Editing IPCC:AR6/SRCCL/Chapter-2 (section)

== 2.3 Greenhouse gas fluxes between land and atmosphere ==

<div id="article-2-3-greenhouse-gas-fluxes-between-land-and-atmosphere-block-1"></div>

Land is simultaneously a source and sink for several GHGs. Moreover, both natural and anthropogenic processes determine fluxes of GHGs, making it difficult to separate ‘anthropogenic’ and ‘non-anthropogenic’ emissions and removals. A meeting report by the IPCC (2010 <sup>[[#fn:r495|495]]</sup> ) divided the processes responsible for fluxes from land into three categories: (i) the ''direct effects'' of anthropogenic activity due to changing land cover and land management, (ii) the ''indirect effects'' of anthropogenic environmental change, such as climate change, carbon dioxide (CO <sub>2</sub> ) fertilisation, nitrogen deposition, and (iii) ''natural'' climate variability and natural disturbances (e.g., wildfires, windrow, disease). The meeting report (IPCC 2010) noted that it was impossible with any direct observation to separate direct anthropogenic effects from non-anthropogenic (indirect and natural) effects in the land sector.

As a result, different approaches and methods for estimating the anthropogenic fluxes have been developed by different communities to suit their individual purposes, tools and data availability.

The major GHGs exchanged between land and the atmosphere discussed in this chapter are CO <sub>2</sub> (Section 2.3.1), methane (CH <sub>4</sub> ) (Section 2.3.2) and nitrous oxide (N2O) (Section 2.3.3). We estimate the total emissions from AFOLU to be responsible for approximately 23% of global anthropogenic GHG emissions over the period 2007–2016 (Smith et al. 2013a <sup>[[#fn:r496|496]]</sup> ; Ciais et al. 2013a <sup>[[#fn:r497|497]]</sup> ) (Table 2.2). The estimate is similar to that reported in AR5 (high confidence), with slightly more than half these emissions coming as non-CO <sub>2</sub> GHGs from agriculture. Emissions from AFOLU have remained relatively constant since AR4, although their relative contribution to anthropogenic emissions has decreased due to increases in emissions from the energy sector.

<div id="article-2-3-greenhouse-gas-fluxes-between-land-and-atmosphere-block-2"></div>

<span id="table-2.2"></span>
<!-- START IMG -->
<!-- TABLE IMG -->
<!-- IMG TITLE -->
'''Table 2.2'''

<span id="net-anthropogenic-emissions-due-to-agriculture-forestry-and-other-land-use-afolu-and-non-afolu-average-for-20072016"></span>
<!-- IMG CAPTION -->
'''Net anthropogenic emissions due to Agriculture, Forestry, and other Land Use (AFOLU) and non-AFOLU (average for 2007–2016)'''

Positive value represents emissions; negative value represents removals.
<!-- IMG FILE -->
[[File:81d4f01b49473a5863a598546c0e0273 table2.2.png]]

# Estimates are only given until 2016 as this is the latest date when data are available for all gases.
# Net anthropogenic flux of CO <sub>2</sub> due to land cover change such as deforestation and afforestation, and land management including wood harvest and regrowth, as well as peatland burning, based on two bookkeeping models as used in the Global Carbon Budget and for AR5. Agricultural soil carbon stock change under the same land use is not considered in these models.
# Estimates show the mean and assessed uncertainty of two databases, FAOSTAT and USEPA 2012.
# Total non-AFOLU emissions were calculated as the sum of total CO <sub>2</sub> -eq emissions values for energy, industrial sources, waste and other emissions with data from the Global Carbon Project for CO <sub>2</sub> , including international aviation and shipping and from the PRIMAP database for CH <sub>4</sub> and N <sub>2</sub> O averaged over 2007–2014 only as that was the period for which data were available.
# The natural response of land to human-induced environmental changes is the response of vegetation and soils to environmental changes such as increasing atmospheric CO <sub>2</sub> concentration, nitrogen deposition, and climate change. The estimate shown represents the average from Dynamic Global Vegetation Models.
# All values expressed in units of CO <sub>2</sub> -eq are based on AR5 100-year Global Warming Potential (GWP) values without climate-carbon feedbacks (N <sub>2</sub> O = 265; CH <sub>4</sub> = 28). Note that the GWP has been used across fossil fuel and biogenic sources of methane. If a higher GWP for fossil fuel CH <sub>4</sub> (30 per AR5), then total anthropogenic CH <sub>4</sub> emissions expressed in CO <sub>2</sub> -eq would be 2% greater.

<!-- END IMG -->
<div id="article-2-3-greenhouse-gas-fluxes-between-land-and-atmosphere-block-3"></div>

<span id="figure-2.4"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 2.4'''

<span id="net-and-gross-fluxes-of-co2-from-land-annual-averages-for-20082017.left-the-total-net-flux-of-co2-between-land-and-atmosphere-grey-is-shown-with-its-two-component-fluxes-i-net-afolu-emissions-blue-and-ii-the-net-land-sink-brown-due-to-indirect-environmental-effects-and-natural-effects-on-managed-and-unmanaged-lands."></span>
<!-- IMG CAPTION -->
'''Net and gross fluxes of CO2 from land (annual averages for 2008–2017).Left: The total net flux of CO2 between land and atmosphere (grey) is shown with its two component fluxes, (i) net AFOLU emissions (blue), and (ii) the net land sink (brown), due to indirect environmental effects and natural effects on managed and unmanaged lands. […]'''
<!-- IMG FILE -->
[[File:1fbfd3f647f1eab0ca905af3512853e4 Figure-2.4-1024x599.jpg]]

Net and gross fluxes of CO <sub>2</sub> from land (annual averages for 2008–2017).Left: The total net flux of CO <sub>2</sub> between land and atmosphere (grey) is shown with its two component fluxes, (i) net AFOLU emissions (blue), and (ii) the net land sink (brown), due to indirect environmental effects and natural effects on managed and unmanaged lands. Middle: The gross emissions and removals contributing to the net AFOLU flux. Right: The gross emissions and removals contributing to the land sink.

<!-- END IMG -->
<span id="carbon-dioxide"></span>
=== 2.3.1 Carbon dioxide ===

<div id="section-2-3-1-carbon-dioxide-block-1"></div>

This section is divided into four sub-sections (Figure 2.4): (i) the total net flux of CO <sub>2</sub> between land and atmosphere, (ii) the contributions of AFOLU fluxes and the non-AFOLU land sink to that total net CO <sub>2</sub> flux, (iii) the gross emissions and removals comprising the net AFOLU flux, and (iv) the gross emissions and removals comprising the land sink. Emissions to the atmosphere are positive; removals from the atmosphere are negative.

<div id="section-2-3-1-carbon-dioxide-block-2"></div>

<span id="figure-2.4-1"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 2.4'''

<span id="net-and-gross-fluxes-of-co2-from-land-annual-averages-for-20082017.left-the-total-net-flux-of-co2-between-land-and-atmosphere-grey-is-shown-with-its-two-component-fluxes-i-net-afolu-emissions-blue-and-ii-the-net-land-sink-brown-due-to-indirect-environmental-effects-and-natural-effects-on-managed-and-unmanaged-lands.-1"></span>
<!-- IMG CAPTION -->
'''Net and gross fluxes of CO2 from land (annual averages for 2008–2017).Left: The total net flux of CO2 between land and atmosphere (grey) is shown with its two component fluxes, (i) net AFOLU emissions (blue), and (ii) the net land sink (brown), due to indirect environmental effects and natural effects on managed and unmanaged lands. […]'''
<!-- IMG FILE -->
[[File:1fbfd3f647f1eab0ca905af3512853e4 Figure-2.4-1024x599.jpg]]

Net and gross fluxes of CO <sub>2</sub> from land (annual averages for 2008–2017).Left: The total net flux of CO <sub>2</sub> between land and atmosphere (grey) is shown with its two component fluxes, (i) net AFOLU emissions (blue), and (ii) the net land sink (brown), due to indirect environmental effects and natural effects on managed and unmanaged lands. Middle: The gross emissions and removals contributing to the net AFOLU flux. Right: The gross emissions and removals contributing to the land sink.

<!-- END IMG -->
<div id="section-2-3-1-1-the-total-net-flux-of-co2-between-land-and-atmosphere"></div>

<span id="the-total-net-flux-of-co2-between-land-and-atmosphere"></span>
==== 2.3.1.1 The total net flux of CO2 between land and atmosphere ====

<div id="section-2-3-1-1-the-total-net-flux-of-co2-between-land-and-atmosphere-block-1"></div>

The net effects of all anthropogenic and non-anthropogenic processes on managed and unmanaged land result in a net removal of CO <sub>2</sub> from the atmosphere ( ''high confidence'' ). This total net land-atmosphere removal (defined here as ''the total net land flux'' ) is estimated to have averaged 6.0 ± 2.0 GtCO <sub>2</sub> yr <sup>–1</sup> ( ''likely range'' ) from 2007–2016 (Table 2.3). The estimate is determined from summing the AFOLU and non-AFOLU fluxes due to transient climate change, CO <sub>2</sub> fertilisation and nitrogen deposition calculated by models in the global carbon budget (Le Quéré et al. 2018 <sup>[[#fn:r498|498]]</sup> ), and is consistent with inverse modelling techniques based on atmospheric CO <sub>2</sub> concentrations and air transport (range: 5.1–8.8 GtCO <sub>2</sub> yr <sup>–1</sup> ) (Peylin et al. 2013 <sup>[[#fn:r499|499]]</sup> ; Van Der Laan-Luijkx et al. 2017 <sup>[[#fn:r500|500]]</sup> ; Saeki and Patra 2017 <sup>[[#fn:r501|501]]</sup> ; Le Quéré et al. 2018 <sup>[[#fn:r502|502]]</sup> ) (see Box 2.2 for methods). A recent inverse analysis, considering carbon transport in rivers and oceans, found a net flux of CO <sub>2</sub> for land within this range, but a lower source from southern lands and a lower sink in northern lands (Resplandy et al. 2018 <sup>[[#fn:r503|503]]</sup> ).

