Anthropogenic and natural methane fluxes in Switzerland synthesized within a spatially-explicit inventory

: We present the first high-resolution (500 m × 500 m) gridded methane (CH4) emission inventory for Switzerland, which integrates the national emission totals reported to the United Nations Framework Convention on Climate Change (UNFCCC) and recent CH4 flux studies conducted by research groups across Switzerland. In addition to anthropogenic emissions, we also include natural and semi-natural CH4 fluxes, i.e., emissions from lakes and reservoirs, wetlands, wild animals as well as uptake by forest soils. National CH4 emissions were disaggregated using detailed geostatistical information on source locations and their spatial extent and process- or area-specific emission factors. In Switzerland, the highest CH4 emissions in 2011 originated from the agricultural sector (150 Gg CH4 yr−1), mainly produced by ruminants and manure management, followed by emissions from waste management (15 Gg CH4 yr−1) mainly from landfills and the energy sector (12 Gg CH4 yr−1), which was dominated by emissions from natural gas distribution. Compared to the anthropogenic sources, emissions from natural and semi-natural sources were relatively small (6 Gg CH4 yr−1), making up only 3 % of the total emissions in Switzerland. CH4 fluxes from agricultural soils were estimated to be not significantly different from zero (between −1.5 and 0 Gg CH4 yr−1), while forest soils are a CH4 sink (approx. −2.8 Gg CH4 yr−1), partially offsetting other natural emissions. Estimates of uncertainties are provided for the different sources, including an estimate of spatial disaggregation errors deduced from a comparison with a global (EDGAR v4.2) and a European CH4 inventory (TNO/MACC). This new spatially-explicit emission inventory for Switzerland will provide valuable input for regional scale atmospheric modeling and inverse source estimation. Abstract We present the ﬁrst high-resolution (500m × 500m) gridded methane 4 ) emission inventory for Switzerland, which integrates the national emission totals reported to the United Nations Framework Convention on Climate Change (UNFCCC) and recent CH 4 ﬂux studies conducted by research groups across Switzerland. In addition to anthro- 5 pogenic emissions, we also include natural and semi-natural CH 4 ﬂuxes, i.e., emissions from lakes and reservoirs, wetlands, wild animals as well as uptake by forest soils. National CH 4 emissions were disaggregated using detailed geostatistical information on source locations and their spatial extent and process- or area-speciﬁc emission factors. In Switzerland, the highest CH 4 emissions in 2011 originated from the agricultural 10 sector (150 Gg CH 4 yr − 1 ), mainly produced by ruminants and manure management, followed by emissions from waste management (15 Gg CH 4 yr − 1 ) mainly from landﬁlls and the energy sector (12 Gg CH 4 yr − 1 ), which was dominated by emissions from natural gas distribution. Compared to the anthropogenic sources, emissions from natural and semi-natural sources were relatively small (6 Gg CH 4 yr − 1 ), making up only 3 % of 15 the total emissions in Switzerland. CH 4 ﬂuxes from agricultural soils were estimated to be not signiﬁcantly di ﬀ erent from zero (between − 1.5 and 0 Gg CH 4 yr − 1 ), while forest soils are a CH 4 sink (approx. − 2.8 Gg CH 4 yr − 1 ), partially o ﬀ setting other natural emissions. Estimates of uncertainties are provided for the di ﬀ erent sources, including an estimate of spatial disaggregation errors deduced from a comparison with a global 20 (EDGAR v4.2) and a European CH 4 inventory (TNO/MACC). This new spatially-explicit emission inventory for Switzerland will provide valuable input for regional scale atmospheric modeling and inverse source estimation.

pogenic emissions, we also include natural and semi-natural CH 4 fluxes, i.e., emissions from lakes and reservoirs, wetlands, wild animals as well as uptake by forest soils. National CH 4 emissions were disaggregated using detailed geostatistical information on source locations and their spatial extent and process-or area-specific emission factors. In Switzerland, the highest CH 4 emissions in 2011 originated from the agricultural sector (150 Gg CH 4 yr −1 ), mainly produced by ruminants and manure management, followed by emissions from waste management (15 Gg CH 4 yr −1 ) mainly from landfills and the energy sector (12 Gg CH 4 yr −1 ), which was dominated by emissions from natural gas distribution. Compared to the anthropogenic sources, emissions from natural and semi-natural sources were relatively small (6 Gg CH 4 yr −1 ), making up only 3 % of 15 the total emissions in Switzerland. CH 4 fluxes from agricultural soils were estimated to be not significantly different from zero (between −1.5 and 0 Gg CH 4 yr −1 ), while forest soils are a CH 4 sink (approx. −2.8 Gg CH 4 yr −1 ), partially offsetting other natural emissions. Estimates of uncertainties are provided for the different sources, including an estimate of spatial disaggregation errors deduced from a comparison with a global (Xinmin, 2004). However, this replacement process is now completed (Xinmin, 2004) and, combined with the projected higher demand for natural gas due to the new Swiss energy strategy (SFOE, 2012a), emissions are expected to increase again. Disposal of combustible waste in landfills has been prohibited since 2000 in Switzerland and therefore emissions from this source are decreasing (FOEN, 2013). Counteracting this 15 trend, emissions from biogas production have more than doubled since 1990 and are projected to rise even further (FOEN, 2013). Thus, a reliable base-line inventory for CH 4 emissions from anthropogenic activities and natural processes is urgently needed. Inventory estimates such as those presented in the SGHGI are based on numerous assumptions and statistical data that are associated with large uncertainties in several 20 categories. Hence, validation by independent means is essential. Attempts have been made to constrain regional or national CH 4 emissions by atmospheric concentration measurements using boundary-layer budgets (Gallagher et al., 1994;Choularton et al., 1995;Fowler et al., 1996;Beswick et al., 1998;Wratt et al., 2001), inverse emission modeling (Vermeulen et al., 1999;Bergamaschi et al., 2005;Manning et al., 2011;Introduction Conclusions References Tables  Figures   Back  Close Full Screen / Esc

