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

== 6.2 Global and Regional Temporal Evolution of SLCF Emissions ==

<div id="h1-3-siblings" class="h1-siblings"></div>

SLCF emissions originate from a variety of sources driven by anthropogenic activities and natural processes. The natural sources include vegetation, soil, fire, lightning, volcanoes and oceans. Changes in SLCF emissions from natural systems occur either due to human activities, such as land-use change, or due to global changes. Their sensitivity to climate change thus induces climate feedbacks (see Section 6.4.5 for a quantification of these feedbacks). This section reviews the current understanding of historical emissions for anthropogenic, natural, and open biomass burning sources. A detailed discussion of methane sources, sinks, trends are provided in Chapter 5, [[IPCC:Wg1:Chapter:Chapter-5#5.2.2|Section 5.2.2]] .

<div id="6.2.1" class="h2-container"></div>

<span id="anthropogenic-sources"></span>
=== 6.2.1 Anthropogenic Sources ===

<div id="h2-10-siblings" class="h2-siblings"></div>

Estimates of global anthropogenic (human-caused) SLCF emissions and their historical evolution that were used in AR5 (CMIP5; [[#Lamarque--2010|Lamarque et al., 2010]] ) have been revised for use in CMIP6 ( [[#Hoesly--2018|Hoesly et al., 2018]] ). The update considered new data and assessment of the impact of environmental policies, primarily regarding air pollution control (R. Wang et al. , 2014; S.X. Wang et al. , 2014; Montzka et al. , 2015; Crippa et al. , 2016; Turnock et al. , 2016; Klimont et al. , 2017a; Zanatta et al. , 2017; Prinn et al. , 2018) . Additionally, [[#Hoesly--2018|Hoesly et al. (2018)]] have extended estimates of anthropogenic emissions back to 1750 and developed an updated and new set of spatial proxies allowing for more differentiated (source sector-wise) gridding of emissions ( [[#Feng--2020|Feng et al., 2020]] ). The CMIP6 emissions inventory has been developed with the Community Emissions Data System (CEDS) that improves upon existing inventories with a more consistent and reproducible methodology, similar to approaches used in, for example, the EDGAR database ( [[#Crippa--2016|Crippa et al., 2016]] ) and the GAINS model ( [[#Amann--2011|Amann et al., 2011]] ; [[#Klimont--2017a|Klimont et al., 2017a]] ; [[#Höglund-Isaksson--2020|Höglund-Isaksson et al., 2020]] ) where emissions of all compounds are consistently estimated using the same emissions drivers and propagating individual components (activity data and emissions factors) separately to capture fuel and technology trends affecting emissions trajectories over time. This contrasts with the approach used to establish historical emissions for CMIP5 where different datasets available at the time were combined. The CMIP6 exercise is based on the first release of the CEDS emissions dataset (version 2017-05-18, sometimes referred to hereafter as CMIP6 emissions) whose main features regarding SLCFs are described hereafter. The CEDS has been and will be regularly updated and extended; the recent update of the CEDS ( [[#Hoesly--2019|Hoesly et al., 2019]] ) and consequences for this Assessment is discussed when necessary. Some details on how SLCF emissions have been represented in scenarios used by IPCC assessments can be found in [[IPCC:Wg1:Chapter:Chapter-1|Chapter 1]] ( [[IPCC:Wg1:Chapter:Chapter-1#1.6.1|Section 1.6.1]] and Cross-Chapter Box 1.4 and in Section 6.7.1.1).

For most of the SLCF species, the global and regional anthropogenic emissions trends developed for CMIP6 for the period 1850–2000 are not substantially different from those used in CMIP5 (Figures 6.18 and 6.19) despite the different method used to derive them. Hoesly et al. (2018, CEDS) developed independent time series capturing trends in fuel use, technology and level of control, whereas CMIP5 combined different emissions datasets. However, for the period after 1990, the CMIP6 dataset shows for all species, except for SO <sub>2</sub> , CO, and (since 2011) for NO <sub>x</sub> , a different trend than CMIP5 (i.e., continued strong growth of emissions driven primarily by developments in Asia (Figure 6.19)). The unprecedented growth of emissions from Eastern and Southern Asia since 2000 changed the global landscape of emissions, making Asia the dominant SLCF source region (Figures 6.3 and 6.19). The Representative Concentration Pathways (RCP) scenarios used in AR5 started from the year 2000 ( [[#van%20Vuuren--2011|van Vuuren et al., 2011]] ) and did not capture the SLCF emissions which actually occurred until 2015. The CEDS inventory ( [[#Hoesly--2018|Hoesly et al., 2018]] ) includes improved representation of these trends and the estimate for 2014. These findings have been largely supported by several independent emissions inventory studies and remote-sensing data analysis. However, for the last decade the decline of Asian emissions of SO <sub>2</sub> and NO <sub>x</sub> appears underestimated while growth of BC and OC emissions in Asia and Africa seems overestimated in CMIP6, compared to most recent regional evaluations ( [[#Klimont--2017a|Klimont et al., 2017a]] ; [[#Zheng--2018b|Zheng et al., 2018b]] ; [[#Elguindi--2020|Elguindi et al., 2020]] ; [[#Kanaya--2020|Kanaya et al., 2020]] ; [[#McDuffie--2020|McDuffie et al., 2020]] ), which are largely considered in the updated release of the CEDS ( [[#Hoesly--2019|Hoesly et al., 2019]] ). Consequently, global CMIP6 anthropogenic emissions for 2014 are likely overestimated by about 10% for SO <sub>2</sub> and NO <sub>x</sub> and by about 15% for BC and OC.