The net removal of CO <sub>2</sub> by land has generally increased over the last 60 years in proportion to total emissions of CO <sub>2</sub> ( ''high confidence'' ). Although land has been a net sink for CO <sub>2</sub> since around the middle of last century, it was a net source to the atmosphere before that time, primarily as a result of emissions from AFOLU (Le Quéré et al. 2018 <sup>[[#fn:r504|504]]</sup> ).

<div id="section-2-3-1-2-separation-of-the-total-net-land-flux-into-afolu-fluxes-and-the-land-sink"></div>

<span id="separation-of-the-total-net-land-flux-into-afolu-fluxes-and-the-land-sink"></span>
==== 2.3.1.2 Separation of the total net land flux into AFOLU fluxes and the land sink ====

<div id="section-2-3-1-2-separation-of-the-total-net-land-flux-into-afolu-fluxes-and-the-land-sink-block-1"></div>

The total net flux of carbon between land and the atmosphere can be divided into fluxes due to direct human activities (i.e., AFOLU) and fluxes due to indirect anthropogenic and natural effects (i.e., the land sink) (Table 2.3). These two components are less certain than their sums, the total net flux of CO <sub>2</sub> between atmosphere and land. The land sink, estimated with DGVMs, is least certain (Figure 2.5).

''Fluxes attributed to AFOLU''

The modelled AFOLU flux was a net emission of 5.2 ± 2.6 GtCO <sub>2</sub> yr <sup>–1</sup> ( ''likely range'' ) for 2007–2016, approximately 13% of total anthropogenic CO <sub>2</sub> emissions (Le Quéré et al. 2018 <sup>[[#fn:r505|505]]</sup> ) (Table 2.3). This net flux was due to direct anthropogenic activities, predominately tropical deforestation, but also afforestation/reforestation, and fluxes due to forest management (e.g., wood harvest) and other types of land management, as well as peatland drainage and burning. The AFOLU flux is the mean of two estimates from bookkeeping models (Hansis et al. 2015 <sup>[[#fn:r506|506]]</sup> ; Houghton and Nassikas 2017 <sup>[[#fn:r507|507]]</sup> ), and this estimated mean is consistent with the mean obtained from an assemblage of DGVMs (Le Quéré et al. 2018 <sup>[[#fn:r508|508]]</sup> ) (Box 2.2 and Figure 2.5), although not all individual DGMVs include the same types of land use. Net CO <sub>2</sub> emissions from AFOLU have been relatively constant since 1900. AFOLU emissions were the dominant anthropogenic emissions until around the middle of the last century when fossil fuel emissions became dominant (Le Quéré et al. 2018 <sup>[[#fn:r509|509]]</sup> ). AFOLU activities have resulted in emissions of CO <sub>2</sub> over recent decades ( ''robust evidence, high agreement'' ) although there is a wide range of estimates from different methods and approaches (Smith et al. 2014 <sup>[[#fn:r510|510]]</sup> ; Houghton et al. 2012 <sup>[[#fn:r511|511]]</sup> ; Gasser and Ciais 2013 <sup>[[#fn:r512|512]]</sup> ; Pongratz et al. 2014 <sup>[[#fn:r513|513]]</sup> ; Tubiello et al. 2015 <sup>[[#fn:r514|514]]</sup> ; Grassi et al. 2018 <sup>[[#fn:r515|515]]</sup> ) (Box 2.2, Figure 2.5 and Figure 2.7).

DGVMs and one bookkeeping model (Hansis et al. 2015 <sup>[[#fn:r516|516]]</sup> ) used spatially explicit, harmonised land-use change data (LUH2) (Hurtt et al. 2017 <sup>[[#fn:r517|517]]</sup> ) based on HYDE 3.2. The HYDE data, in turn, are based on changes in the areas of croplands and pastures. In contrast, the Houghton bookkeeping approach (Houghton and Nassikas 2017 <sup>[[#fn:r518|518]]</sup> ) used primarily changes in forest area from the FAO Forest Resource Assessment (FAO 2015 <sup>[[#fn:r519|519]]</sup> ) and FAOSTAT to determine changes in land use. To the extent that forests are cleared for land uses other than crops and pastures, estimates from Houghton and Nassikas (2017 <sup>[[#fn:r520|520]]</sup> , 2018 <sup>[[#fn:r521|521]]</sup> ) are higher than estimates from DGMVs. In addition, both bookkeeping models (Hansis et al. 2015 <sup>[[#fn:r522|522]]</sup> ; Houghton and Nassikas 2017 <sup>[[#fn:r523|523]]</sup> ) included estimates of carbon emissions in Southeast Asia from peat burning from GFED4s (Randerson et al. 2015 <sup>[[#fn:r524|524]]</sup> ) and from peat drainage (Hooijer et al. 2010 <sup>[[#fn:r525|525]]</sup> ).

Satellite-based estimates of CO <sub>2</sub> emissions from losses of tropical forests during 2000–2010 corroborate the modelled emissions but are quite variable; 4.8 GtCO <sub>2</sub> yr <sup>–1</sup> (Tyukavina et al. 2015 <sup>[[#fn:r526|526]]</sup> ), 3.0 GtCO <sub>2</sub> yr <sup>–1</sup> (Harris et al. 2015 <sup>[[#fn:r527|527]]</sup> ), 3.2 GtCO <sub>2</sub> yr <sup>–1</sup> (Achard et al. 2014 <sup>[[#fn:r528|528]]</sup> ) and 1.6 GtCO <sub>2</sub> yr <sup>–1</sup> (Baccini et al. 2017 <sup>[[#fn:r529|529]]</sup> ). Differences in estimates can be explained to a large extent by the different approaches used. For example, the analysis by Tyukavina et al. (2015 <sup>[[#fn:r530|530]]</sup> ) led to a higher estimate because they used a finer spatial resolution. Three of the estimates considered losses in forest area and ignored degradation and regrowth of forests. Baccini et al. (2017 <sup>[[#fn:r531|531]]</sup> ) in contrast, included both losses and gains in forest area and losses and gains of carbon within forests (i.e., forest degradation and growth). The four remote sensing studies cited above also reported committed emissions; in essence, all of the carbon lost from deforestation was assumed to be released to the atmosphere in the year of deforestation. In reality, only some of the carbon in trees is released immediately to the atmosphere at the time of deforestation. The unburned portion is transferred to woody debris and wood products. Both bookkeeping models and DGVMs account for the delayed emissions in growth and decomposition. Finally, the satellite-based estimates do not include changes in soil carbon.

In addition to differences in land-cover data sets between models and satellites, there are many other methodological reasons for differences (Houghton et al. 2012 <sup>[[#fn:r532|532]]</sup> ; Gasser and Ciais 2013 <sup>[[#fn:r533|533]]</sup> ; Pongratz et al. 2014 <sup>[[#fn:r534|534]]</sup> ; Tubiello et al. 2015 <sup>[[#fn:r535|535]]</sup> ) (Box 2.2). There are different definitions of land-cover type, including forest (e.g., FAO uses a tree cover threshold for forests of 10%, Tyukavina et al. (2017 <sup>[[#fn:r536|536]]</sup> ) used 25%), different estimates of biomass and soil carbon density (MgC ha–1), different approaches to tracking emissions through time (legacy effects) and different types of activity included (e.g., forest harvest, peatland drainage and fires). Most DGVMS only recently (since AR5) included forest management processes, such as tree harvesting and land clearing for shifting cultivation, leading to larger estimates of CO <sub>2</sub> emissions than when these processes are not considered (Arneth et al. 2017 <sup>[[#fn:r537|537]]</sup> ; Erb et al. 2018 <sup>[[#fn:r538|538]]</sup> ). Grazing management has likewise been found to have large effects (Sanderman et al. 2017 <sup>[[#fn:r539|539]]</sup> ), and is not included in most DGVMs (Pugh et al. 2015 <sup>[[#fn:r540|540]]</sup> ; Pongratz et al., 2018 <sup>[[#fn:r541|541]]</sup> ).

''Nationally reported greenhouse gas inventories versus global model estimates''

There are large differences globally between estimates of net anthropogenic land-atmosphere fluxes of CO <sub>2</sub> from national GHGIs and from global models, and the same is true in many regions (Figure 2.5). Fluxes reported to the UNFCCC through country GHGIs were noted as about 4.3 GtCO <sub>2</sub> yr <sup>–1</sup> lower (Grassi et al. 2018 <sup>[[#fn:r542|542]]</sup> ) than estimates from the bookkeeping model (Houghton et al. 2012) used in the carbon budget for AR5 (Ciais et al. 2013a <sup>[[#fn:r543|543]]</sup> ). The anthropogenic emissions of CO <sub>2</sub> from AFOLU reported in countries’ GHG inventories were 0.1 ± 1.0 GtCO <sub>2</sub> yr <sup>–1</sup> globally during 2005–2014 (Grassi et al. 2018 <sup>[[#fn:r544|544]]</sup> ) much lower than emission estimates from the two global bookkeeping models of 5.1 ± 2.6 GtCO <sub>2</sub> yr <sup>–1</sup> ( ''likely range'' ) over the same time period (Le Quéré et al. 2018 <sup>[[#fn:r545|545]]</sup> ). Transparency and comparability in estimates can support measuring, reporting and verifying GHG fluxes under the UNFCCC, and also the global stocktake, which will assess globally the progress towards achieving the long-term goals of the Paris Agreement. These differences can be reconciled largely by taking account of the different approaches to defining ‘anthropogenic’ in terms of different areas of land and treatment of indirect environmental change (Grassi et al. 2018 <sup>[[#fn:r546|546]]</sup> ).