Printer-friendly Version
Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | rather than on providing representative numbers for national emission budgets. Only one single attempt has so far been made to upscale CH 4 flux measurements to national totals or to validate the Swiss CH 4 inventory with atmospheric measurements (Hiller, 2012). In order to model the influence of CH 4 emissions on atmospheric concentrations, 5 spatially-explicit inventories are needed in addition to total national emissions (Bun et al., 2010). To disaggregate emissions to a higher spatial resolution, detailed knowledge on the location and the activity of each source is required, leading to additional uncertainty (Ciais et al., 2010). In recent years, the increasing targeting of the atmospheric and inverse modeling community on the regional and urban scale led to 10 a clear trend towards high-resolution inventories. Currently, four different CH 4 inventories include Switzerland. EDGARv4.2, EDGAR-HTAP and TNO-MACC focus on anthropogenic emissions, NatAir considers only natural and biogenic emissions (details are shown in Table 1); thus, no inventory combines all CH 4 sources. Although these inventories have benefited from a considerable increase in resolution (e.g. EDGAR 15 changed in 2009 from 1 • × 1 • to 0.1 • × 0.1 • ), they are still limited to cell sizes of about 10 km × 10 km. For a spatially heterogeneous country such as Switzerland, this resolution is still too coarse to capture local variations. The goal of this study was to produce the first gridded, high-resolution (500 m×500 m) CH 4 inventory for Switzerland. Anthropogenic emission estimates followed the method- 20 ologies of the SGHGI (FOEN, 2013). National totals were either spatially disaggregated across Switzerland using respective correlated geostatistical data with at least 500 m × 500 m spatial resolution, or generated in a bottom-up approach using emission factors (EFs). The SAEFL (1996) fluxes were updated and total emissions for all relevant categories are reported together with their uncertainties. The additional uncer- 25 tainty at the grid level introduced by the spatial disaggregation was estimated by comparing different inventories. For each source category, a brief review of recent research studies in Switzerland is presented and the results from field studies are compared with the inventory estimates where possible. Introduction

Data collection and processing
Our spatially-explicit inventory is based on the SGHGI for anthropogenic emissions and additionally comprises natural fluxes. National emission totals were distributed onto a 500 m × 500 m grid according to correlated geostatistical data. Natural sources without up-to-date national totals available were up-scaled and spatially attributed using 5 geostatistical data. Emission factors were adapted from the literature, including dedicated studies for Switzerland. The spatially-explicit inventory was generated for the year 2011, which is the latest year available from the SGHGI. Sources are represented by positive numbers, sinks by negative numbers. 10 The latest submission of the SGHGI to the UNFCCC on 15 April 2013 reports greenhouse gas (GHG) emissions by sources and removals by sinks between 1990 (base year) and 2011 (FOEN, 2013). A detailed description of the institutional arrangements for inventory preparation, data sources and methodologies, uncertainty evaluations as well as quality assurance/quality control (QA/QC) activities are given in the SGHGI 15 (FOEN, 2013). The inventory preparation follows the reporting guidelines developed by the Intergovernmental Panel on Climate Change (IPCC) (IPCC, 1997(IPCC, , 2000(IPCC, , 2003(IPCC, , 2006. To estimate GHG emissions and removals following the IPCC methodology, three approaches differing in complexity (so-called tiers) can be used. The Tier 1 methodology uses generalized default equations and parameters provided by the IPCC 20 guidelines. Tier 2 employs country-specific input data, providing more detail of the underlying processes with regional specificities. Tier 3 is the most complex approach in terms of capturing dynamic processes and their spatial stratification, involving domestic measurements and/or modeling. The UNFCCC encourages parties to develop Tier 3 methods for large sources and sinks as well as for those with temporal trends, the Introduction For the spatially-explicit inventory, the eight strongest CH 4 sources out of a total of 620 listed in the SGHGI (FOEN, 2013) were selected, adding up to about 90 % of all anthropogenic CH 4 emissions. These eight sources include emissions from the agricultural sector (41 % from enteric fermentation of dairy cattle, 17 % from young cattle, 5 % from suckler cows, 2 % from sheep, 9 % from manure of dairy cattle and 5 5 % of swine), the waste sector (5 % from landfills), and the energy sector (5 % from losses from natural gas distribution), where percentages in parentheses represent the share of the total 2011 anthropogenic emission estimate of 178 Gg. Additionally, we also compiled a spatial inventory of the emissions from wastewater treatment plants as these act as strong local sources.