For SO <sub>2</sub> , independent emissions inventories and observational evidence show that on a global scale strong growth of Asian emissions has been countered by reduction in North America and Europe (Reis et al. , 2012; Amann et al. , 2013; Crippa et al. , 2016; Aas et al. , 2019) . However, Chinese emissions declined by nearly 70% between about 2006 and 2017 ( ''high confidence'' ) (Silver et al. , 2018; Zheng et al. , 2018b; Mortier et al. , 2020; Tong et al. , 2020) . The estimated reduction in China contrasts with continuing strong growth of SO <sub>2</sub> emissions in Southern Asia (Figure 6.19). In 2014, over 80% of anthropogenic SO <sub>2</sub> emissions originated from power plants and industry, with Asian sources contributing more than 50% of the total (Figure 6.3).

<div id="_idContainer013" class="Basic-Text-Frame"></div>

[[File:b69959c7568e4a2c1f6ea659d4ec8885 IPCC_AR6_WGI_Figure_6_3.png]]

'''Figure 6.3 |''' '''Relative regional and sectoral contributions to the present day (year 2014) anthropogenic emissions of short-lived climate forcers (SLCFs).''' Emissions data are from the Community Emissions Data System (CEDS; [[#Hoesly--2018|Hoesly et al., 2018]] ). Emissions are aggregated into the following sectors: fossil fuel production and distribution (coal mining, oil and gas production, upstream gas flaring, gas distribution networks), fossil fuel combustion for energy (power plants), residential and commercial (fossil and biofuel use for cooking and heating), industry (combustion and production processes, solvent-use losses from production and end use), transport (road and off-road vehicles), shipping (including international shipping), aviation (including international aviation), agriculture (livestock and crop production), waste management (solid waste, including landfills and open trash burning, residential and industrial waste water), and other. Further details on data sources and processing are available in the chapter data table (Table 6.SM.3).

Global emissions of NO <sub>x</sub> have been growing in spite of the successful reduction of emissions in North America, Europe, Japan and Korea ( [[#Crippa--2016|Crippa et al., 2016]] ; [[#Turnock--2016|Turnock et al., 2016]] ; [[#Miyazaki--2017|Miyazaki et al., 2017]] ; [[#Jiang--2018|Jiang et al., 2018]] ), partly driven by continuous efforts to strengthen the emissions standards for road vehicles in most countries (Figures 6.18 and 6.19). In many regions, an increase in vehicle fleet as well as non-compliance with emissions standards (Anenberg et al. , 2017, 2019; Jonson et al. , 2017; Jiang et al. , 2018) , growing aviation ( [[#Grewe--2019|Grewe et al., 2019]] ; [[#Lee--2021|Lee et al., 2021]] ) and demand for energy, and consequently a large number of new fossil fuel power plants, have more than compensated for these reductions. Since about 2011, global NO <sub>x</sub> emissions appear to have stabilized or slightly declined ( ''medium confidence'' ) but the global rate of decline has been underestimated in the CEDS, as recent data suggest that emissions reductions in China were larger than included in the CEDS (Figure 6.19 and [[#Hoesly--2018|Hoesly et al., 2018]] ). Recent bottom-up emissions estimates ( [[#Zheng--2018b|Zheng et al., 2018b]] ) largely confirm what has been shown in satellite data (F. [[#Liu--2016|]] [[#Liu--2016|Liu et al., 2016]] ; [[#Miyazaki--2017|Miyazaki et al., 2017]] ; [[#Silver--2018|Silver et al., 2018]] ): a strong decline of NO <sub>2</sub> column over eastern China ( ''high confidence'' ) (Section 6.3.3.1). At a global level, the estimated CEDS CO emissions trends are comparable to NO <sub>x</sub> , which has been confirmed by several inverse modelling studies (Section 6.3.3.2). The transport sector (including international shipping and aviation) was the largest anthropogenic source of NO <sub>x</sub> (about 50% of the total) and also contributed over 25% of CO emissions in 2014; Asia represented 50% and North America and Europe about 20% of global total NO <sub>x</sub> and CO emissions (Figure 6.3).

