European anthropogenic AFOLU emissions and their uncertainties : a review and benchmark data

25 Emission of greenhouse gases (GHG) and removals from land, including both anthropogenic and natural fluxes, require reliable quantification, along with estimates of their inherent uncertainties, in order to support credible mitigation action under the Paris Agreement. This study provides a state-of-the-art scientific overview of bottom-up anthropogenic emissions data from agriculture, forestry and other land use (AFOLU) in Europe. The data integrates recent AFOLU emission inventories with ecosystem data and land carbon models, covering the European Union 30 (EU28) and summarizes GHG emissions and removals over the period 1990-2016, of relevance for UNFCCC. This compilation of bottom-up estimates of the AFOLU GHG emissions of European national greenhouse gas inventories (NGHGI) with those of land carbon models and observation-based estimates of large-scale GHG fluxes, aims at improving the overall estimates of the GHG balance in Europe with respect to land GHG emissions and removals. Particular effort is devoted to the estimation of uncertainty, its propagation and role in the comparison of different 35 estimates. While NGHGI data for EU28 provides consistent quantification of uncertainty following the established IPCC guidelines, uncertainty in the estimates produced with other methods will need to account for both within model uncertainty and the spread from different model results. At EU28 level, the largest inconsistencies between estimates are mainly due to different sources of data related to human activity which result in emissions or removals taking place during a given period of time (IPCC 2006) referred here as activity data (AD) and methodologies (Tiers) used 40 for calculating emissions/removals from AFOLU sectors. The referenced datasets related to figures are visualised at http://doi.org/doi:10.5281/zenodo.3460311, Petrescu et al., 2019. https://doi.org/10.5194/essd-2019-199 O pe n A cc es s Earth System Science Data D icu ssio n s Preprint. Discussion started: 4 November 2019 c © Author(s) 2019. CC BY 4.0 License.


50
The atmospheric concentrations of the main greenhouse gases (GHG) have increased significantly since preindustrial times (pre-1750), by 46 % for carbon dioxide (CO2), 257 % for methane (CH4) and 122 % for nitrous oxide (N2O) (WMO 2019). The rise of CO2 levels is caused primarily by fossil fuel combustion, with substantial contributions from land use change. Increases in emissions of CH4 are mainly driven by agriculture and by fossil fuel 55 extraction activities, while increases in natural emissions post-2006 cannot be ruled out (e.g. Worden et al., 2017).
Increases in N2O emissions are largely due to anthropogenic activities, mainly in relation to the application of nitrogen (N) fertilizers in agriculture (FAO 2015;IPCC SRCCL 2019). Globally, fossil fuel emissions grew at a rate of 1.5 % yr −1 for the decade 2008-2017 and account for 87 % of the anthropogenic sources in the total carbon budget (Le Quéré et al., 2018). In contrast, global emissions from land-use change were estimated from bookkeeping models and land 60 carbon models (DGVMs) to be approximately stable in the same period, albeit with large uncertainties (Le Quéré et al., 2018). Emissions from land management changes were not estimated in the global budget from Le Quéré et al.

2018.
National greenhouse gas inventories (NGHGI) are prepared and reported by countries based on IPCC Guidelines using national data and different calculation methods (Tiers) for well-defined sectors. The IPCC tiers Emissions from FOLU (i.e. LULUCF in this study) represented in 2016 a sink of about 300 Mt CO2, and this sink has increased 15 % from 1990 to 2016. Bioenergy emissions are reported as a memo in the energy sector, as the emissions are captured in the forestry and land-use sector. The memo implies that bioenergy is carbon neutral in the energy sector but to compensate, the emissions are captured as a harvest (stock change) in the LULUCF sector.

