et al., 2015]] ; [[#Niemeier--2017|Niemeier and Schmidt, 2017]] ; [[#Kleinschmitt--2018|Kleinschmitt et al., 2018]] ). The magnitude of the maximum achievable ERF depends on the chosen aerosol type and mixture, internal structure and size, or precursor gas (e.g., SO <sub>2</sub> ), as well as the injection strategy (latitude, altitude, magnitude and season of injections), plume dispersal, model representation of aerosol microphysics, and ambient aerosol concentrations (Rasch et al. , 2008; Robock et al. , 2008; Pierce et al. , 2010; Weisenstein et al. , 2015; Laakso et al. , 2017; MacMartin et al. , 2017; Dai et al. , 2018; Kleinschmitt et al. , 2018; Vattioni et al. , 2019; Visioni et al. , 2019) . For sulphur, the radiative forcing efficiency is of around –0.1 to –0.4 W m <sup>–2</sup> /(TgS yr <sup>–1</sup> <sup>)</sup> ( [[#Niemeier--2015|Niemeier and Timmreck, 2015]] ; [[#Weisenstein--2015|Weisenstein et al., 2015]] ; [[#Niemeier--2017|Niemeier and Schmidt, 2017]] ). Different manufactured aerosols, such as ZrO <sub>2</sub> , TiO <sub>2</sub> and Al <sub>2</sub> O <sub>3</sub> , have different ERF efficiencies compared to sulphate (Ferraro et al. , 2011; Weisenstein et al. , 2015; Dykema et al. , 2016; Jones et al. , 2016) . The aerosol size distribution influences the optical properties of an aerosol layer, and hence the ERF efficiency, which also depends on the dispersion, transport, and residence time of the aerosols.

For marine cloud brightening (MCB), seeded aerosols may affect both cloud microphysical and macrophysical properties ( [[IPCC:Wg1:Chapter:Chapter-7#7.3.3.2|Section 7.3.3.2]] ). By principle, MCB relies on ERFaci through the so-called Twomey effect ( [[#Twomey--1977|Twomey, 1977]] ), but ERFari may be of equal magnitude as shown in studies that consider spraying of sea salt outside tropical marine cloud areas ( [[#Jones--2012|Jones and Haywood, 2012]] ; [[#Partanen--2012|Partanen et al., 2012]] ; [[#Alterskjaer--2013|Alterskjaer and Kristjánsson, 2013]] ; [[#Ahlm--2017|Ahlm et al., 2017]] ). The maximum negative ERF estimated from modelling is within the range of –5.4 to –0.8 W m <sup>–2</sup> (Latham et al. , 2008; Rasch et al. , 2009; Jones et al. , 2011; Partanen et al. , 2012; [[#Alterskjaer--2013|Alterskjaer and Kristjánsson, 2013]] ) . For dry sea salt, the ERF efficiency is estimated to be within the range of –3 to –10 W m <sup>–2</sup> /(Pg yr <sup>–1</sup> ), when emitted over tropical oceans in ESMs in the Geoengineering Intercomparison Project (GeoMIP; [[#Ahlm--2017|Ahlm et al., 2017]] ). Cloud-resolving models reveal the complex behaviour and response of stratocumulus clouds to seeding, in that the ERF efficiency depends on meteorological conditions, and the ambient aerosol composition, where lower background particle concentrations may increase the ERFaci efficiency ( [[#Wang--2011|Wang et al., 2011]] ). Seeding could suppress precipitation formation and drizzle, and hence increase the lifetime of clouds, preserving their cooling effect ( [[#Ferek--2000|Ferek et al., 2000]] ). In contrast, cloud lifetime could be decreased by making the smaller droplets more susceptible to evaporation. Modelling studies have shown that a positive ERFaci <sub></sub> (warming) could also result from seeding clouds with too large aerosols ( [[#Pringle--2012|Pringle et al., 2012]] ; [[#Alterskjaer--2013|Alterskjaer and Kristjánsson, 2013]] ). These individual and combined processes are not well understood, and may have a limited representation in models, or counteracting errors ( [[#Mülmenstädt--2018|Mülmenstädt and Feingold, 2018]] ), lending ''low'' to ''medium confidence'' to the ERF estimates.

Modelled ERFaci associated with cirrus cloud thinning (CCT) cover a wide range in the literature, and the maximum are of the order of –0.8 to –3.5 W m <sup>–2</sup> , though they are of ''low confidence'' , with some studies using more simplified representations ( [[#Mitchell--2009|Mitchell and Finnegan, 2009]] ; [[#Storelvmo--2013|Storelvmo et al., 2013]] ; [[#Kristjánsson--2015|Kristjánsson et al., 2015]] ; [[#Jackson--2016|Jackson et al., 2016]] ; [[#Muri--2018|Muri et al., 2018]] ; [[#Gasparini--2020|Gasparini et al., 2020]] ). ERFaci for CCT is mainly affected by particle seeding concentrations, with an optimum around 20 L <sup>–1</sup> , according to limited evidence from models ( [[#Storelvmo--2013|Storelvmo et al., 2013]] ). Seeding leading to higher particle concentrations could lead to a warming ( [[#Storelvmo--2013|Storelvmo et al., 2013]] ; [[#Penner--2015|Penner et al., 2015]] ; [[#Gasparini--2016|Gasparini and Lohmann, 2016]] ). The lack of representation of processes related to, for example, heterogeneous and homogeneous freezing and their prevalence, is a dominant source of uncertainty in ERF estimates, in addition to less research activity.

In summary, the aerosol and cloud microphysics involved with SRM are not well understood, notably due to insufficient (and varying degrees of) representation of relevant processes in models. ERF of up to several W m <sup>–2</sup> is reported in the literature, with SAI at the higher end and CCT with lower potentials, though it remains a challenge to establish ERF potentials and efficacies with confidence. Modelling studies have been published with more sophisticated treatment of SRM since AR5, but the uncertainties, such as cloud–aerosol radiation interactions, remain large ( ''high confidence'' ).

<div id="6.5" class="h1-container"></div>

<span id="implications-of-changing-climate-on-aq"></span>