Oil production-distribution and transport sectors have dominated anthropogenic NMVOC emissions for most of the 20th century ( [[#Hoesly--2018|Hoesly et al., 2018]] ) and still represent a large share (Figure 6.3). Efforts to control transport emissions (i.e., increasing stringency of vehicle emissions limits) were largely offset by the fast growth of emissions from chemical industries and solvent use, as well as from fossil fuel production and distribution, resulting in continued growth of global anthropogenic NMVOC emissions since 1900 ( ''high confidence'' ) (Figure 6.18). Since AR5, there is ''high confidence'' that motor vehicle NMVOC emissions have sharply declined in North America and Europe in the last decades ( [[#Rossabi--2018|Rossabi and Helmig, 2018]] ), for example, by about an order of magnitude in major US cities since 1990 ( [[#Bishop--2018|Bishop and Haugen, 2018]] ; [[#McDonald--2018|McDonald et al., 2018]] ). Increasing (since 2008) oil- and gas-extraction activities in North America lead to a strong growth of NMVOC emissions ( ''high confidence'' ) as shown by analysis of ethane column data ( [[#Franco--2016|Franco et al., 2016]] ), but absolute emission amounts remain uncertain ( [[#Pétron--2014|Pétron et al., 2014]] ; [[#Tzompa-Sosa--2019|Tzompa-Sosa et al., 2019]] ). In Eastern Asia, there is ''medium confidence'' in a decreasing trend of motor vehicle emissions, suggested by ambient measurements in Beijing since 2002 ( [[#Wang--2015|Wang et al., 2015]] ) and by bottom-up estimates ( [[#Zheng--2018b|Zheng et al., 2018b]] ), and a decrease in residential heating emissions due to declining coal and biofuel use since 2005 ( [[#Zheng--2018b|Zheng et al., 2018b]] ; [[#Li--2019|]] [[#Li--2019|M Li et al., 2019]] ). However, total anthropogenic NMVOC emissions have increased steadily in China since the mid-20th century, largely due to the growing importance of the solvent-use and industrial sectors ( ''medium evidence'' , ''high agreement'' ) (Sun et al. , 2018; Zheng et al. , 2018b; M. Li et al. , 2019) . Resulting changes in the NMVOC speciated emissions might be underestimated in the current regional and global inventories. For example, in the USA, a recent study suggested an emergent shift in urban NMVOC sources from transportation to chemical products (i.e., household chemicals, personal care products, solvents, etc.), which is not in accordance with emissions inventories currently used ( [[#McDonald--2018|McDonald et al., 2018]] ). In many European regions and cities, wood burning has been increasingly used for residential heating, partly for economic reasons and because it is considered CO <sub>2</sub> -neutral ( [[#Athanasopoulou--2017|Athanasopoulou et al., 2017]] ); in situ measurements in several cities, including Paris, suggest that wood burning explains up to half of the NMVOC emissions during winter ( [[#Kaltsonoudis--2016|Kaltsonoudis et al., 2016]] ; [[#Languille--2020|Languille et al., 2020]] ). Due to the vast heterogeneity of sources and components of NMVOCs, uncertainty in regional emissions and trends is higher than for most other components.

Emissions of carbonaceous aerosols (BC, OC) have been steadily increasing and their emissions have almost doubled since 1950 ( ''medium confidence'' ) ( [[#Hoesly--2018|Hoesly et al., 2018]] ). Before 1950, North America and Europe contributed about half of the global total but successful introduction of diesel particulate filters on road vehicles ( [[#Fiebig--2014|Fiebig et al., 2014]] ; [[#Robinson--2015|Robinson et al., 2015]] ; [[#Klimont--2017a|Klimont et al., 2017a]] ) and declining reliance on solid fuels for heating brought in large reductions ( ''high confidence'' ) (Figure 6.19). Currently, global carbonaceous aerosol emissions originate primarily from Asia and Africa ( [[#Bond--2013|Bond et al., 2013]] ; [[#Hoesly--2018|Hoesly et al., 2018]] ; [[#Elguindi--2020|Elguindi et al., 2020]] ; [[#McDuffie--2020|McDuffie et al., 2020]] ), representing about 80% of the global total ( ''high confidence'' ) (Figure 6.3). Consideration, in CMIP6, of emissions from kerosene lamps and gas flaring, revised estimates for open burning of waste, regional coal consumption, and new estimates for Russia ( [[#Stohl--2013|Stohl et al., 2013]] ; [[#Huang--2015|Huang et al., 2015]] ; [[#Huang--2016|Huang and Fu, 2016]] ; [[#Kholod--2016|Kholod et al., 2016]] ; [[#Conrad--2017|Conrad and Johnson, 2017]] ; [[#Evans--2017|Evans et al., 2017]] ; [[#Klimont--2017a|Klimont et al., 2017a]] ) resulted in over 15% higher global emissions of OC and BC than in the CMIP5 estimates for the first decade of the 21st century (Figure 6.18). However, the continued increase of BC emissions over Eastern Asia after 2005, estimated in CMIP6 (Figure 6.19), has been questioned recently as a steady decline of BC concentrations was measured in the air masses flowing out from the east coast of China ( [[#Kanaya--2020|Kanaya et al., 2020]] ), which has been also estimated in recent regional bottom-up and top-down inventories (Zheng et al. , 2018a; Elguindi et al. , 2020; McDuffie et al. , 2020) . Since AR5, confidence in emissions estimates and trends in North America and Europe has increased, but high uncertainties remain for Asia and Africa, despite their major contribution to global emissions. The size distribution of emitted species, of importance for climate and health impacts, remains uncertain and the CEDS inventory does not provide such information. Overall, a factor two uncertainty in global estimates of BC and OC emissions remains, with post-2005 emissions overestimated in Asia ( ''high confidence'' ) and Africa ( ''medium confidence'' ).

Bottom-up global emissions estimates of methane ( [[#Lamarque--2010|Lamarque et al., 2010]] ; [[#Hoesly--2018|Hoesly et al., 2018]] ; [[#Janssens-Maenhout--2019|Janssens-Maenhout et al., 2019]] ; [[#Höglund-Isaksson--2020|Höglund-Isaksson et al., 2020]] ) for the last two decades are higher than top-down assessments (e.g., [[#Saunois--2016|Saunois et al., 2016]] , 2020) but trends from the two methods are similar and indicate continued growth ( ''high confidence'' ). Larger discrepancies exist at the sectoral and regional levels, notably for coal mining ( [[#Peng--2016|Peng et al., 2016]] ; [[#Miller--2019|Miller et al., 2019]] ) and the oil and gas sector due to the growth of unconventional production and higher loss estimates [[IPCC:Wg1:Chapter:Chapter-5#5.2.2|Section 5.2.2]] ; Franco et al. , 2016; Alvarez et al. , 2018; Dalsøren et al. , 2018).