115
For N2O, the largest EU28 sources are agriculture and the industrial processes and product use (IPPU) sectors, while the AFOLU sub-sectors that cover carbon stocks in agriculture and forests is a small source. Agriculture contributes emissions largely from the use of fertilizers in agricultural soils, while industrial production of nitric and adipic acid dominates IPPU-related emissions. These sources accounted for 85 % of N2O emissions in 2016, that is 5 % of total EU28 GHG emissions (Mt CO2e) in 2016. From 1990-2016, the total N2O emissions decreased by 35 % 120 (251 Mt CO2e). The top five EU28 emitters of N2O are France (18 %), Germany (16 %), UK (9 %), Poland (8 %) and Italy (8 %) that account for 59 % of the total N2O EU28 emissions (excl. LULUCF sector).
Zooming on recent trends, non-CO2 emissions show a very small decrease (-0.4 %) from 2004 to 2014 and an increase (+0.8 %) from 2015 to 2017 (Olivier and Peters, 2018). This recent growth is principally determined by the increase of N2O which have offset the declining CH4 emissions. The continued CH4 emissions decrease is mainly 125 due to shifts in the fossil fuel production from coal to natural gas in Germany, Italy and the Netherlands (BP, 2018).
The main objective of the present study is to present a synthesis of AFOLU GHG emission estimates from bottom-up approaches that can serve as a benchmark for future assessments, important during the reconciliation process with top-down GHG emissions. We use existing officially reported data from NGHGI submitted under the UNFCCC as well as other emission estimates based on research data, from global emissions datasets to detailed 130 biogeochemical models. A synthesis of available top-down non-CO2 estimates has already been undertaken by Bergamaschi et al. (2015) and will not be discussed further here. The bottom-up approaches considered, although based on independent efforts form those in the NGHGI, have some level of redundancy among them and the inventories, since they use similar activity data (AD) and largely apply the current IPCC (2006) methodology, albeit using different 'Tiers'.

135
Independent bottom-up estimates are valuable to compare with estimates officially reported to the UNFCCC and may identify differences that need closer investigation. The uncertainties presented in this paper are taken from the UNFCCC NGHGI 2018 submissions. For the global emissions dataset EDGAR uncertainties are only calculated for the year 2012 as described in the Appendix B. We evaluate the reason for differences in emissions by carefully comparing the estimates, quantifying uncertainties and detecting discrepancies. We compare the inconsistencies 140 (defined by differences between estimates) to the uncertainties (error associated to each estimate) and identify those https://doi. org/10.5194/essd-2019-199

Compilation of AFOLU emission estimates
We collected available data of AFOLU emissions and removals (  Table A2. Whenever necessary we provide details on individual countries separating CO2, CH4 and N2O. The units are based on the metric tonne (t) [1kt = 10 9 g; 1Mt = 10 12 g] for individual gases and [Mt =10 12 g; 1Tg=10 12 g] for CO2 and carbon (C) from AFOLU sectors. We rely on observational data-streams to quantify GHG fluxes from bottom-up models together with country specific inventory from NGHGI 155 official statistics (UNFCCC), global inventory datasets (EDGAR), global statistics (FAOSTAT) and global land GHG biogeochemical models used for research assessments (e.g. DGVMs TRENDY-GCP, bookkeeping models).

2015)
EFISCEN  FAOSTAT  TRENDY.v6  Bookkeeping model H&N  Bookkeeping model BLUE  References 2006  As an overview of potential uncertainty sources, Appendix Tables A1a and A1b present the use of emission factor data (EF), activity data (AD) and uncertainty estimation methods used for all agriculture and forestry data sources used in this study. The referenced data used for figures are available for download at http://doi.org/doi:10.5281/zenodo.3460311 (Petrescu et al., 2019).

Emission estimates
As part of the AFOLU sectors, agricultural activities play a significant role in non-CO2 GHG emissions (IPCC SRCCL 2019;FAO 2015). The two major gases emitted by the agricultural sector are CH4 and N2O. According

175
Regarding the forestry sub-sector of AFOLU, LULUCF, the major GHG gas is CO2. According to NGHGI  As a consequence of the similar trends and distribution of emissions to sectors presented in Table 2, we notice 220 a small but consistent variability of total emissions between the five data sources (Fig. 2).

230
One possible cause for the similarity lays in the fact that almost all sources use EFs from the same IPCC guidelines (2006). In EU28, AD are produced by four main sources and further dissiminated to the end users (see Fig.   4) and this can be subject to a certain amount of commonalities. Therefore, excluding AD and EFs, we might conclude that differences shown in Figure 2 are mainly due to the choice of the Tier method for calculating emissions (e.g. in CAPRI as shown in Appendix A, Table A1a).

235
To better understand the differences between emissions in EU28 we plotted in Figure 3 the CH4 emission percent difference between 2005 and 1990, and between last reported year, 2010 and 2005. We obeserve that for the  change there is a major reduction in CH4 emissions for all data sources due to the implementation in the 1990s of European and country specific emission reduction policies on agriculture and environment, and socioeconomic changes in the sector resulting overall into lower agricultural livestock, lower emissions from managed 240 waste disposal on land and from agricultural soils. For the other two periods considered, the relative agricultural CH4 reduction is smaller but still consistent between all data sources.