To date there has been one study that quantitatively reconciles the global model estimates with GHGIs (Grassi et al. 2018 <sup>[[#fn:r547|547]]</sup> ). The separation of anthropogenic from non-anthropogenic effects is impossible with direct observation (IPCC 2010 <sup>[[#fn:r548|548]]</sup> ). The different approaches of models and GHGIs to estimating anthropogenic emissions and removals are shown in (Figure 2.6). The difficulty is that ''indirect'' effects of environmental changes (e.g., climate change and rising atmospheric CO <sub>2</sub> ) affect both manged and unmanaged lands, and some approaches treat these as anthropogenic while others do not. Bookkeeping models (e.g., Houghton and Nassikas 2017 <sup>[[#fn:r549|549]]</sup> ) attempt to estimate the fluxes of CO <sub>2</sub> driven by direct anthropogenic effects alone. DGVMs model the ''indirect'' environmental effects of climate and CO <sub>2</sub> . If the indirect effects happen on land experiencing anthropogenic land cover change or management (harvest and regrowth), DGVMs treat this as anthropogenic. Country GHGIs separately report fluxes due to land conversion (e.g., forests to croplands) and fluxes due to land management (e.g., forest land remaining forest land). The ‘managed land proxy’ is used as a pragmatic approach to estimate anthropogenic fluxes on managed lands, whereby countries define the areas they consider managed and include all of the emissions and removals that occur on those lands. Emissions and removals are caused simultaneously by direct, indirect and natural drivers and are captured in the reporting, which often relies on inventories.

Grassi et al. (2018 <sup>[[#fn:r550|550]]</sup> ) demonstrated that estimates of CO <sub>2</sub> emissions from global models and from nationally reported GHGIs were similar for deforestation and afforestation, but different for managed forests. Countries generally reported larger areas of managed forests than the models and the carbon removals by these managed forests were also larger. The flux due to indirect effects on managed lands was quantified using post-processing of results from DGVMs, looking at the indirect effects of CO <sub>2</sub> and climate change on secondary forest areas. The derived DGVM indirect managed forest flux was found to account for most of the difference between the bookkeeping models and the inventories.

''Regional differences''

Figure 2.7 shows regional differences in emissions due to AFOLU. Recent increases in deforestation rates in some tropical countries have been partially balanced by increases in forest area in India, China, the USA and Europe (FAO-FRA 2015 <sup>[[#fn:r551|551]]</sup> ). The trend in emissions from AFOLU since the 1990s is ''uncertain'' because some data suggest a declining rate of deforestation (FAO-FRA 2015 <sup>[[#fn:r552|552]]</sup> ), while data from satellites suggest an increasing rate (Kim 2014 <sup>[[#fn:r553|553]]</sup> ; Hansen et al. 2012 <sup>[[#fn:r554|554]]</sup> ). The disagreement results in part from differences in the definition of forest and approaches to estimating deforestation. The FAO defines deforestation as the conversion of forest to another land use (FAO-FRA 2015 <sup>[[#fn:r555|555]]</sup> ), while the measurement of forest loss by satellite may include wood harvests (forests remaining forests) and natural disturbances that are not directly caused by anthropogenic activity (e.g., forest mortality from droughts and fires). Trends in anthropogenic and natural disturbances may be in opposite directions. For example, recent drought-induced fires in the Amazon have increased the emissions from wildfires at the same time that emissions from anthropogenic deforestation have declined (Aragão et al. 2018 <sup>[[#fn:r556|556]]</sup> ). Furthermore, there have been advances since AR5 in estimating the GHG effects of different types of forest management (e.g., Valade et al. 2017 <sup>[[#fn:r557|557]]</sup> ). Overall, there is ''robust evidence and high agreement'' for a net loss of forest area and tree cover in the tropics and a net gain, mainly of secondary forests and sustainably managed forests, in the temperate and boreal zones (Chapter 1).

''Processes responsible for the land sink''

Just over half of total net anthropogenic CO <sub>2</sub> emissions (AFOLU and fossil fuels) were taken up by oceanic and land sinks ( ''robust evidence, high agreement'' ) (Table 2.3). The land sink was referred to in AR5 as the ‘residual terrestrial flux’, as it was not estimated directly, but calculated by difference from the other directly estimated fluxes in the budget (Table 2.3). In the 2018 budget (Le Quéré et al. 2018 <sup>[[#fn:r558|558]]</sup> ), the land sink term was instead estimated directly by DGVMs, leaving a budget imbalance of 2.2 GtCO <sub>2</sub> yr <sup>–1</sup> (sources overestimated or sinks underestimated). The budget imbalance may result from variations in oceanic uptake or from uncertainties in fossil fuel or AFOLU emissions, as well as from land processes not included in DGVMs.

The land sink is thought to be driven largely by the indirect effects of environmental change (e.g., climate change, increased atmospheric CO <sub>2</sub> concentration, nitrogen deposition) on unmanaged and managed lands ( ''robust evidence, high agreement'' ). The land sink has generally increased since 1900 and was a net sink of 11.7 ± 3.7 GtCO <sub>2</sub> yr <sup>–1</sup> during the period 2008–2017 (Table 2.3), absorbing 29% of global anthropogenic emissions of CO <sub>2</sub> . The land sink has slowed the rise in global land-surface air temperature by 0.09 ± 0.02°C since 1982 (medium confidence) (Zeng et al. 2017 <sup>[[#fn:r559|559]]</sup> ).

The rate of CO <sub>2</sub> removal by land accelerated from –0.026 ± 0.24 GtCO <sub>2</sub> yr <sup>–1</sup> during the warming period (1982–1998) to –0.436 ± 0.260 GtCO <sub>2</sub> yr <sup>–1</sup> during the warming hiatus (1998–2012). One explanation is that respiration rates were lower during the warming hiatus (Ballantyne et al. 2017 <sup>[[#fn:r560|560]]</sup> ). However, the lower rate of growth in atmospheric CO <sub>2</sub> during the warming hiatus may have resulted, not from lower rates of respiration, but from declining emissions from AFOLU (lower rates of tropical deforestation and increased forest growth in northern mid-latitudes) (Piao et al. 2018 <sup>[[#fn:r561|561]]</sup> ). Changes in the growth rate of atmospheric CO <sub>2</sub> , by themselves, do not identify the processes responsible and the cause of the variation is uncertain.

While year-to-year variability in the indirect land sink is high in response to climate variability, DGVM fluxes are influenced far more on decadal timescales by CO <sub>2</sub> fertilisation. A DGVM intercomparison (Sitch et al. 2015 <sup>[[#fn:r562|562]]</sup> ) for 1990–2009 found that CO <sub>2</sub> fertilisation alone contributed a mean global removal of –10.54 ± 3.68 GtCO <sub>2</sub> yr <sup>–1</sup> (trend –0.444 ± 0.202 GtCO <sub>2</sub> yr <sup>–1</sup> ). Data from forest inventories around the world corroborate the modelled land sink (Pan et al. 2011). The geographic distribution of the non-AFOLU land sink is less certain. While it seems to be distributed globally, its distribution between the tropics and non-tropics is estimated to be between 1:1 (Pan et al. 2011 <sup>[[#fn:r563|563]]</sup> ) and 1:2 (Houghton et al. 2018 <sup>[[#fn:r564|564]]</sup> ).

As described in Box 2.3, rising CO <sub>2</sub> concentrations have a fertilising effect on land, while climate has mixed effects; for example, rising temperature increases respiration rates and may enhance or reduce photosynthesis depending on location and season, while longer growing seasons might allow for higher carbon uptake. However, these processes are not included in DGVMs, which may account for at least some of the land sink. For example, a decline in the global area burned by fires each year (Andela et al. 2017 <sup>[[#fn:r565|565]]</sup> ) accounts for an estimated net sink (and/or reduced emissions) of 0.5 GtCO <sub>2</sub> yr <sup>–1</sup> (limited evidence, medium agreement) (Arora and Melton 2018 <sup>[[#fn:r566|566]]</sup> ).

Boreal forests represent an exception to this decline (Kelly et al. 2013 <sup>[[#fn:r567|567]]</sup> ). The reduction in burning not only reduces emissions, but also allows more growth of recovering forests. There is also an estimated net carbon sink of about the same magnitude (0.5 GtCO <sub>2</sub> yr <sup>–1</sup> ) as a result of soil erosion from agricultural lands and redeposition in anaerobic environments where respiration is reduced (limited evidence, low agreement) (Wang et al. 2017d <sup>[[#fn:r568|568]]</sup> ). A recent study attributes an increase in land carbon to a longer-term (1860–2005) aerosol-induced cooling (Zhang et al. 2019 <sup>[[#fn:r569|569]]</sup> ). Recent evidence also suggests that DGVMs and ESMs underestimate the effects of drought on CO <sub>2</sub> emissions (Humphrey et al. 2018 <sup>[[#fn:r570|570]]</sup> ; Green et al. 2019 <sup>[[#fn:r571|571]]</sup> ; Kolus et al. 2019 <sup>[[#fn:r572|572]]</sup> ).

<div id="section-2-3-1-2-separation-of-the-total-net-land-flux-into-afolu-fluxes-and-the-land-sink-block-2"></div>

<span id="table-2.3"></span>
<!-- START TABLE -->
'''Table 2.3'''

<span id="perturbation-of-the-global-carbon-cycle-caused-by-anthropogenic-activities-gtco-2-yr1"></span>
'''Perturbation of the global carbon cycle caused by anthropogenic activities (GtCO <sub>2</sub> yr–1)'''

Source: Le Quéré et al. (2018 <sup>[[#fn:r573|573]]</sup> ).