Agricultural production (livestock and mineral nitrogen fertilizer application) is the primary source of ammonia in the atmosphere with more than half of present-day emissions originating in Asia (Hoesly et al. , 2018; Figure 6.3, [[#EC-JRC/PBL--2020|EC-JRC/PBL, 2020]] ; Vira et al. , 2020) . NH <sub>3</sub> emissions are estimated to have grown strongly since 1850, especially since 1950, driven by continuously increasing livestock production, widespread application of mineral nitrogen fertilizers, and lack of action to control ammonia ( ''high confidence'' ) (Erisman et al. , 2008; Riddick et al. , 2016; Hoesly et al. , 2018; Fowler et al. , 2020) . The trends estimated in CMIP5 and CMIP6 are similar, while in absolute terms, CMIP6 has somewhat higher emissions as it includes emissions from wastewater and human waste that were largely missing in CMIP5 ( [[#Hoesly--2018|Hoesly et al., 2018]] ). CMIP6 has improved spatial and temporal distribution of emissions ( [[#Lamarque--2013a|Lamarque et al., 2013a]] ) relying on the EDGAR v4.3 database and [[#Paulot--2014|Paulot et al. (2014)]] , but important uncertainties remain for regionally specific temporal patterns (Riddick et al. , 2016; Liu et al. , 2019; Feng et al. , 2020; Vira et al. , 2020) . The continuing increase in global NH <sub>3</sub> emissions is driven primarily by growing livestock and crop production in Asia while emissions in the USA and Europe remain about constant or have slightly declined in the last decade ( [[#Hoesly--2018|Hoesly et al., 2018]] ). Recent satellite and ground observations support trends estimated in CMIP6 dataset (Section 6.3.3.4).

To summarize, there are significant differences in spatial and temporal patterns of SLCF emissions across global regions (Figure 6.18). Until the 1950s, the majority of SLCF emissions were associated with fossil fuel use (SO <sub>2</sub> , NO <sub>x</sub> , NMVOCs, CO) and about half of BC and OC originated from North America and Europe ( [[#Lamarque--2010|Lamarque et al., 2010]] ; [[#Hoesly--2018|Hoesly et al., 2018]] ). Since the 1990s a large redistribution of emissions was associated with strong economic growth in Asia and declining emissions in North America and Europe due to air-quality legislation and the declining capacity of energy-intensive industry; currently more than 50% of anthropogenic emissions of each SLCF species (including methane and NH <sub>3</sub> ) originates from Asia (Figure 6.3; Amann et al. , 2013; Bond et al. , 2013; Fiore et al. , 2015; Crippa et al. , 2016, 2018; Klimont et al. , 2017a; Hoesly et al. , 2018) . The dominance of Asia for SLCF emissions is corroborated by growing remote-sensing capacity that has been providing an independent evaluation of estimated pollution trends in the last decade (Duncan et al. , 2013; Lamsal et al. , 2015; Luo et al. , 2015; Fioletov et al. , 2016; Geddes et al. , 2016; Irie et al. , 2016; Krotkov et al. , 2016; Wen et al. , 2018) .

Since AR5, the quality and completeness of activity and emission-factor data and applied methodology, including spatial allocation together with independent satellite-derived observations, have improved, raising confidence in methods used to derive emissions. There is ''high confidence'' in the sign of global trends of SLCF emissions until the year 2000. However, only ''medium confidence'' for the rate of change in the two last decades, owing primarily to uncertainties in the actual application of reduction technologies in fast-growing economies of Asia. At a regional level, bottom-up derived SLCF emissions trends, and magnitudes in regions with strong economic growth and changing air-quality regulation, are highly uncertain but can be better constrained with top-down methods (Section 6.3). For most SLCF species, there is ''high confidence'' in trends and magnitudes for affluent countries from the Organisation for Economic Co- operation and Development (OECD) regions where accurate and detailed information about drivers of emissions exists; ''medium confidence'' is assessed for regional emissions of NH <sub>3</sub> , methane and NMVOC.

<div id="6.2.2" class="h2-container"></div>

<span id="emissions-by-natural-systems"></span>
=== 6.2.2 Emissions by Natural Systems ===

<div id="h2-11-siblings" class="h2-siblings"></div>

This section assesses our current understanding of SLCF emissions by natural systems. Many naturally occurring emission processes in the Earth system have been perturbed by the growing influence of human activities either directly (e.g., deforestation, agriculture) or via human-induced atmospheric CO <sub>2</sub> increase and climate change, and therefore cannot be considered as purely natural emissions. The temporal evolution and spatial distribution of natural SLCF emissions are highly variable and their estimates rely on models with rather uncertain parametrizations for production mechanisms. For some SLCFs, the natural processes by which emissions occur are also not well understood. In the following sections, we assess the level of confidence in present-day SLCF emissions by natural systems, in their perturbation since the pre-industrial period and their sensitivity to future changes. When available, the assessment also includes estimates from the CMIP6 model ensemble. Note that volcanic SO <sub>2</sub> emissions are discussed in [[IPCC:Wg1:Chapter:Chapter-2#2.2.2|Section 2.2.2]] and natural sources of methane and N <sub>2</sub> O are assessed in Sections 5.2.2 and 5.2.3.