245
We could therefore conclude that all inventory-based data sources are consistent with each other for capturing recent CH4 emissions reductions, or that they are not independent because they use similar methodology with different versions of the same AD ( Fig. 4) which is mostly the case for the EU28 countries. The AD follows also a different course than the emissions data (see Fig. 4). The AD used is highly uncertain due to the collection process from surveys 250 and different national reporting systems. FAOSTAT statistics use a relative value of 20 % uncertainty that is within the range for the confidence interval that IPCC (2006) suggests.  [1989][1990][1991] and the consequent structural changes. The worst match between data sources in EU28 is found for Malta, Cyprus and Croatia but their emissions represent in the UNFCCC reporting less than 1 % of the total EU28 agricultural CH4 emissions. UNFCCC uncertainties for CH4 emissions are between 10-50 % but can be larger for some countries and sectors, e.g. Romania reporting a 500 % uncertainty for emissions from rice cultivation.

265
To exemplify the shares of CH4 emission from agriculture, in Figure 5 we present the total sub-sectoral CH4 emissions for three example countries.

(GAINS) and 2013 (CAPRI).
The highest share is attributed to enteric fermentation which for almost all countries count as ~80 % of total agricultural CH4 emissions. We notice that a very good consistency between emission estimates are found in Figure   5a for France while, on contrary a worse consistency is presented in Figure 5b for Cyprus, which might not report AD

295
to FAOSTAT from its entire territory. Figure 5c exemplifies the high 1990 CH4 emissions for Hungary in the former Eastern European Block and the lower subsequent estimates, mainly caused by political and economic changes after the dissolution of the Soviet Union (1989)(1990)(1991) Table   3 we present the allocation of emissions by subsector following the IPCC classification and we notice that each data 305 source has its own particular way of grouping emissions.      The two most important sources for N2O emissions from agriculture pertain to direct (synthetic fertilizer, 335 manure application to soils, histosols, crop residues and biological nitrogen fixation) and indirect (ammonia vollatilization, leaching and atmopsheric deposition) emissions. We exemplify this in Figure 8 where we present the N2O split in sub-activities.
We agricultural soils, count as much as 200 % to 300 %. We notice that a very good match between emission estimates is found in Figure 8a for Germany while on contrary a worse match is presented in Figure 8b for Estonia with no FAO data available in 1990 (only for former USSR). Figure 8c exemplifies the high 1990 N2O emissions for Romania (former Eastern European Block) and is due to irregularities in reporting during the dissolution of the Soviet Union 380 (1989)(1990)(1991).

Natural CH4 emissions
In recent assessments of the global CH4 budget ( Wetlands are sinks for CO2 and sources of CH4. Their net GHG emissions therefore depends on the relative sign and magnitude of the land-atmosphere exchange of these two major GHGs. Undisturbed wetlands are estimated 395 have a great carbon sequestration potential because near water-logged conditions reduce or inhibit microbial respiration, but CH4 production may partially or completely counteract carbon uptake (Petrescu et al., 2015).

405
Since CH4 emissions are highly variable in time and space as a function of climate and disturbances, it makes EF-based methods impractical and national budget estimates difficult, making it challenging to accurately estimate CH4 emissions in NGHGI. There is also a risk of double counting with emissions from inland waters as discussed e.g.
For the same period, GHGI 2019 reports an average of 10.34 kton CH4, a highly underestimated value compared to the modelled results, due to non-reporting and accounting under NGHGI.

425
Given this current gap between modelled and GHGI reported data on CH4 emission from wetlands in EU28, we stress the need of investing in better modelling methodologies for emission calculation and verification. Out of all EU28 countries, for the purpose of reporting, only Finland developed its own biogeochemical CH4 model to provide to GHGI a very detailed list of estimates for all CH4 sub-activities. forest productivity and composition (Lindner et al., 2015). Several studies analyzed the European forest carbon budget from different perspectives and over several time periods (Kauppi et al., 1992;Karjalainen et al., 2003), using GHG budgets from fluxes, inventories and inversions (Lyussaert et al,, 2012), flux towers (Valentini et al., 2000), forest inventories (Liski et al., 2000, Pilli et al., 2017 and IPCC guidelines (Federici et al., 2015).