<!-- TABLE -->
{| class="wikitable"
|-

CO <sub>2</sub> flux (GtCO <sub>2</sub> y <sup>–1</sup> ), 10-year mean

|-
|

1960–1969

1970–1979

1980–1989

1990–1999

2000–2009

2008–2017

|-

Emissions

|-

Fossil CO <sub>2</sub> emissions

11.4 ± 0.7

17.2 ± 0.7

19.8 ± 1.1

23.1 ± 1.1

28.6 ± 1.5

34. ± 1.8

|-

AFOLU net emissions

5.5 ± 2.6

4.4 ± 2.6

4.4 ± 2.6

5.1 ± 2.6

4.8 ± 2.6

5.5 ± 2.6

|-

Partitioning

|-

Growth in atmosphere

6.2 ± 0.3

10.3 ± 0.3

12.5 ± 0.07

11.4 ± 0.07

14.7 ± 0.07

17.2 ± 0.07

|-

Ocean sink

3.7 ± 1.8

4.8 ± 1.8

6.2 ± 1.8

7.3 ± 1.8

7.7 ± 1.8

8.8 ± 1.8

|-

Land sink (non-AFOLU)

4.4 ± 1.8

7.7 ± 1.5

6.6 ± 2.2

8.8 ± 1.8

9.9 ± 2.6

11.7 ± 2.6

|-

Budget imbalance

2.2

–1.1

–1.1

0.7

0.7

1.8

|-

Total net land flux (AFOLU – land sink)

+1.1 ± 3.2

–3.3 ± 3.0

–2.2 ± 3.4

–3.7 ± 2.2

–5.1 ± 3.2

–6.2 ± 3.7

|}
<!-- END TABLE -->

<div id="section-2-3-1-2-separation-of-the-total-net-land-flux-into-afolu-fluxes-and-the-land-sink-block-3"></div>

<span id="figure-2.5"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 2.5'''

<span id="global-net-co2-emissions-due-to-afolu-from-different-approaches-in-gtco2-yr1.brown-line-the-mean-and-individual-estimates-brown-shading-from-two-bookkeeping-models-houghton-and-nassikas-2017-hansis-et-al.-2015.-blue-line-the-mean-from-dgvms-run-with-the-same-driving-data-with-the-pale-blue-shading-showing-the-1-standard-deviation"></span>
<!-- IMG CAPTION -->
'''Global net CO2 emissions due to AFOLU from different approaches (in GtCO2 yr–1).Brown line: the mean and individual estimates (brown shading) from two bookkeeping models (Houghton and Nassikas 2017; Hansis et al. 2015). Blue line: the mean from DGVMs run with the same driving data with the pale blue shading showing the ±1 standard deviation […]'''
<!-- IMG FILE -->
[[File:d2d43ba6645535ca1b26c05b8836a7ed Figure-2.5-1024x431.jpg]]

Global net CO <sub>2</sub> emissions due to AFOLU from different approaches (in GtCO <sub>2</sub> yr <sup>–1</sup> ).Brown line: the mean and individual estimates (brown shading) from two bookkeeping models (Houghton and Nassikas 2017 <sup>[[#fn:r574|574]]</sup> ; Hansis et al. 2015 <sup>[[#fn:r575|575]]</sup> ). Blue line: the mean from DGVMs run with the same driving data with the pale blue shading showing the ±1 standard deviation range. Yellow line: data downloaded from FAOSTAT website (Tubiello et al. 2013 <sup>[[#fn:r576|576]]</sup> ); the dashed line is primarily forest-related emissions, while the solid yellow line also includes emissions from peat fires and peat draining. Orange line: Greenhouse Gas Inventories (GHGI) based on country reports to UNFCCC (Grassi et al. 2018 <sup>[[#fn:r577|577]]</sup> ), data are shown only from 2005 because reporting in many developing countries became more consistent/reliable after this date. For more details on methods see Box 2.2.

<!-- END IMG -->
<div id="section-2-3-1-2-separation-of-the-total-net-land-flux-into-afolu-fluxes-and-the-land-sink-block-4"></div>

<span id="figure-2.6"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 2.6'''

<span id="summary-of-the-main-conceptual-differences-between-ghg-inventories-and-global-models-in-considering-what-is-the-anthropogenic-land-co2-flux.-adapted-from-grassi-et-al.-2018-effects-of-key-processes-on-the-land-flux-as-defined-by-ipcc-2010-including-where-these-effects-occur-in-managed-andor-unmanaged-lands-and-how-these-effects-are"></span>
<!-- IMG CAPTION -->
'''Summary of the main conceptual differences between GHG Inventories and global models in considering what is the ‘anthropogenic land CO2 flux’. Adapted from Grassi et al. (2018), effects of key processes on the land flux as defined by IPCC (2010) including where these effects occur (in managed and/or unmanaged lands) and how these effects are […]'''
<!-- IMG FILE -->
[[File:5d2dee5ed5f22d7ddea96a7298575f3d Figure-2.6-1024x622.jpg]]

Summary of the main conceptual differences between GHG Inventories and global models in considering what is the ‘anthropogenic land CO <sub>2</sub> flux’. Adapted from Grassi et al. (2018) <sup>[[#fn:r578|578]]</sup> , effects of key processes on the land flux as defined by IPCC (2010) <sup>[[#fn:r579|579]]</sup> including where these effects occur (in managed and/or unmanaged lands) and how these effects are captured in (a) bookkeeping models that do not explicitly model the effects of environmental change (although some is implicitly captured in data on carbon densities and growth and decay rates), (b) DGVMs that include the effects of environmental change on all lands, and run the models with and without land use change to diagnose ‘land use change’. The ‘land sink’ is then conceptually assumed to be a natural response of land to the anthropogenic perturbation of environmental change, DGVMs include the effects of inter-annual climate variability, and some include fires but no other natural disturbances, and (c) GHG Inventories reported by countries to the UNFCCC that report all fluxes in areas the countries define as ‘managed land’ but do not report unmanaged land. This is the CO <sub>2</sub> flux due to Land Use Land Use Change and Forestry (LULUCF) which is a part of the overall AFOLU flux. The area of land considered as managed in the inventories is greater than that considered as subject to direct management activities (harvest and regrowth) in the models.

<!-- END IMG -->
<div id="section-2-3-1-2-separation-of-the-total-net-land-flux-into-afolu-fluxes-and-the-land-sink-block-5"></div>

<span id="figure-2.7"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 2.7'''

<span id="regional-trends-in-net-anthropogenic-land-atmosphere-co2-flux-from-a-range-of-different-approaches-in-gtco2-yr1.red-symbols-bookkeeping-models-hexagon-houghton-and-nassikas-2017-square-hansis-et-al.-2015.-blue-cross-the-mean-from-dgmvs-with-the-box-showing-the-1-standard-deviation-range.-green-triangles-downloaded-from-faostat-website-the-open-triangle-is"></span>
<!-- IMG CAPTION -->
'''Regional trends in net anthropogenic land-atmosphere CO2 flux from a range of different approaches (in GtCO2 yr–1).Red symbols: bookkeeping models (hexagon: Houghton and Nassikas 2017; square: Hansis et al. 2015). Blue cross: the mean from DGMVs with the box showing the 1 standard deviation range. Green triangles: downloaded from FAOSTAT website; the open triangle is […]'''
<!-- IMG FILE -->
[[File:34c8313cefcdc0cc91b0aebb3814647d Figure-2.7-970x1024.jpg]]

Regional trends in net anthropogenic land-atmosphere CO <sub>2</sub> flux from a range of different approaches (in GtCO <sub>2</sub> yr <sup>–1</sup> ).Red symbols: bookkeeping models (hexagon: Houghton and Nassikas 2017 <sup>[[#fn:r580|580]]</sup> ; square: Hansis et al. 2015 <sup>[[#fn:r581|581]]</sup> ). Blue cross: the mean from DGMVs with the box showing the 1 standard deviation range. Green triangles: downloaded from FAOSTAT website; the open triangle is primarily forest-related emissions, while the closed triangle includes emission from peat fires and peat drainage. Yellow inverted triangle: GHGI LULUCF flux based on country reports to UNFCCC (Grassi et al. 2018 <sup>[[#fn:r582|582]]</sup> ). Data for developing countries are only shown for 2006–2015 because reporting in many developing countries became more consistent/reliable after 2005. For more details on methods see Box 2.2.

<!-- END IMG -->
<div id="section-2-3-1-3-gross-emissions-and-removals-contributing-to-afolu-emissions"></div>

<span id="gross-emissions-and-removals-contributing-to-afolu-emissions"></span>
==== 2.3.1.3 Gross emissions and removals contributing to AFOLU emissions ====

<div id="section-2-3-1-3-gross-emissions-and-removals-contributing-to-afolu-emissions-block-1"></div>

The modelled AFOLU flux of 5.5 ± 3.7 GtCO <sub>2</sub> yr <sup>–1</sup> over the period 2008–2017 represents a net value. It consists of both gross emissions of CO <sub>2</sub> from deforestation, forest degradation and the oxidation of wood products, as well as gross removals of CO <sub>2</sub> in forests and soils recovering from harvests and agricultural abandonment (Figure 2.4). The uncertainty of these gross fluxes is high because few studies report gross fluxes from AFOLU. Houghton and Nassikas (2017 <sup>[[#fn:r583|583]]</sup> ) estimated gross emissions to be as high as 20.2 GtCO <sub>2</sub> yr <sup>–1</sup> ( ''limited evidence, low agreement'' ) (Figure 2.4), and even this may be an underestimate because the land-use change data used from FAOSTAT (Tubiello et al. 2013 <sup>[[#fn:r584|584]]</sup> ) is itself a net of all changes within a country.

Gross emissions and removals of CO <sub>2</sub> result from rotational uses of land, such as wood harvest and shifting cultivation, including regrowth. These gross fluxes are more informative for assessing the timing and potential for mitigation than estimates of net fluxes, because the gross fluxes include a more complete accounting of individual activities. Gross emissions from rotational land use in the tropics are approximately 37% of total CO <sub>2</sub> emissions, rather than 14%, as suggested by net AFOLU emissions (Houghton and Nassikas 2018 <sup>[[#fn:r585|585]]</sup> ). Further, if the forest is replanted or allowed to regrow, gross removals of nearly the same magnitude would be expected to continue for decades.