<div id="6.2.2.1" class="h3-container"></div>

<span id="lightning-no-x"></span>
==== 6.2.2.1 Lightning NO <sub>x</sub> ====

<div id="h3-1-siblings" class="h3-siblings"></div>

Lightning contributes approximately 10% of the total NO <sub>x</sub> emissions ( [[#Murray--2016|Murray, 2016]] ). Since lightning NO <sub>x</sub> (LNO <sub>x</sub> ) is predominantly released in the upper troposphere, it has a disproportionately large impact on ozone and OH, and on the lifetime of methanecompared with surface NO <sub>x</sub> emissions. Whereas the global spatial and temporal distribution of lightning flashes can be characterized thanks to satellite-borne and ground sensors ( [[#Virts--2013|Virts et al., 2013]] ; [[#Cecil--2014|Cecil et al., 2014]] ), constraining the amount of NO <sub>x</sub> produced per flash (Miyazaki et al. , 2014; Medici et al. , 2017; Nault et al. , 2017; Marais et al. , 2018; D.J. Allen et al. , 2019; Bucsela et al. , 2019) and its vertical allocation ( [[#Koshak--2014|Koshak et al., 2014]] ; [[#Medici--2017|Medici et al., 2017]] ) has been more elusive.

Atmospheric Chemistry and Climate Model Intercomparison Project (ACCMIP) models in CMIP5 used a range of LNO <sub>x4</sub> between 1.2–9.7 TgN y <sup>−1</sup> (Lamarque et al., 2013b). In CMIP6, the corresponding LNO <sub>x</sub> range is between 3.2–7.6 TgN y <sup>−1</sup> ( [[#Griffiths--2021|Griffiths et al., 2021]] ). All CMIP6 models (as well as most models included in CMIP5, [[#Young--2013|Young et al., 2013]] ) apply a parametrization that relates cloud-top height to lightning intensity ( [[#Price--1992|Price and Rind, 1992]] ), projecting an increase in LNO <sub>x</sub> in a warmer world in the range of 0.27–0.61 TgN yr <sup>−1</sup> °C <sup>–1</sup> ( [[#Thornhill--2021a|Thornhill et al., 2021a]] ). However, models using parametrizations based on convection ( [[#Grewe--2001|Grewe et al., 2001]] ), updraft mass flux ( [[#Allen--2002|Allen and Pickering, 2002]] ) or ice flux ( [[#Finney--2016a|Finney et al., 2016a]] ) show either much less sensitivity or a negative response ( [[#Finney--2016b|Finney et al., 2016b]] , 2018; [[#Clark--2017|Clark et al., 2017]] ).

In summary, the total present-day global lightning NO <sub>x</sub> emissions are still estimated to be within a factor of two. There is ''high confidence'' that LNO <sub>x</sub> are perturbed by climate change; however, there is ''low confidence'' in the sign of the change due to fundamental uncertainties in parametrizations. <sub></sub>

<div id="6.2.2.2" class="h3-container"></div>

<span id="no-x-emissions-from-soils"></span>
==== 6.2.2.2 NO <sub>x</sub> emissions from Soils ====

<div id="h3-2-siblings" class="h3-siblings"></div>

Soil NO <sub>x</sub> (SNO <sub>x</sub> ) emissions occur in connection with complex biogenic/microbial nitrification and denitrification processes ( [[#Ciais--2013|Ciais et al., 2013]] ), which in turn are sensitive – in a non-linear manner – to temperature, precipitation, soil moisture, carbon and nutrient content, and the biome itself (e.g., [[#Hudman--2012|Hudman et al., 2012]] ). Global SNO <sub>x</sub> estimates, based on observationally constrained chemistry-transport model and vegetation model studies, show a broad range between 4.7–16.8 TgN yr <sup>–1</sup> ( [[#Young--2018|Young et al., 2018]] ). This estimate is generally larger than the current source strength used in CMIP6 simulations, which is prescribed using an early empirical estimate, typically scaled to about 5 TgN yr <sup>–1</sup> ( [[#Yienger--1995|Yienger and Levy, 1995]] ).

By the end of the 21st century, the overall nitrogen fixation in non-agricultural ecosystems could be 40% larger than in 2000, due to increased enzyme activity with growing temperatures, but the emission rates of NO (and N <sub>2</sub> O) could be dominated by changes in precipitation patterns and evapotranspiration fluxes ( [[#Fowler--2015|Fowler et al., 2015]] ). Current Earth system models (ESMs) incorporate biophysical and biogeochemical processes only to a limited extent ( [[#Jia--2019|Jia et al., 2019]] ), precluding adequate climate sensitivity studies for SNO <sub>x</sub> . Hence, while the current strength source of soil NO <sub>x</sub> has been better constrained over the last decade, adequate representations of SNO <sub>x</sub> and how it escapes from the canopy, which could provide quantitative estimates of climate-driven changes in SNO <sub>x</sub> , are still missing in ESMs.

<div id="6.2.2.3" class="h3-container"></div>

<span id="vegetation-emissions-of-organic-compounds"></span>
==== 6.2.2.3 Vegetation Emissions of Organic Compounds ====