440
Achieving the well-below 2 o C temperature goal of the PA requires forest-based mitigation (Grassi et al., 2018, Nabuurs et al. 2017. Currently, the EU28 forests act as a sink and forest management will continue to be the main driver affecting the productivity of European forests for the next decades (Koehl et al., 2010). Forest management, however, can enhance (Schlamadinger et al., 1996) or weaken (Searchinger et al., 2018) this sink. Forest management not only influences the sink strength, it also changes forest composition and structure, which affects the 445 exchange of energy with the atmosphere (Naudts et al., 2016), therefore the potential of mitigating climate change (Luyssaert et al., 2018;Grassi et al., 2019).
We compared CO2 net emissions/removals from the LULUCF sector reported by UNFCCC NGHGI 2018 to those included in FAOSTAT and to the carbon balance here termed as the Net Biome Production (NBP) from different models ( Table 4). The categories presented in this study are forest land, cropland and grassland. We present separate 450 the results from forest land and land use because, some models (e.g. CBM and EFISCEN) use a different definition of forest land than the Dynamic Global Vegetation Models (DGVMs) ensemble TRENDY (Sitch et al., 2008, Le Quéré et al., 2009 or bookkeeping models (Houghton & Nassikas 2017, Hansis et al., 2015.
To better illustrate differences between estimates we exemplify how four of the data sources interpret and calculate the NBP:

455
-UNFCCC NBP definition depends on the method used by each country; -CBM calculates NBP as the total ecosystem and stock change the difference between net ecosystem production (NEP) and the direct losses due to harvest and natural disturbances (e.g., fires) (Pilli et al., 2017, Kurz et al., 2009). Adding to the NBP the total changes in the harvested wood product (HWP) carbon stock, CBM estimates the net sector exchange (NSE) (Karjalainen et al., 2003, Pilli et al., 2017;

460
-EFISCEN's NBP is derived from total tree gross growth minus soil losses, minus (density related) mortality minus harvest. Natural disturbances tend to occur relatively little in Europe and, if are happening, are included in regular harvest, therefore EFISCEN does not consider them in addition for the NBP calculation; -DGVMs calculate NBP as the net flux between land and atmosphere defined as photosynthesis minus the sum of plant respiration and soil heterotrophic respiration, carbon emissions from fire (some models and CO2 465 emissions from harvested wood products) and harvest. Land use change emissions are calculated as the imbalance between photosynthesis and respiration over land areas that followed a transition. Positive flux is into the land. NBP should be equal to changes in total carbon reservoirs. The net land use change flux is derived by differencing the NBP of a simulation with and without land use change.

510
In 2015, most of the differences between FAOSTAT estimates and UNFCCC country data were generated by few countries: for Finland there is a disagreement from neutrality (due to extrapolation of previous data) in FAOSTAT to a large sink of 38 Mt CO2 yr -1 reported to UNFCCC. For Romania and Latvia we find that the FAOSTAT sink is a factor 7 larger than the reported UNFCCC, and for Denmark we find a sink according to FAOSTAT estimates 515 and a source reported to UNFCCC. When comparing NGHGI and FAO-FRA data, it should be considered that NGHGIs specifically report emissions and removals, and are formally reviewed annually, while FAO-FRA reports are not primarily for reporting CO2 emissions and removals, and are not formally reviewed . Similar as for FL, fluxes include effects from both environmental changes and from land management and land use change. In FAOSTAT GHG emissions in the domain "Cropland" and "Grassland" are currently limited to the CO2 525 emissions from cropland/grassland organic soils associated with carbon losses from drained histosols under cropland/grassland. This can be one of the reasons for differences between estimates reported by the two sources ( Fig.   11).
Cropland definition in IPCC includes arable and tillage land, and agro-forestry systems where vegetation falls below the threshold used for the forest land category, consistent with the selection of national definitions (IPCC 530 glossary). According to EUROSTAT, the term 'crop' within cropland covers a very broad range of cultivated plants.
In 2015 more than one fifth (22 %) of the EU28's area was covered by cropland (EUROSTAT, updated in 2019).
Denmark ( Grassland definition in IPCC includes rangelands and pasture land that is not considered as cropland, as well as systems with vegetation that fall below the threshold used in the forest land category.. This category also includes all grassland from wild lands to recreational areas as well as agricultural and silvo-pastural systems, subdivided into 540 managed and unmanaged, consistent with national definitions. Grasslands tend to be concentrated in regions with less favorable conditions for growing crops or where forests have been cut down. In 2015 just above one fifth of the EU28's (21 %) was covered by grassland. Some of these are found in northern Europe (  Climate change and climate effects on soil temperature and moisture are key drivers in the 21 st century increase of soil decomposition and decrease of the soil carbon stock (Smith et al., 2005). Avoiding soil carbon losses or restoring stocks requires practices that increase C input in excess of losses from erosion and decomposition, such 560 as diminished grazing intensity for grasslands, higher return of residues or reduced tillage for croplands, and manure additions for both. Further change in land use and management will also affect the soil carbon stock of European cropland and grasslands (Smith et al., 2005). that is, the sink that favorable environmental changes, in particular CO2 fertilization. This sink created over forest land in the simulation without land use change is "lost" in the simulation with land use change because agricultural land lacks the woody material and thus has a higher carbon turnover (Gasser et al., 2013. This different definition from bookkeeping models historically implies higher carbon emissions from DGVMs, even if all post-conversion carbon stocks changes were the same in DGVMs and bookkeeping models.