<div id="section-2-3-1-4-gross-emissions-and-removals-contributing-to-the-non-anthropogenic-land-sink"></div>

<span id="gross-emissions-and-removals-contributing-to-the-non-anthropogenic-land-sink"></span>
==== 2.3.1.4 Gross emissions and removals contributing to the non-anthropogenic land sink ====

<div id="section-2-3-1-4-gross-emissions-and-removals-contributing-to-the-non-anthropogenic-land-sink-block-1"></div>

The net land sink averaged 11.2 ± 2.6 GtCO <sub>2</sub> yr <sup>–1</sup> (l ''ikely range'' ) over 2007–2016 (Table 2.3.2), but its gross components have not been estimated at the global level. There are many studies that suggest increasing emissions of carbon are due to indirect environmental effects and natural disturbance, for example, temperature-induced increases in respiration rates (Bond-Lamberty et al. 2018 <sup>[[#fn:r586|586]]</sup> ), increased tree mortality (Brienen et al. 2015 <sup>[[#fn:r587|587]]</sup> ; Berdanier and Clark 2016 <sup>[[#fn:r588|588]]</sup> ; McDowell et al. 2018 <sup>[[#fn:r589|589]]</sup> ) and thawing permafrost (Schuur et al. 2015 <sup>[[#fn:r590|590]]</sup> ). The global carbon budget indicates that land and ocean sinks ''have increased'' over the last six decades in proportion to total CO <sub>2</sub> emissions (Le Quéré et al. 2018 <sup>[[#fn:r591|591]]</sup> ) ( ''robust evidence, high agreement'' ). That means that any emissions must have been balanced by even larger removals (likely driven by CO <sub>2</sub> fertilisation, climate change, nitrogen deposition, erosion and redeposition of soil carbon, a reduction in areas burned, aerosol-induced cooling and changes in natural disturbances) (Box 2.3).

Climate change is expected to impact terrestrial biogeochemical cycles via an array of complex feedback mechanisms that will act to either enhance or decrease future CO <sub>2</sub> emissions from land. Because the gross emissions and removals from environmental changes are not constrained at present, the balance of future positive and negative feedbacks remains uncertain. Estimates from climate models in Coupled Model Intercomparison Project 5 (CMIP5) exhibit large differences for different carbon and nitrogen cycle feedbacks and how they change in a warming climate (Anav et al. 2013 <sup>[[#fn:r592|592]]</sup> ; Friedlingstein et al. 2006 <sup>[[#fn:r593|593]]</sup> ; Friedlingstein, et al. 2014 <sup>[[#fn:r594|594]]</sup> ). The differences are in large part due to the uncertainty regarding how primary productivity and soil respiration will respond to environmental changes, with many of the models not even agreeing on the sign of change. Furthermore, many models do not include a nitrogen cycle, which may limit the CO <sub>2</sub> fertilisation effect in the future (Box 2.3). There is an increasing amount of observational data available and methods to constrain models (e.g., Cox et al. 2013 <sup>[[#fn:r595|595]]</sup> ; Prentice, et al., 2015 <sup>[[#fn:r596|596]]</sup> ) which can reduce uncertainty.

<div id="section-2-3-1-5-potential-impact-of-mitigation-on-atmospheric-co2-concentrations"></div>

<span id="potential-impact-of-mitigation-on-atmospheric-co-2-concentrations"></span>
==== 2.3.1.5 Potential impact of mitigation on atmospheric CO <sub>2</sub> concentrations ====

<div id="section-2-3-1-5-potential-impact-of-mitigation-on-atmospheric-co2-concentrations-block-1"></div>

If CO <sub>2</sub> concentrations decline in the future as a result of low 2 emissions and large negative emissions, the global land and<br />
ocean sinks are expected to weaken (or even reverse). The oceans<br />
are expected to release CO <sub>2</sub> back to the atmosphere when the concentration declines (Ciais et al. 2013a <sup>[[#fn:r597|597]]</sup> ; Jones et al. 2016 <sup>[[#fn:r598|598]]</sup> ).

This means that to maintain atmospheric CO <sub>2</sub> and temperature at low levels, both the excess CO <sub>2</sub> from the atmosphere and the CO <sub>2</sub> progressively outgassed from the ocean and land sinks will need to be removed. This outgassing from the land and ocean sinks is called the ‘rebound effect’ of the global carbon cycle (Ciais et al. 2013a <sup>[[#fn:r599|599]]</sup> ). It will reduce the effectiveness of negative emissions and increase the deployment level needed to achieve a climate stabilisation target (Jackson et al. 2017 <sup>[[#fn:r600|600]]</sup> ; Jones et al. 2016 <sup>[[#fn:r601|601]]</sup> ) ( ''limited evidence, high agreement'' ).

<span id="methane"></span>
=== 2.3.2 Methane ===

<div id="section-2-3-2-1-atmospheric-trends"></div>

<span id="atmospheric-trends"></span>
==== 2.3.2.1 Atmospheric trends ====

<div id="section-2-3-2-1-atmospheric-trends-block-1"></div>

In 2017, the globally averaged atmospheric concentration of CH <sub>4</sub> was 1850 ± 1 ppbv (Figure 2.8A). Systematic measurements of atmospheric CH <sub>4</sub> concentrations began in the mid-1980s and trends show a steady increase between the mid-1980s and early- 1990s, slower growth thereafter until 1999, a period of no growth between 1999 and 2006, followed by a resumption of growth in 2007. The growth rates show very high inter-annual variability with a negative trend from the beginning of the measurement period until about 2006, followed by a rapid recovery and continued high inter- annual variability through 2017 (Figure 2.8B). The growth rate has been higher over the past 4 years ( ''high confidence'' ) (Nisbet et al. 2019 <sup>[[#fn:r602|602]]</sup> ). The trend in δ <sup>13</sup> C-CH <sub>4</sub> prior to 2000 with less depleted ratios indicated that the increase in atmospheric concentrations was due to thermogenic (fossil) CH <sub>4</sub> emissions; the reversal of this trend after 2007 indicates a shift to biogenic sources (Figure 2.8C).

Understanding the underlying causes of temporal variation in atmospheric CH <sub>4</sub> concentrations is an active area of research. Several studies concluded that inter-annual variability of CH <sub>4</sub> growth was driven by variations in natural emissions from wetlands (Rice et al. 2016 <sup>[[#fn:r603|603]]</sup> ; Bousquet et al. 2006 <sup>[[#fn:r604|604]]</sup> ; Bousquet et al. 2011 <sup>[[#fn:r605|605]]</sup> ). These modelling efforts concluded that tropical wetlands were responsible for between 50 and 100% of the inter-annual fluctuations and the renewed growth in atmospheric concentrations after 2007. However, results were inconsistent for the magnitude and geographic distribution of the wetland sources between the models. Pison et al. (2013) <sup>[[#fn:r606|606]]</sup> used two atmospheric inversion models and the ORCHIDEE model and found greater uncertainty in the role of wetlands in inter-annual variability between 1990 and 2009 and during the 1999–2006 pause. Poulter et al. (2017) <sup>[[#fn:r607|607]]</sup> used several biogeochemical models and inventory-based wetland area data to show that wetland CH <sub>4</sub> emissions increases in the boreal zone have been offset by decreases in the tropics, and concluded that wetlands have not contributed significantly to renewed atmospheric CH <sub>4</sub> growth.

The models cited above assumed that atmospheric hydroxyl radical (OH) sink over the period analysed did not vary. OH reacts with CH <sub>4</sub> as the first step toward oxidation to CO <sub>2</sub> . In global CH <sub>4</sub> budgets, the atmospheric OH sink has been difficult to quantify because its short lifetime (about 1 second) and its distribution is controlled by precursor species that have non-linear interactions (Taraborrelli et al., 2012 <sup>[[#fn:r608|608]]</sup> ; Prather et al., 2017 <sup>[[#fn:r609|609]]</sup> ). Understanding of the atmospheric OH sink has evolved recently. The development of credible time series of methyl chloroform (MCF: CH3CCl3) observations offered a way to understand temporal dynamics of OH abundance and applying this to global budgets further weakened the argument for the role of wetlands in determining temporal trends since 1990. Several authors used the MCF approach and concluded that changes in the atmospheric OH sink explained a large portion of the suppression in global CH <sub>4</sub> concentrations relative to the pre-1999 trend (Turner et al. 2017 <sup>[[#fn:r610|610]]</sup> ; Rigby et al. 2013 <sup>[[#fn:r611|611]]</sup> ; McNorton et al. 2016 <sup>[[#fn:r612|612]]</sup> ). These studies could not reject the null hypothesis that OH has remained constant in recent decades and they did not suggest a mechanism for the inferred OH concentration changes (Nisbet et al. 2019 <sup>[[#fn:r613|613]]</sup> ). Nicely et al. (2018) <sup>[[#fn:r614|614]]</sup> used a mechanistic approach and demonstrated that variation in atmospheric OH was much lower than what MCF studies claimed that positive trends in OH due to the effects of water vapour, nitrogen oxides (NOx), tropospheric ozone and expansion of the tropical Hadley cells offsets the decrease in OH that is expected from increasing atmospheric CH <sub>4</sub> concentrations.