<div id="h3-3-siblings" class="h3-siblings"></div>

A wide range of BVOCs are emitted from vegetation with the dominant compounds being isoprene and monoterpenes but also including sesquiterpenes, alkenes, alcohols, aldehydes and ketones. The photooxidation of BVOC emissions plays a fundamental role in atmospheric composition by controlling the regional and global budgets of ozone and organic aerosols, and impacting the lifetime of methane and other reactive components ( [[#Arneth--2010b|Arneth et al., 2010b]] ; [[#Heald--2015|Heald and Spracklen, 2015]] ). Substantial uncertainty exists across different modelling frameworks for estimates of global total BVOC emissions and individual compound emissions ( [[#Messina--2016|Messina et al., 2016]] ). Global isoprene emissions estimates differ by a factor of two from 300–600 TgC yr <sup>–1</sup> and global monoterpene emissions estimates by a factor of five from 30–150 TgC yr <sup>–1</sup> ( [[#Messina--2016|Messina et al., 2016]] ). A main driver of the uncertainty ranges is the choice of basal emissions rates assigned to different plant functional types in the model; however, the smaller uncertainty range for isoprene than for monoterpenes is not fully understood ( [[#Arneth--2008|Arneth et al., 2008]] ). The evaluation of global BVOC emissions is challenging because of poor measurement data coverage in many regions and the lack of year-round measurements ( [[#Unger--2013|Unger et al., 2013]] ). Several observational approaches have been developed in the past few years to improve understanding of BVOC emissions, including indirect methods such as the measurement of the OH loss rate in forested environments ( [[#Yang--2016|Yang et al., 2016]] ) and application of the variability in satellite formaldehyde concentrations ( [[#Palmer--2006|Palmer et al., 2006]] ; [[#Barkley--2013|Barkley et al., 2013]] ; [[#Stavrakou--2014|Stavrakou et al., 2014]] ). Direct space-borne isoprene retrievals using infrared radiance (IR) measurements have very recently become available ( [[#Fu--2019|Fu et al., 2019]] ; [[#Wells--2020|Wells et al., 2020]] ). Collectively these approaches have identified weaknesses in the ability of the parametrizations in global models to reproduce BVOC emissions hotspots ( [[#Wells--2020|Wells et al., 2020]] ). However, none of the current observational approaches have yet been able to reduce the uncertainty ranges in global emissions estimates.

At the plant level, BVOC emissions rates and composition depend strongly on plant species with plants tending to emit either isoprene or monoterpenes but not both. Photosynthetic activity is a main driver of isoprene and monoterpene production. Therefore, radiation and temperature, along with leaf-water status, phenological state and atmospheric CO <sub>2</sub> mixing ratio, affect emissions directly (on the leaf scale) and indirectly (via plant productivity; Guenther et al. , 2012; Loreto et al. , 2014; Niinemets et al. , 2014) . CO <sub>2</sub> directly influences the isoprene-synthesis process, with inhibition under increasing atmospheric CO <sub>2</sub> ( [[#Rosenstiel--2003|Rosenstiel et al., 2003]] ; [[#Possell--2005|Possell et al., 2005]] ; [[#Wilkinson--2009|Wilkinson et al., 2009]] ). Direct CO <sub>2</sub> inhibition has been observed for some monoterpene compounds ( [[#Loreto--2001|Loreto et al., 2001]] ; [[#Llorens--2009|Llorens et al., 2009]] ). Severe/long-term water stress may reduce emissions whilst mild/short-term water stress may temporarily amplify or maintain BVOC emissions to protect plants against ongoing stress ( [[#Peñuelas--2010|Peñuelas and Staudt, 2010]] ; [[#Potosnak--2014|Potosnak et al., 2014]] ; [[#Genard-Zielinski--2018|Genard-Zielinski et al., 2018]] ). Furthermore, observations in the Amazon indicate that the chemical composition of monoterpene emissions could also change under elevated temperature conditions ( [[#Jardine--2016|Jardine et al., 2016]] ). In addition, all these processes are investigated over short time scales but the long-term response of BVOC emissions depends on how the vegetation itself responds to the altered climate state (including temperature and water stress).

Global BVOC emissions are highly sensitive to environmental changes including changes in climate, atmospheric CO <sub>2</sub> <sub>,</sub> and vegetation composition and cover changes in natural and managed lands. Recent global modelling studies agree that global isoprene emissions have declined since the pre-industrial period, driven predominantly by anthropogenic land-use and land-cover change (LULCC) with results converging on a 10–25% loss of isoprene emissions between 1850 and the present day ( [[#Lathière--2010|Lathière et al., 2010]] ; [[#Unger--2013|Unger, 2013]] , 2014; [[#Acosta%20Navarro--2014|Acosta Navarro et al., 2014]] ; [[#Heald--2016|Heald and Geddes, 2016]] ; [[#Hantson--2017|Hantson et al., 2017]] ; [[#Hollaway--2017|Hollaway et al., 2017]] ; [[#Scott--2017|Scott et al., 2017]] ). The historical evolution of monoterpene and sesquiterpene emissions is less well studied and there is no robust consensus on even the sign of the change ( [[#Acosta%20Navarro--2014|Acosta Navarro et al., 2014]] ; [[#Hantson--2017|Hantson et al., 2017]] ). Future global isoprene and monoterpene emissions depend strongly on the climate and land-use scenarios considered ( [[#Hantson--2017|Hantson et al., 2017]] ; [[#Szogs--2017|Szogs et al., 2017]] ). BVOC emissions will be sensitive to future land-based climate change mitigation strategies including afforestation and bioenergy, with impacts of bioenergy depending on the choice of crops ( [[#Szogs--2017|Szogs et al., 2017]] ).

Most CMIP6 models use overly simplistic parametrizations and project an increase in global BVOC emissions in response to warming temperatures ( [[#Turnock--2020|Turnock et al., 2020]] ). This good agreement actually reflects the lack of diversity in BVOC-emissions parametrizations in global models that do not fully account for the complex processes influencing emissions that are discussed above.