580
The key difference between DGVMs and bookkeeping models, on the one hand, and FAO and UNFCCC methodology, on the other, is that the latter are based on the managed land proxy (Grassi et al., 2018a) (Fig. 12).  Fig. 3 in Grassi et al., 2018a).
Land fluxes can be differentiated into three processes: 1) Direct anthropogenic effects (land-use and land use change, e.g., harvest, other management, deforestation), 2) Indirect anthropogenic effects (e.g., changes induced by 595 climate change, CO2 fertilization), and 3) Natural effects (i.e., that would happen without human caused climate  National Greenhouse Gas Inventories (NGHGI) use the notion of "managed land" as a proxy for direct "anthropogenic" emissions, hence in practice include most or all (depending on the specific method) indirect emissions into their anthropogenic estimates. In addition, the area considered "managed" by countries is typically much greater 610 than the area used by biophysical models to simulate the direct anthropogenic effects, as it includes areas that are not actively managed (for instance, forest parks or forest seldomly harvested) (Grassi et al. 2018a).
The differences between biogeochemical models and NGHGI of around 4-5Gt CO2 yr -1 globally is to a large part attributable to the accounting of indirect effects on managed land towards AFOLU emissions for NGHGI (Grassi et al., 2018a, IPCC SRCCL). The differences at the EU28 level are much smaller, because nearly all forest land is 615 managed in the EU.
Independent estimates of the land-related flux for the EU28 are presented in Figure 13. The data behind the three main estimates, bookkeeping models, NGHGI and FAOSTAT represent the total net land use emissions/removal from forest, cropland and grassland, including conversions to and from one category to another. Next to them, we plotted each of the net land use change flux (in grey) (difference of simulation with and without land use change) from 620 eight of the DGVMs TRENDYv6 with their mean, as they mostly simulate the indirect and natural sink considered unmanaged. FAOSTAT includes emissions from peatland drainage and fires, and from biomass fires (not considered herein  (Grassi et al., 2018a). Differences between the two bookkeeping models, BLUE and H&N, relate to the different forcing applied by each of the models and differences in biome types. The forcing used by H&N is based directly on FAOSTAT/FRA agricultural and wood harvest data, while BLUE uses LUH2 (Hurtt et al., 2011(Hurtt et al., , 2018. LUH2 is based on HYDE3.  At European level the largest inconsistencies between estimates from AFOLU emission sources/sinks were found to be mainly caused by the use of different methodologies, including use of different AD and/or Tier level. When looking at final emission estimates, inconsistencies in methodology and Tier application in calculating emissions give as much as 10-20 % variation across estimates (e.g. CH4 from agriculture),. Higher tiers require more detailed AD for calculating emissions/removals from AFOLU sectors.

665
Within the UNFCCC practice, for agriculture, each country uses its own country specific method which takes considers specific national circumstances (as long as they are in accordance with the 2006 IPCC Guidelines) as well as IPCC default values, which are usually more conservative. The EU GHG inventory underlies the assumption that the individual use of national country specific methods leads to more accurate GHG estimates than the implementation of a single EU wide approach (UNFCCC, 2018b). The Tier level a country applies depends on the national 670 circumstances, which explains the variability of uncertainties among the sector itself as well as among EU countries. For the UNFCCC, throughout the variability of the analyzed national GHG inventories, it turned out that N2O emissions from manure management and direct and indirect emissions together with CH4 emissions from rice 700 cultivation have the largest uncertainties. When we aggregated UNFCCC uncertainties at country level (using the methodology described in Appendix C), we also noticed the fact that not all countries report sub-sectoral uncertainties (e.g. Greece for grazing) and some countries (Sweden, Poland, Croatia and Czech Republic) had no uncertainty analysis performed for all sub-activities due to lack of data (e.g. confidential data).
There is as well the need to define a common methodology for overall uncertainty calculation while checking 705 for consistency in the way uncertainties are calculated for different data sources and the way data is aggregated for different sectors. We noticed that for agricultural N2O emissions the split in sub-activities is not always consistent with IPCC sectors and this leaves room to differences when aggregating the results (Table 3).