The depletion of δ <sup>13</sup> C <sub>atm</sub> beginning in 2009 could be due to changes in several sources. Decreased fire emissions combined with increased tropical wetland emissions compared to earlier years could explain the δ <sup>13</sup> C perturbations to atmospheric CH <sub>4</sub> sources (Worden et al. 2017 <sup>[[#fn:r615|615]]</sup> ; Schaefer et al. 2016 <sup>[[#fn:r616|616]]</sup> ). However, because tropical wetland emissions are higher in the southern hemisphere, and the remote sensing observations show that CH <sub>4</sub> emissions increases are largely in the north tropics (Bergamaschi et al. 2013 <sup>[[#fn:r617|617]]</sup> ; Melton et al. 2013 <sup>[[#fn:r618|618]]</sup> ; Houweling et al. 2014 <sup>[[#fn:r619|619]]</sup> ), an increased wetland source does not fit well with the southern hemisphere δ <sup>13</sup> C observations. New evidence shows that tropical wetland CH <sub>4</sub> emissions are significantly underestimated, perhaps by a factor of 2, because estimates do not account for release by tree stems (Pangala et al. 2017 <sup>[[#fn:r620|620]]</sup> ). Several authors have concluded that agriculture is a more probable source of increased emissions, particularly from rice and livestock in the tropics, which is consistent with inventory data (Wolf et al. 2017 <sup>[[#fn:r621|621]]</sup> ; Patra et al. 2016 <sup>[[#fn:r622|622]]</sup> ; Schaefer et al. 2016 <sup>[[#fn:r623|623]]</sup> ).

The importance of fugitive emissions in the global atmospheric accumulation rate is growing ( ''medium evidence, high agreement'' ). The increased production of natural gas in the US from the mid 2000s is of particular interest because it coincides with renewed atmospheric CH <sub>4</sub> growth (Rice et al. 2016 <sup>[[#fn:r624|624]]</sup> ; Hausmann et al. 2015 <sup>[[#fn:r625|625]]</sup> ). Reconciling increased fugitive emissions with increased isotopic depletion of atmospheric CH <sub>4</sub> indicates that there are ''likely'' multiple changes in emissions and sinks that affect atmospheric accumulation ( ''medium confidence'' ).

With respect to atmospheric CH <sub>4</sub> growth rates, we conclude that there is significant and ongoing accumulation of CH <sub>4</sub> in the atmosphere ( ''very high confidence'' ). The reason for the pause in growth rates and subsequent renewed growth is at least partially associated with land use and land use change. Evidence that variation in the atmospheric OH sink plays a role in the year-to-year variation of the CH <sub>4</sub> is accumulating, but results are contradictory ( ''medium evidence, low agreement'' ) and refining this evidence is constrained by lack of long-term isotopic measurements at remote sites, particularly in the tropics. Fugitive emissions likely contribute to the renewed growth after 2006 ( ''medium evidence, high agreement'' ). Additionally, the recent depletion trend of <sup>13</sup> C isotope in the atmosphere indicates that growth in biogenic sources explains part of the current growth and that biogenic sources make up a larger proportion of the source mix compared to the period before 1997 ( ''robust evidence, high agreement'' ). In agreement with the findings of AR5, we conclude that wetlands are important drivers of inter- annual variability and current growth rates ( ''medium evidence, high agreement'' ). Ruminants and the expansion of rice cultivation are also important contributors to the current growth trend ( ''medium evidence, high agreement'' ).

<div id="section-2-3-2-1-atmospheric-trends-block-2"></div>

<span id="figure-2.8"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 2.8'''

<span id="globally-averaged-atmospheric-ch4-mixing-ratios-frame-a-and-instantaneous-rates-of-change-frame-b-and-c-isotopevariation-frame-c.-data-sources-noaaesrl-www.esrl.noaa.govgmdccggtrends_ch4-dlugokencky-et-al.-1994-and-schaefer-et-al.-2016."></span>
<!-- IMG CAPTION -->
'''Globally averaged atmospheric CH4 mixing ratios (Frame A) and instantaneous rates of change (Frame B) and C isotope/variation (Frame C). Data sources: NOAA/ESRL (www.esrl.noaa.gov/gmd/ccgg/trends_ch4); Dlugokencky et al. (1994) and Schaefer et al. (2016).'''
<!-- IMG FILE -->
[[File:ba5541e1bc4f3f3fc46c2d300188606f Figure-2.8-1018x1024.jpg]]

Globally averaged atmospheric CH <sub>4</sub> mixing ratios (Frame A) and instantaneous rates of change (Frame B) and C isotope/variation (Frame C). Data sources: NOAA/ESRL (www.esrl.noaa.gov/gmd/ccgg/trends_ch4); Dlugokencky et al. (1994) <sup>[[#fn:r626|626]]</sup> and Schaefer et al. (2016) <sup>[[#fn:r627|627]]</sup> .

<!-- END IMG -->
<div id="section-2-3-2-2-land-use-effects"></div>

<span id="land-use-effects"></span>
==== 2.3.2.2 Land use effects ====

<div id="section-2-3-2-2-land-use-effects-block-1"></div>

Agricultural emissions are predominantly from enteric fermentation and rice, with manure management and waste burning contributing<br />
small amounts (Figure 2.9). Since 2000, livestock production has been responsible for 33% of total global emissions and 66% of agricultural emissions (EDGAR 4.3.2 database, May 2018; USEPA 2012 <sup>[[#fn:r628|628]]</sup> ; Tubiello et al. 2014 <sup>[[#fn:r629|629]]</sup> ; Janssens-Maenhout et al. 2017b <sup>[[#fn:r630|630]]</sup> ). Asia has the largest livestock emissions (37%) and emissions in the region have been growing by around 2% per year over the same period.

North America is responsible for 26% and emissions are stable; Europe is responsible for around 8% of emissions, and these are decreasing slightly (&lt;1% per year). Africa is responsible for 14%, but emissions are growing fastest in this region at around 2.5% y <sup>–1</sup> . In Latin America and the Caribbean, livestock emissions are decreasing at around 1.6% per year and the region makes up 16% of emissions.

Rice emissions are responsible for about 24% of agricultural emissions and 89% of these are from Asia. Rice emissions are increasing by 0.9% per year in that region. These trends are predicted to continue through 2030 (USEPA 2013 <sup>[[#fn:r631|631]]</sup> ).

Upland soils are a net sink of atmospheric CH <sub>4</sub> , but soils both produce and consume the gas. On the global scale, climatic zone, soil texture and land cover have an important effect on CH <sub>4</sub> uptake in upland soils (Tate 2015 <sup>[[#fn:r632|632]]</sup> ; Yu et al. 2017 <sup>[[#fn:r633|633]]</sup> ; Dutaur and Verchot 2007 <sup>[[#fn:r634|634]]</sup> ). Boreal soils take up less than temperate or tropical soils, coarse textured soils take up more CH <sub>4</sub> than medium and fine textured soils, and forests take up more than other ecosystems. Low levels of nitrogen fertilisation or atmospheric deposition can affect the soil microbial community and stimulate soil CH <sub>4</sub> uptake in nitrogen-limited soils, while higher fertilisation rates decrease uptake (Edwards et al. 2005 <sup>[[#fn:r635|635]]</sup> ; Zhuang et al., 2013 <sup>[[#fn:r636|636]]</sup> ). Globally, nitrogen fertilisation on agricultural lands may have suppressed CH <sub>4</sub> oxidation by as much as 26 Tg between 1998 and 2004 ( ''low confidence, low agreement'' ) (Zhuang et al., 2013 <sup>[[#fn:r637|637]]</sup> ). The effect of nitrogen additions is cumulative and repeated fertilisation events have progressively greater suppression effects ( ''robust evidence, high agreement'' ) (Tate 2015 <sup>[[#fn:r638|638]]</sup> ). Other factors such as higher temperatures, increased atmospheric concentrations and changes in rainfall patterns stimulate soil CH <sub>4</sub> consumption in unfertilised ecosystems. Several studies (Yu et al. 2017 <sup>[[#fn:r639|639]]</sup> ; Xu et al. 2016 <sup>[[#fn:r640|640]]</sup> ; Curry 2009 <sup>[[#fn:r641|641]]</sup> ) have shown that globally, uptake has been increasing during the second half of the 20th century and is expected to continue to increase by as much as 1 Tg in the 21st century, particularly in forests and grasslands ( ''medium evidence, high agreement'' ).

Northern peatlands (40–70°N) are a significant source of atmospheric CH <sub>4</sub> , emitting about 48 TgCH <sub>4</sub> , or about 10% of the total emissions to the atmosphere (Zhuang et al. 2006 <sup>[[#fn:r642|642]]</sup> ; Wuebbles and Hayhoe 2002 <sup>[[#fn:r643|643]]</sup> ). CH <sub>4</sub> emissions from natural northern peatlands are highly variable, with the highest rate from fens ( ''medium evidence, high agreement'' ). Peatland management and restoration alters the exchange of CH <sub>4</sub> with the atmosphere ( ''medium evidence, high agreement'' ). Management of peat soils typically converts them from CH <sub>4</sub> sources to sinks (Augustin et al. 2011 <sup>[[#fn:r644|644]]</sup> ; Strack and Waddington 2008 <sup>[[#fn:r645|645]]</sup> ; Abdalla et al. 2016 <sup>[[#fn:r646|646]]</sup> ) ( ''robust evidence, high agreement'' ). While restoration decreases CO <sub>2</sub> emissions (Section 4.9.4), CH <sub>4</sub> emissions often increase relative to the drained conditions ( ''robust evidence, high agreement'' ) (Osterloh et al. 2018 <sup>[[#fn:r647|647]]</sup> ; Christen et al. 2016 <sup>[[#fn:r648|648]]</sup> ; Koskinen et al. 2016 <sup>[[#fn:r649|649]]</sup> ; Tuittila et al. 2000 <sup>[[#fn:r650|650]]</sup> ; Vanselow-Algan et al. 2015 <sup>[[#fn:r651|651]]</sup> ; Abdalla et al. 2016 <sup>[[#fn:r652|652]]</sup> ). Drained peatlands are usually considered to be negligible methane sources, but they emit CH <sub>4</sub> under wet weather conditions and from drainage ditches (Drösler et al. 2013 <sup>[[#fn:r653|653]]</sup> ; Sirin et al. 2012 <sup>[[#fn:r654|654]]</sup> ). While ditches cover only a small percentage of the drained area, emissions can be sufficiently high that drained peatlands emit comparable CH <sub>4</sub> as undrained ones ( ''medium evidence, medium agreement'' ) (Sirin et al. 2012 <sup>[[#fn:r655|655]]</sup> ; Wilson et al. 2016 <sup>[[#fn:r656|656]]</sup> ).