Overall, we assess that historical global isoprene emissions declined between the pre-industrial period and the present day by 10–25% ( ''low confidence'' ) but historical changes in global monoterpenes and sesquiterpenes are too uncertain to provide an assessment. Future changes in BVOCs depend strongly on the evolution of climate and land use and are strongly sensitive to land-based climate change mitigation strategies. However, the net response of BVOC emissions is uncertain due to the complexity of processes that are hard to constrain observationally and are considered with various degrees of details in models.

<div id="6.2.2.4" class="h3-container"></div>

<span id="land-emissions-of-dust-particles"></span>
==== 6.2.2.4 Land Emissions of Dust Particles ====

<div id="h3-4-siblings" class="h3-siblings"></div>

The emission of dust particles into the atmosphere results from a natural process, namely saltation bombardment of the soil by large wind-blown particles, such as sand grains, and from disintegration of saltating particle clusters ( [[#Kok--2012|Kok et al., 2012]] ). The occurrence and intensity of dust emissions are controlled by soil properties, vegetation and near-surface wind, making dust emissions sensitive to climate change and LULCC ( [[#Jia--2019|Jia et al., 2019]] ). In addition, dust can be directly emitted through human activities, such as agriculture, off-road vehicles, building construction and mining, and indirectly emitted through hydrological changes due to human actions such as water diversion for irrigation (e.g., [[#Ginoux--2012|Ginoux et al., 2012]] ). Estimates of the anthropogenic fraction of global dust vary from less than 10% to over 60% suggesting that the human contribution to the global dust budget is quite uncertain ( [[#Ginoux--2012|Ginoux et al., 2012]] ; [[#Stanelle--2014|Stanelle et al., 2014]] ; [[#Xi--2016|Xi and Sokolik, 2016]] ). Reconstruction of global dust (deposition) from paleo records indicate factor of two to four changes between the different climate regimes in the glacial and interglacial periods ( [[IPCC:Wg1:Chapter:Chapter-2#2.2.6|Section 2.2.6]] ).

An extremely limited number of studies have explored the evolution of global dust sources since pre-industrial times ( [[#Mahowald--2010|Mahowald et al., 2010]] ; [[#Stanelle--2014|Stanelle et al., 2014]] ). A modelling study estimated a 25% increase in global dust emissions between the late 19th century and the present, due to agricultural land expansion and climate change ( [[#Stanelle--2014|Stanelle et al., 2014]] ). CMIP5 models were unable to capture the observed variability of annual and longer time scales in North African dust emissions ( [[#Evan--2014|Evan et al., 2014]] ), however, more recent ESMs with process-based dust emissions schemes that account for changes in vegetation and climate in a more consistent manner, better match the observations ( [[#Kok--2014|Kok et al., 2014]] ; [[#Evans--2016|Evans et al., 2016]] ). Feedbacks between the global dust cycle and the climate system (Section 6.4.5) could account for a substantial fraction of the total aerosol feedbacks in the climate system with an order of magnitude enhancement on a regional scale ( [[#Kok--2018|Kok et al., 2018]] ). In summary, there is ''high confidence'' that atmospheric dust source and loading are sensitive to changes in climate and land use, however, there is ''low confidence'' in quantitative estimates of dust emission response to climate change.

<div id="6.2.2.5" class="h3-container"></div>

<span id="oceanic-emissions-of-marine-aerosols-and-precursors"></span>
==== 6.2.2.5 Oceanic Emissions of Marine Aerosols and Precursors ====

<div id="h3-5-siblings" class="h3-siblings"></div>

Oceans are a significant source of marine aerosols that influence climate directly by scattering and absorbing solar radiation or indirectly through the formation of cloud condensation nuclei (CCN) and ice nucleating particles (INPs). Marine aerosols consist of primary sea-spray particles and secondary aerosols produced by the oxidation of emitted precursors, such as dimethylsulphide (DMS) and numerous other BVOCs. Sea-spray particles, composed of sea salt and primary organic aerosols (POA), are produced by wind-induced wave breaking as well as the direct mechanical disruption of waves. The understanding of sea-spray emissions has increased substantially over the last five years, however, the knowledge of formation pathways and factors influencing their emissions continue to have large uncertainties ( [[#Forestieri--2018|Forestieri et al., 2018]] ; [[#Saliba--2019|Saliba et al., 2019]] ). The emission rate of sea-spray particles is predominantly controlled by wind speed. Since AR5, the influence of other factors, including sea surface temperature, wave history and salinity is increasingly evident (Callaghan et al. , 2014; Grythe et al. , 2014; Ovadnevaite et al. , 2014; Salter et al. , 2014; Barthel et al. , 2019) . Marine POA, often the dominant submicron component of sea spray, are emitted as a result of oceanic biological activity, however the biological processes by which these particles are produced remain poorly characterized contributing to large uncertainties in global marine POA emissions estimates (Tsigaridis et al. , 2014; Cravigan et al. , 2020; Hodzic et al. , 2020) '''.''' Furthermore, the particle size and chemical composition of sea-spray particles, and how these evolve in response to changing climate factors and dynamic oceanic biology, continue to have large uncertainties.

DMS, the largest natural source of sulphur in the atmosphere, is produced by marine phytoplankton and is transferred from ocean water to the atmosphere due to wind-induced mixing of surface water. DMS oxidizes to produce sulphate aerosols and contributes to the formation of CCN. Since AR5, the range in global DMS flux estimates reduced from 10–40 TgS yr <sup>–1</sup> to 9–34 TgS yr <sup>–1</sup> with a ''very likely'' range of 18–24 TgS yr <sup>–1</sup> based on sea-surface measurements and satellite observations ( [[#Lana--2011|Lana et al., 2011]] ). DMS production, and consequently emissions, have been shown to respond to multiple stressors, including climate warming, eutrophication, and ocean acidification. However, large uncertainties in process-based understanding of the mechanisms controlling DMS emissions, from physiological to ecological, limit our knowledge of past variations and our capacity to project future changes.