710
For the LULUCF sector, methods for the estimation of GHGs and CO2 fluxes differ enormously among countries and land use categories. Within the UNFCCC practice, each country uses its own country specific method which considers specific national circumstances (as long as they are in accordance with the 2006 IPCC GLs), as well as IPCC default values, which result in higher uncertainties. When we analyze the estimates from multiple sources 715 (inventories and models) we observe that, published estimates contain two main sources of uncertainties: a) differences due to input data and structural/parametric uncertainty of models (Houghton et al., 2012); b) differences in definition Grassi et al., 2018b). These differences result from choices in the simulation setup, and are partly predetermined (for b) in particular) by the type of model used: bookkeeping models, DGVMs, or inventorybasedand whether fluxes are attributed to LULUCF emissions due to the cause or place of occurrence (indirect 720 fluxes on managed land included in GHGIs and FAOSTAT). Differences in definitions and methodology calculation of estimates across model types is crucial and may lead to model-to-model variability. In Figure 13 the variability between the mean of the DGVMs ranges between 44 % in 1996 and 186 % in 2016 (distance between interquartile range and median across models for each year).
Depending on the degree of independence between assumptions, variability can become a reliable proxy for 725 structural uncertainty when more accurate estimates are lacking (Solazzo et al., 2017). In general the definition of NBP denotes the net gain or loss of carbon from a region. NBP is equal to the Net Ecosystem Production (NEP) minus the carbon lost due to a disturbance (e.g., a forest fire, freshwater CO2 emissions or a forest harvest) taking into account as well the net C balance of harvested products ( The different definitions and concepts used by the global models and inventory communities mean that the land fluxes cannot necessarily be consistently compared. The framework developed by Grassi et al. (2018a) and shown 755 in Figure 12 can be generalized to make a more direct comparison. Figure 14 disaggregates managed forest land into components that are reported in the UNFCCC CRFs: converted land (e.g., land changing from cropland to forest land), HWPs, and the remaining land (e.g., forest land remaining forest land) is split into land that is "production" (forestry) or land that is used for "ecological or social functions", based on the definitions of managed land. Unmanaged land cannot have direct human induced effects.   (Hansis et al., 2015;Houghton &  Overall, our results suggest that most of the LULUCF emissions in the EU28 are from direct effects in the 775 managed forest sector, including age-legacy effects (forest expansion and regrowth after WWII), with small net emissions from land conversion as they are largely compensated by deforestation (from CRFs). With appropriate data and models, it is theoretically possible to expand and enumerate the estimates more accurately.

780
There are many independent estimates of GHG emissions, but adequate understanding of their differences (either qualitatively or quantitatively) is lacking. For CH4 and N2O the main differences between countries reports and models are the use of tiers and methodologies (for both emissions and uncertainty calculation). Countries reporting to UNFCCC use an inconsistent mix of tiers depending on the animal type and activity following the approach described 785 by the 2006 IPCC Guidelines, while models run with more accurate data being able to disaggregate better the activities.
One detected similarity between all sources is the use of EFs, as almost all sources make use of the IPCC defaults.
AD is as well somehow shared, coming mostly from the MS, FAOSTAT, Eurostat or UNFCCC, with the flow between these four sources not totally understood. At EU28 level, countries are generally doing well in reporting their total GHG emissions but there is large 790 room for improvement mainly when looking at differences between UNFCCC Tier use and models (e.g. for CH4 from agriculture 10-20 % difference). We stress the need of looking as well on the LULUCF CO2 estimates, where a quantification of differences between net emission estimates (inventories, models etc.) caused by inconsistencies in methodology and/or Tier application in the EU28 is not yet available. More data is needed to account for and reduce these differences. Narrowing down the analysis to sensitive parameters (e.g. AD) which may trigger the differences 795 (e.g. Appendix A, Table A1) also requires more information on uncertainties.
As previously discussed, it is of great importance to better distinguish between direct and indirect effects on land use emissions especially for the purpose of reconciling land-related emissions from global datasets and NGHGI.
Currently our comparisons give significant uncertainty, mostly related to coverage of different land use practices and the differences in definitions (Fig. 12).