Because of the large uncertainty in the tropical peatland area, estimates of the global flux are highly uncertain. A meta-analysis of the effect of conversion of primary forest to rice production showed that emissions increased by a factor of four ( ''limited evidence, high agreement'' ) (Hergoualc’h and Verchot, 2012 <sup>[[#fn:r657|657]]</sup> ). For land uses that required drainage, emissions decreased by a factor of three ( ''limited evidence, high agreement'' ).There are no representative measurements of emissions from drainage ditches in tropical peatlands.

<div id="section-2-3-2-2-land-use-effects-block-2"></div>

<span id="figure-2.9"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 2.9'''

<span id="average-agricultural-ch4-emissions-estimates-from-1990.-sub-sectorial-agricultural-emissions-are-based-on-the-emissions-database-for-global-atmospheric-research-edgar-v4.3.2-janssens-maenhout-et-al.-2017a-faostat-tubiello-et-al.-2013-and-national-ghgi-data-grassi-et-al.-2018.-ghgi-data-are-aggregate-values-for-the-sector.-note-that-edgar-data-are-complete-only-through"></span>
<!-- IMG CAPTION -->
'''Average agricultural CH4 emissions estimates from 1990. Sub-sectorial agricultural emissions are based on the Emissions Database for Global Atmospheric Research (EDGAR v4.3.2; Janssens-Maenhout et al. 2017a); FAOSTAT (Tubiello et al. 2013); and National GHGI data (Grassi et al. 2018). GHGI data are aggregate values for the sector. Note that EDGAR data are complete only through […]'''
<!-- IMG FILE -->
[[File:36b982b2b8d180f89286cbcde6c789c0 Figure-2.9-1024x599.jpg]]

Average agricultural CH <sub>4</sub> emissions estimates from 1990. Sub-sectorial agricultural emissions are based on the Emissions Database for Global Atmospheric Research (EDGAR v4.3.2; Janssens-Maenhout et al. 2017a <sup>[[#fn:r658|658]]</sup> ); FAOSTAT (Tubiello et al. 2013 <sup>[[#fn:r659|659]]</sup> ); and National GHGI data (Grassi et al. 2018 <sup>[[#fn:r660|660]]</sup> ). GHGI data are aggregate values for the sector. Note that EDGAR data are complete only through 2012; the data in the right-hand panel represent the three years 2010–2012 and are presented for comparison.

<!-- END IMG -->
<span id="nitrous-oxide"></span>
=== 2.3.3 Nitrous oxide ===

<div id="section-2-3-3-1-atmospheric-trends"></div>

<span id="atmospheric-trends-1"></span>
==== 2.3.3.1 Atmospheric trends ====

<div id="section-2-3-3-1-atmospheric-trends-block-1"></div>

The atmospheric abundance of N <sub>2</sub> O has increased since 1750, from a pre-industrial concentration of 270 ppbv to 330 ppbv in 2017 (high agreement, robust evidence) (US National Oceanographic and Atmospheric Agency, Earth Systems Research Laboratory)

(Figure 2.10). The rate of increase has also increased, from approximately 0.15 ppbv yr <sup>–1</sup> 100 years ago, to 0.85 ppbv yr <sup>–1</sup> over the period 2001–2015 (Wells et al. 2018 <sup>[[#fn:r661|661]]</sup> ). Atmospheric N <sub>2</sub> O isotopic composition (14/15N) was relatively constant during the pre-industrial period (Prokopiou et al. 2018 <sup>[[#fn:r662|662]]</sup> ) and shows a decrease in the δ15N as the N <sub>2</sub> O mixing ratio in the atmosphere has increased between 1940 and 2005. This recent decrease indicates that terrestrial sources are the primary driver of increasing trends and marine sources contribute around 25% (Snider et al. 2015 <sup>[[#fn:r663|663]]</sup> ). Microbial denitrification and nitrification processes are responsible for more than 80% of total global N <sub>2</sub> O emissions, which includes natural soils, agriculture and oceans, with the remainder coming from non-biological sources such as biomass burning and fossil-fuel combustion (Fowler et al. 2015 <sup>[[#fn:r664|664]]</sup> ). The isotopic trend also indicates a shift from denitrification to nitrification as the primary source of N <sub>2</sub> O as a result of the use of synthetic nitrogen fertiliser (high evidence, high agreement) (Park et al. 2012 <sup>[[#fn:r665|665]]</sup> ; Toyoda et al. 2013 <sup>[[#fn:r666|666]]</sup> ; Snider et al. 2015 <sup>[[#fn:r667|667]]</sup> ; Prokopiou et al. 2018 <sup>[[#fn:r668|668]]</sup> ).

The three independent sources of N <sub>2</sub> O emissions estimates from agriculture at global, regional and national levels are: USEPA, EDGAR and FAOSTAT (USEPA 2013 <sup>[[#fn:r669|669]]</sup> ; Tubiello et al. 2015 <sup>[[#fn:r670|670]]</sup> ; Janssens-Maenhout et al. 2017a <sup>[[#fn:r671|671]]</sup> ). EDGAR and FAOSTAT have temporal resolution beyond 2005 and these databases compare well with national inventory data (Figure 2.10). USEPA has historical estimates through 2005 and projections thereafter. The independent data use IPCC methods, with Tier 1 emission factors and national reporting of activity data. Tier 2 approaches are also available based on top-down and bottom-up approaches. Recent estimates using inversion modelling and process models estimate total annual global N <sub>2</sub> O emissions of 16.1–18.7 (bottom-up) and 15.9–17.7 TgN (top-down), demonstrating relatively close agreement (Thompson et al. 2014 <sup>[[#fn:r672|672]]</sup> ). Agriculture is the largest source and has increased with extensification and intensification. Recent modelling estimates of terrestrial sources show a higher emissions range that is slightly more constrained than what was reported in AR5: approximately 9 (7–11) TgN <sub>2</sub> O-N yr <sup>–1</sup> (Saikawa et al. 2014 <sup>[[#fn:r673|673]]</sup> ; Tian et al. 2016 <sup>[[#fn:r674|674]]</sup> ) compared to 6.6 (3.3–9.0) TgN <sub>2</sub> O-N yr <sup>–1</sup> (Ciais et al. 2013a <sup>[[#fn:r675|675]]</sup> ). Estimates of marine N <sub>2</sub> O emissions are between 2.5 and 4.6 TgN <sub>2</sub> O-N yr <sup>–1</sup> (Buitenhuis et al., 2018 <sup>[[#fn:r676|676]]</sup> ; Saikawa et al., 2014 <sup>[[#fn:r677|677]]</sup> ).

To conclude, N <sub>2</sub> O is continuing to accumulate in the atmosphere at an increasingly higher rate (very high confidence), driven primarily by increases in manure production and synthetic nitrogen fertiliser use from the mid-20th century onwards (high confidence). Findings since AR5 have constrained regional and global estimates of annual N <sub>2</sub> O emissions and improved our understanding of the spatio-temporal dynamics of N <sub>2</sub> O emissions, including soil rewetting and freeze-thaw cycles which are important determinants of total annual emission fluxes in some regions (medium confidence).

<div id="section-2-3-3-1-atmospheric-trends-block-2"></div>

<span id="figure-2.10"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 2.10'''

<span id="globally-averaged-atmospheric-n2o-mixing-ratios-since-1984.-data-source-noaaesrl-global-monitoring-division-www.esrl.noaa.govgmdhatscombinedn2o.html."></span>
<!-- IMG CAPTION -->
'''Globally averaged atmospheric N2O mixing ratios since 1984. Data source: NOAA/ESRL Global Monitoring Division (www.esrl.noaa.gov/gmd/hats/combined/N2O.html).'''
<!-- IMG FILE -->
[[File:7682712740be97b970691aad4a5a9f5f Figure-2.10-1024x599.jpg]]

Globally averaged atmospheric N <sub>2</sub> O mixing ratios since 1984. Data source: NOAA/ESRL Global Monitoring Division ( [[IPCC:Srccl:Chapter:Chapter-2:Www.esrl.noaa.gov:Gmd:Hats:Combined:N2o.html|www.esrl.noaa.gov/gmd/hats/combined/N <sub>2</sub> O.html]] ).

<!-- END IMG -->
<div id="section-2-3-3-1-atmospheric-trends-block-3"></div>

<span id="figure-2.11"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 2.11'''

<span id="average-agricultural-n2o-emissions-estimates-from-1990.-sub-sectorial-agricultural-emissions-are-based-on-the-emissions-database-for-global-atmospheric-research-edgar-v4.3.2-janssens-maenhout-et-al.-2017a-faostat-tubiello-et-al.-2013-and-national-ghgi-data-grassi-et-al.-2018.-ghgi-data-are-aggregate-values-for-the-sector.-note-that-edgar-data-are-complete-only-through"></span>
<!-- IMG CAPTION -->
'''Average agricultural N2O emissions estimates from 1990. Sub-sectorial agricultural emissions are based on the Emissions Database for Global Atmospheric Research (EDGAR v4.3.2; Janssens-Maenhout et al. 2017a); FAOSTAT (Tubiello et al. 2013) and National GHGI data (Grassi et al. 2018). GHGI data are aggregate values for the sector. Note that EDGAR data are complete only through […]'''
<!-- IMG FILE -->
[[File:42cde1d261437e74b8cd743bfb1e2db2 Figure-2.11-1024x577.jpg]]

Average agricultural N <sub>2</sub> O emissions estimates from 1990. Sub-sectorial agricultural emissions are based on the Emissions Database for Global Atmospheric Research (EDGAR v4.3.2 <sup>[[#fn:r2128|2128]]</sup> ; Janssens-Maenhout et al. 2017a <sup>[[#fn:r2129|2129]]</sup> ); FAOSTAT (Tubiello et al. 2013 <sup>[[#fn:r2130|2130]]</sup> ) and National GHGI data (Grassi et al. 2018 <sup>[[#fn:r2131|2131]]</sup> ). GHGI data are aggregate values for the sector. Note that EDGAR data are complete only through 2012; the EDGAR data in the right-hand panel represent the three years 2010–2012 and are presented for comparison.