Overall, there is ''low confidence'' in the magnitude and changes in marine aerosol emissions in response to shifts in climate and marine ecosystem processes.

<div id="6.2.2.6" class="h3-container"></div>

<span id="open-biomass-burning-emissions"></span>
==== 6.2.2.6 Open Biomass Burning Emissions ====

<div id="h3-6-siblings" class="h3-siblings"></div>

Emissions from open biomass burning (including forest, grassland, peat fires and agricultural waste burning) represent about 30%, 10%, 15% and 40% of present-day global emissions of CO, NO <sub>x</sub> , BC and OC, respectively ( [[#van%20Marle--2017|van Marle et al., 2017]] ; [[#Hoesly--2018|Hoesly et al., 2018]] ). Wildfires also play an important role in several atmospheric chemistry–climate feedback mechanisms ( [[#Bowman--2009|Bowman et al., 2009]] ; [[#Fiore--2012|Fiore et al., 2012]] ) and fire events occurring near populated areas induce severe air pollution episodes ( [[#Marlier--2020|Marlier et al., 2020]] ; [[#Rooney--2020|Rooney et al., 2020]] ; [[#Yu--2020|Yu et al., 2020]] ).

For the last two decades, model-based emissions estimates are constrained by remote-sensing capacity to detect active fires and area burned. In AR5, biomass burning emissions were derived from a satellite product ( [[#Lamarque--2010|Lamarque et al., 2010]] ). Since then, improvements in detection of small fires has enhanced the agreement with higher-resolution and ground-based data on burned area in several regions ( [[#Randerson--2012|Randerson et al., 2012]] ; [[#Mangeon--2015|Mangeon et al., 2015]] ), especially for areas subjected to agricultural waste burning ( [[#Chuvieco--2016|Chuvieco et al., 2016]] , 2019). The updated emissions factors and the contribution of forest versus savanna fires lead to significantly higher global emissions of NO <sub>x</sub> and lower emissions of OC and CO in CMIP6, compared with CMIP5. A recent compilation and assessment of emissions factors ( [[#Andreae--2019|Andreae, 2019]] ) indicates that the emissions factors from [[#Akagi--2011|Akagi et al. (2011)]] , primarily used to produce the CMIP6 datasets, differ by ±50% for CO, OC, BC and NO <sub>x</sub> , depending on the biome, and would imply, for example, up to 10–30% higher OC and BC emissions from tropical forest fires.

The historical (pre-satellite era) dataset for CMIP6 considers advances in knowledge of past fire dynamics (new fire proxy datasets, such as charcoal in sediments and levoglucosan in ice cores) and visibility records from weather stations ( [[#Marlon--2016|Marlon et al., 2016]] ; [[#van%20Marle--2017|van Marle et al., 2017]] ). At a global level, CMIP5 and CMIP6 emissions trends are similar, however, there are substantial differences at the regional level, especially for the USA, South America (south of Amazonia) and Southern Hemisphere Africa ( [[#van%20Marle--2017|van Marle et al., 2017]] ).

Globally, the CMIP5 estimates ( [[#Lamarque--2010|Lamarque et al., 2010]] ), indicated a gradual decline of open biomass burning emissions from 1920 to about 1950 and then steady, and stronger than CMIP6, increase towards 2000. In contrast, CMIP6 biomass burning emissions ( [[#van%20Marle--2017|van Marle et al., 2017]] ) increase only slightly over 1750–2015 – they peak during the 1990s after which they decrease gradually, which is consistent with the assessment of fire trends in Chapter 5. Therefore, the CMIP6 evolution has a smaller difference between pre-industrial and present-day emissions than CMIP5, resulting in a lower radiative forcing of biomass burning SLCFs, possibly leading to a lower effect on climate ( [[#van%20Marle--2017|van Marle et al., 2017]] ).

Climate warming, especially through change in temperature and precipitation, will generally increase the risk of fire ( [[#Jia--2019|Jia et al., 2019]] , see also Chapter 12) and can also affect the fire injection and plume height ( [[#Veira--2016|Veira et al., 2016]] ), but occurrence of fires and their emissions in the future strongly depends on anthropogenic factors, such as population density, land use and fire management ( [[#Veira--2016|Veira et al., 2016]] ). Consequently, future emissions vary widely with increases and decreases amongst the SSP scenarios due to different land-use change scenarios.

In summary, there has been an improvement in the knowledge of biomass burning emissions by reducing key uncertainties highlighted in AR5. However, systematic assessment of remaining uncertainties is limited, with a lower limit of uncertainties due to emissions factors of 30%, and larger uncertainties due to burning-activity estimates, especially at regional level. Overall, a ''medium confidence'' in current global biomass burning SLCF emissions and their evolution over the satellite era is assessed. There is ''low'' to ''medium confidence'' in SLCF emissions from biomass burning from the pre-industrial period to the 1980s, which rely on the incorporation of several proxy data, with limited spatial representativeness. Nevertheless, uncertainties in the absolute value of pre-industrial emissions remain high, limiting confidence in radiative forcing estimates.

<div id="6.3" class="h1-container"></div>

<span id="evolution-of-atmospheric-slcf-abundances"></span>