800
It is also important to recognize that just because independent inventories agree well for a sector, does not necessarily mean that the estimate is better in the sense that it is closer to real emissions. The reason for agreement across inventories may simply be that the different inventories used the same methodology and data sources. In recent years there has been increased attention to the quantitative differences between land-based CO2 emissions, with a much better understanding between inventories and estimates from the scientific community. However, there remain 805 gaps in our understanding of differences between FAOSTAT and UNFCCC and between different DGVMs and bookkeeping models. One explanation can be linked to the fact that models use different methods to estimate emissions/removal then countries use in reporting to UNFCCC.

The current atmospheric GHG network is coordinated by the Integrated Carbon Observation System (ICOS)
infrastructure at the European level. Within the future UNFCCC reporting framework, we argue that countries should 810 use, whenever possible, global inversions to provide additional constraints for the verification and reconciliation purposes. Within the VERIFY project framework, we will use in a following study, inversions based on better, higher resolution, transport models to assimilate the precise ICOS GHG concentration data complemented by satellite retrievals of column CO2, CH4 and N2O concentrations. The main challenge for the inversion community remains the separation of natural and anthropogenic part of the total emission column. For the moment, global inverse models are 815 widely used to estimate emissions of CH4 and N2O at global/continental scale, using mainly high-accuracy surface measurements at remote stations (e.g. Bergamaschi et al., 2013;Bousquet et al.,2006;Mikaloff Fletcher et al., 2004a, b;Saunois et al., 2016, Hirsch et al., 2006Huang et al., 2008;Saikawa et al., 2014;Thompson et al., 2014b;Wells et al., 2018, InGOS JRC report, 2018.

850
However, these requirements only commit developed country parties listed in the Annex I of the UNFCCC. This is explained by the fact that in the 1990s, when the Convention's and the Kyoto Protocol's reporting system was developed and adopted, there was a clear division of the regional distribution of GHG emissions. In industrialized countries, most GHG emissions were released, while in developing and emerging countries, emissions were low (Berger et al., 2016). Therefore, the Convention based on the principle of common but differentiated responsibilities. give the range (confidence interval) within which the underlying value of an uncertain quantity is thought to lie for a specified probability.
According Chapter 3 there are two ways uncertainties can be calculated: a) Where uncertain quantities are to be combined by multiplication, the standard deviation of the sum will be the 885 square root of the sum of the squares of the standard deviations of the quantities that are added. b) Where uncertain quantities are to be combined by addition or subtraction, the standard deviation of the sum will be the square root of the sum of the squares of the standard deviations of the quantities that are added. For this study an analysis of the reported uncertainties under the NGHGI for CO2, CH4 and N2O has been performed for 26 EU countries 4 . The analysis has not been performed for Sweden and Czech Republic due to lack of 890 data (e.g. confidential data). To identify the main uncertainties, the Approach 1: propagation of error, has been applied to each country's uncertainty assessment under the NGHGI.
Since the EU MS report all on different subsectors, the uncertainties have been aggregated to the subsectors per gas that all countries have in common, see the following  Generally, for almost all countries, the uncertainties for CO2, CH4 and N2O in the agriculture sector, LULUCF sector are rather high and variable compared to the other sectors. For the EU as a whole, the level uncertainties vary by sector; for the agriculture sector it is 45.4%, and for the LULUCF sector it is 33% (UNFCCC, 2018b). This is 900 because of the inherently different aspects of these sectors due to their dependencies on a number of variable factors and parameters, which make it harder to measure greenhouse gases accurately. For example, Rypdal & Winiwarter (2001) claim that it is the incomplete understanding of soils that is the largest contribution to national uncertainty assessments, which can be confirmed with the uncertainty analysis. N2O emissions in soil are affected by microbiological activity and processes, the natural variation in soil conditions and the impacts of inter-annual variation 905 in climate on the emissions, making it difficult to measure. Other important contributions to the overall uncertainty are uncertainties about the amount of solid waste (organic material that decomposes to produce methane) that is deposited and the extent of land use change.
Since the 2015, following decisions of COP19 5 ,inventories of Annex I need to be reported annually by 15 th April following the 2006 IPCC Guidelines (Eggleston et al., 2006), using the spreadsheets with the Common Reporting 910 Format (CRF), using the GWP100 of AR4 and following the new structure for sectoral specifications but keeping within the AFOLU sector Agriculture and LULUCF distinguished 6,7 .
The Revised UNFCCC Reporting Guidelines for Greenhouse Gas Inventories of Parties in Annex I to the Convention (UNFCCC, 2013), here after UNFCC Reporting Guidelines, define what and how to report GHG emission by source and removals by sinks in order to comply with requirements. However, these requirements only commit 915 developed country parties listed in the Annex I of the UNFCCC. This is explained by the fact that in the 1990s, when the Convention's and the Kyoto Protocol's reporting system was developed and adopted, there was a clear division of the regional distribution of GHG emissions. In industrialized countries, most GHG emissions were released, while in developing and emerging countries, emissions were low (Berger et al., 2016). Therefore, the Convention also includes the principle of common but differentiated responsibilities. Developing countries are not requested to provide detailed 920 information on national GHG emissions.