<!-- END IMG -->
<div id="section-2-3-3-2-land-use-effects"></div>

<span id="land-use-effects-1"></span>
==== 2.3.3.2 Land use effects ====

<div id="section-2-3-3-2-land-use-effects-block-1"></div>

Agriculture is responsible for approximately two-thirds of N <sub>2</sub> O emissions ( ''robust evidence, high agreement'' ) (Janssens-Maenhout et al. 2017b <sup>[[#fn:r678|678]]</sup> ). Total emissions from this sector are the sum of direct and indirect emissions. Direct emissions from soils are the result of mineral fertiliser and manure application, manure management, deposition of crop residues, cultivation of organic soils and inorganic nitrogen inputs through biological nitrogen fixation. Indirect emissions come from increased warming, enrichment of downstream water bodies from runoff, and downwind nitrogen deposition on soils. The main driver of N <sub>2</sub> O emissions in croplands is a lack of synchronisation between crop nitrogen demand and soil nitrogen supply, with approximately 50% of nitrogen applied to agricultural land not taken up by the crop (Zhang et al. 2017 <sup>[[#fn:r679|679]]</sup> ). Cropland soils emit over 3 TgN <sub>2</sub> O-N yr <sup>–1</sup> ( ''medium evidence, high agreement'' ) (Janssens-Maenhout et al. 2017b <sup>[[#fn:r680|680]]</sup> ; Saikawa et al. 2014 <sup>[[#fn:r681|681]]</sup> ). Regional inverse modelling studies show larger tropical emissions than the inventory approaches and they show increases in N <sub>2</sub> O emissions from the agricultural sector in South Asia, Central America, and South America (Saikawa et al. 2014 <sup>[[#fn:r682|682]]</sup> ; Wells et al. 2018 <sup>[[#fn:r683|683]]</sup> ).

Emissions of N <sub>2</sub> O from pasturelands and rangelands have increased by as much as 80% since 1960 due to increased manure production and deposition ( ''robust evidence, high agreement'' ) (de Klein et al. 2014 <sup>[[#fn:r684|684]]</sup> ; Tian et al. 2018 <sup>[[#fn:r685|685]]</sup> ; Chadwick et al. 2018 <sup>[[#fn:r686|686]]</sup> ; Dangal et al. 2019 <sup>[[#fn:r687|687]]</sup> ; Cardenas et al. 2019 <sup>[[#fn:r689|689]]</sup> ). Studies consistently report that pasturelands and rangelands are responsible for around half of the total agricultural N <sub>2</sub> O emissions (Davidson 2009 <sup>[[#fn:r690|690]]</sup> ; Oenema et al. 2014 <sup>[[#fn:r691|691]]</sup> ; Dangal et al. 2019 <sup>[[#fn:r692|692]]</sup> ). An analysis by Dangal et al. (2019) shows that, while managed pastures make up around one-quarter of the global grazing lands, they contribute 86% of the net global N <sub>2</sub> O emissions from grasslands and that more than half of these emissions are related to direct deposition of livestock excreta on soils.

Many studies calculate N <sub>2</sub> O emissions from a linear relationship between nitrogen application rates and N <sub>2</sub> O emissions. New studies are increasingly finding nonlinear relationships, which means that N <sub>2</sub> O emissions per hectare are lower than the Tier 1 EFs (IPCC 2003 <sup>[[#fn:r693|693]]</sup> ) at low nitrogen application rates, and higher at high nitrogen application rates ( ''robust evidence, high agreement'' ) (Shcherbak et al. 2014 <sup>[[#fn:r694|694]]</sup> ; van Lent et al. 2015 <sup>[[#fn:r695|695]]</sup> ; Satria 2017 <sup>[[#fn:r696|696]]</sup> ). This not only has implications for how agricultural N <sub>2</sub> O emissions are estimated in national and regional inventories, which now often use a linear relationship between nitrogen applied and N <sub>2</sub> O emissions, it also means that in regions of the world where low nitrogen application rates dominate, increases in nitrogen fertiliser use would generate relatively small increases in agricultural N <sub>2</sub> O emissions. Decreases in application rates in regions where application rates are high and exceed crop demand for parts of the growing season are likely to have very large effects on emissions reductions ( ''medium evidence, high agreement'' ).

Deforestation and other forms of land-use change alter soil N <sub>2</sub> O emissions. Typically, N <sub>2</sub> O emissions increase following conversion of native forests and grasslands to pastures or croplands (McDaniel et al. 2019 <sup>[[#fn:r697|697]]</sup> ; van Lent et al. 2015 <sup>[[#fn:r698|698]]</sup> ). This increase lasts from a few years to a decade or more, but there is a trend toward decreased N <sub>2</sub> O emissions with time following land use change and, ultimately, lower N <sub>2</sub> O emissions than had been occurring under native vegetation, in the absence of fertilisation ( ''medium evidence, high agreement'' ) (Meurer et al. 2016 <sup>[[#fn:r2132|2132]]</sup> ; van Lent et al. 2015 <sup>[[#fn:r699|699]]</sup> ) (Figure 2.12). Conversion of native vegetation to fertilised systems typically leads to increased N <sub>2</sub> O emissions over time, with the rate of emission often being a function of nitrogen fertilisation rates, however, this response can be moderated by soil characteristics and water availability ( ''medium evidence, high agreement'' ) (van Lent et al. 2015 <sup>[[#fn:r700|700]]</sup> ; Meurer et al. 2016 <sup>[[#fn:r701|701]]</sup> ). Restoration of agroecosystems to natural vegetation, over the period of one to two decades does not lead to recovery of N <sub>2</sub> O emissions to the levels of the original vegetation (McDaniel et al. 2019 <sup>[[#fn:r702|702]]</sup> ). To conclude, findings since AR5 increasingly highlight the limits of linear N <sub>2</sub> O emission factors, particularly from field to regional scales, with emissions rising nonlinearly at high nitrogen application rates ( ''high confidence'' ). Emissions from unfertilised systems often increase and then decline over time with typically lower emissions than was the case under native vegetation ( ''high confidence'' ).

While soil emissions are the predominant source of N <sub>2</sub> O in agriculture, other sources are important (or their importance is only just emerging). Biomass burning is responsible for approximately 0.7 TgN <sub>2</sub> O-N yr <sup>–1</sup> (0.5–1.7 TgN <sub>2</sub> O-N yr <sup>–1</sup> ) or 11% of total gross anthropogenic emissions due to the release of N <sub>2</sub> O from the oxidation of organic nitrogen in biomass (UNEP 2013 <sup>[[#fn:r703|703]]</sup> ). This source includes crop residue burning, forest fires, household cook stoves and prescribed savannah, pasture and cropland burning. Aquaculture is currently not accounted for in most assessments or compilations. While it is currently responsible for less than 0.1 TgN <sub>2</sub> O-N yr <sup>–1</sup> , it is one of the fastest growing sources of anthropogenic N <sub>2</sub> O emissions (Williams and Crutzen 2010 <sup>[[#fn:r704|704]]</sup> ; Bouwman et al. 2013 <sup>[[#fn:r705|705]]</sup> ) ( ''limited evidence, high agreement'' ). Finally, increased nitrogen deposition from terrestrial sources is leading to greater indirect N <sub>2</sub> O emissions, particularly since 1980 ( ''moderate evidence, high agreement'' ) (Tian et al. 2018 <sup>[[#fn:r706|706]]</sup> , 2016 <sup>[[#fn:r2133|2133]]</sup> ). In marine systems, deposition is estimated to have increased the oceanic N <sub>2</sub> O source by 0.2 TgN <sub>2</sub> O-N yr <sup>–1</sup> or 3% of total gross anthropogenic emissions (Suntharalingam et al. 2012 <sup>[[#fn:r707|707]]</sup> ).

<div id="section-2-3-3-2-land-use-effects-block-2"></div>

<span id="figure-2.12"></span>
<!-- START IMG -->
<!-- IMG TITLE -->
'''Figure 2.12'''

<span id="effect-of-time-since-conversion-on-n2o-fluxes-in-unfertilised-orange-circles-and-fertilised-blue-circles-tropical-croplands-left-frame-and-in-unfertilised-tropical-pastures-right-frame.-average-n2o-flux-and-95-confidence-intervals-are-given-for-upland-forests-orange-inverted-triangle-and-low-canopy-forests-blue-inverted-triangle-for-comparison.-the-solid-lines-represent"></span>
<!-- IMG CAPTION -->
'''Effect of time since conversion on N2O fluxes in unfertilised (orange circles) and fertilised (blue circles) tropical croplands (left frame) and in unfertilised tropical pastures (right frame). Average N2O flux and 95% confidence intervals are given for upland forests (orange inverted triangle) and low canopy forests (blue inverted triangle), for comparison. The solid lines represent […]'''
<!-- IMG FILE -->
[[File:3fe437700a08981212c063d3ec0e286f Figure-2.12-1024x440.jpg]]

Effect of time since conversion on N <sub>2</sub> O fluxes in unfertilised (orange circles) and fertilised (blue circles) tropical croplands (left frame) and in unfertilised tropical pastures (right frame). Average N <sub>2</sub> O flux and 95% confidence intervals are given for upland forests (orange inverted triangle) and low canopy forests (blue inverted triangle), for comparison. The solid lines represent the trends for unfertilised and fertilised cases. Data source: van Lent et al. (2015) <sup>[[#fn:r708|708]]</sup> .

<!-- END IMG -->
<div id="section-2-3-3-2-land-use-effects-block-3" class="box"></div>

<span id="b2.2-methodologies-for-estimating-national-to-global-scale-anthropogenic-land-carbon-fluxes"></span>