950
-When dealing with CO2, full correlation is assumed for subsets sharing the same emission factors (typically fuel-dependent); -Aggregated emissions from same categories but different countries assumes full correlation, unless the emission factor is country-specific, or derived from higher tiers (i.e. not default EF defined by IPCC).
In addition, the following assumption is adopted:

955
-When uncertainty is defined within a range (e.g. for the energy sector, IPCC recommend that the methane emission factors are treated with an uncertainty ranging from 50% to 150%), the upper bound of the range is assigned to developing countries, whilst the lower bound to developed countries. Uncertainty of country or process-specific EF is not propagated (no correlation).
Although assuming full correlation when aggregating emissions is quite conservative (overestimating the 960 uncertainty introduced by emission factors), this approach is intended to balance for other sources of uncertainty that are not taken into account, such as covariance among activity data (deemed negligible), uncertainty of technologies factors (no information available as to how these factors are uncertain, as for example on the different rice cultivar practices), and uncertainty due to the 'fast track', i.e. applying trends to estimate latest year's emissions.
Based on the model framework, each stand is described by area, age and land use classes and up to 10 classifiers based on administrative and ecological information and on silvicultural parameters (such as forest composition and management strategy). A set of yield tables define the merchantable volume production for each 1040 species while species-specific allometric equations convert merchantable volume production into aboveground biomass at stand-level. At the end of each year the model provides data on the net primary production (NPP), carbon stocks and fluxes, as the annual C transfers between pools and to the forest product sector.
The model can support policy anticipation, formulation and evaluation under the LULUCF sector, and it is used to estimate the current and future forest C dynamics, both as a verification tool (i.e. to compare the results with 1045 the estimates provided by other models) and to support the EU legislation on the LULUCF sector (Grassi et al., 2018a).
In the biomass sector, the CBM can be used in combination with other models, to estimate the maximum wood potential and the forest C dynamic under different assumptions of harvest and land use change (Jonsson et al., 2018).

1050
The European Forest Information SCENario Model (EFISCEN) is a large-scale forest model that projects forest resource development on regional to European scale. The model uses national forest inventory data as a main With the help of biomass expansion factors, stem wood volume is converted into whole-tree biomass and subsequently 1055 to whole tree carbon stocks. Information on litter fall rates, felling residues and natural mortality is used as input into the soil module YASSO (Liski et al. 2005), which is dynamically linked to EFISCEN and delivers information on forest soil carbon stocks. The core of the EFISCEN model was developed by Prof. Ola Sallnäs at the Swedish Agricultural University (Sallnäs 1990

Bookkeeping models
The LULUCF chapter makes use of data from two bookkeeping models: H&N (Houghton & Nassikas, 2017) and BLUE (Hansis et al., 2015). Bookkeeping models (Houghton, 1983)       To check for correctness, the total uncertainty for the aggregated sectors can be calculated. If the total uncertainty for the aggregated sectors matches the total uncertainty of the uncertainty assessment, the calculated uncertainties for the subsectors are correct. This was the case for all calculations performed for this analysis.

1145
All raw data files reported in this work which were used for calculations and figures are available for public download at http://doi.org/10.5281/zenodo.3460311 (Petrescu et al., 2019).The data we submitted is reachable with one click (without the need for entering login and password), and a second click to download the data, consistent with the two-click access principle for data published in ESSD (Carlson and Oda, 2018). The data and the DOI number is subject to future updates and it refers only to this version of the manuscript.