Agricultural sector
The largest agricultural source in the SGHGI is 4.A Enteric Fermentation followed by 4.B Manure Management (the headings and numbers correspond to the official nomenclature for reporting, see Table 2). Agricultural residue burning is only a small source in Switzerland and reported in the sector 6. Waste, whereas emissions from agricultural 15 soils, rice production, and burning of savannas are negligible.
The 57 600 registered farms manage about one third of Switzerland's area (15 000 km 2 including alpine pastures; FSO, 2013) and rear animals equivalent to 1 316 600 livestock units (FSO, 2011). Agricultural CH 4 emissions from livestock result from the microbial degradation of 20 carbohydrates present in the rumen of ruminants, and to a lesser extent also in the hindgut of all herbivores (Jensen, 1996). Additionally, carbohydrates that are not digested and thus are excreted as volatile solids can subsequently be converted to CH 4 during manure management. Overall, the CH 4 production from enteric fermentation is primarily related to feed intake, standardized by using gross energy intake (GE) for 25 inventory purposes. Intake differences quite reliably reflect variations in animal weight and performance (milk yield, growth, and pregnancy) and corresponding differences in CH 4 emissions (Soliva, 2006 strates for the methanogenic archaea, is not accounted for. CH 4 production is assumed to decline when forage is partially replaced with concentrate in the ruminant diet (Beauchemin et al., 2008), but this reduction is often smaller than assumed, and about one third of this reduction may be subsequently compensated by correspondingly higher manure-derived CH 4 emissions (Hindrichsen et al., 2006). Since Swiss ruminant diet 5 types are mostly forage-based, CH 4 conversion rates measured in Switzerland are higher than IPCC (2006) default values . On the other hand, experiments on CH 4 emissions from Swiss manure management result in lower emissions than currently estimated in the SGHGI using IPCC (2006) default values . In particular, emissions from liquid manure systems tend to be lower than those currently reported. The influence of animal genotype on the CH 4 emission potential is currently discussed at the global level, but Swiss studies do not indicate significant differences between dairy breeds (Münger and Kreuzer, 2006). In conclusion, preliminary analyses suggest no significant change in CH 4 emissions from livestock by applying Swiss-specific EFs , as different under-and overestimates compen-15 sate each other. However, this conclusion does not yet consider the potential to reduce GHG emissions using different feeding measures (addition of lipids, plant secondary compounds, etc.; see e.g., Beauchemin et al., 2008;Staerfl et al., 2012). For our spatially-explicit inventory, emissions were calculated from livestock numbers in 2007, aggregated by farm (agricultural establishment census 2007;FSO, 2009), 20 and multiplied with animal-specific EFs from the Swiss national air pollution database (EMIS, Federal Office for the Environment). Emissions for 2007 were then scaled to the 2011 value reported in the SGHGI. Following the Swiss husbandry practice, most emissions were assumed to be produced in the stall (80 % for cattle, 20 % for sheep, 100 % for swine) and the remaining fraction on the pastures. The agricultural establish-25 ment census contains the location of the main farm building at one hectare resolution and was assumed identical to the stable and manure storage location (Kupper et al., 2010). Emissions on pastures were attributed to all grid cells covered by this land use type (Swiss land use statistics; FSO GEOSTAT, 2009)  Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | spective farm. As part of the Swiss farming practice, part of the livestock is moved to alpine pastures in summer. Consequently, the CH 4 emissions produced there were also allocated on those alpine pastures (≈ 4 % of the agricultural emissions included in the spatially-explicit inventory). 5 Within the waste management sector, CH 4 emissions originate mainly from 6.A Solid Waste Disposal on Land, 6.B Wastewater Handling, and 6.D Other processes, including composting, digestion of organic waste, and biogas up-grading.

Landfills
Gas production by decomposition of organic material in the anoxic waste body (typically 10 50-70 % CH 4 (v/v), 30-50 % CO 2 , and trace amounts of other gases; Farquhar and Rovers, 1973) leads to advective and diffusive gas transport within the landfill pore system and eventually to emissions to the atmosphere (e.g. Franzidis et al., 2008). Recent research activities related to landfill-derived CH 4 in Switzerland are limited to the municipal waste landfill Lindenstock near Liestal. This 12 ha landfill received 15 ≈ 3.2 × 10 6 m 3 of household, construction, and commercial waste between 1949 and 1994. Following closure, the waste was capped with a 2 to 2.5 m thick cover soil consisting primarily of silty loam, and a gas-collection system was installed constructed of vertical and horizontal, partially-screened, high density polyethylene pipes. However, gas collection has not been attempted in recent years, and gas outlets remain closed 20 with screw-cap lids. This is a unique feature of this landfill as gas-collection systems on several other Swiss landfills are either in continuous operation or absent. Experiments at Lindenstock compared CH 4 fluxes obtained by different methods Eugster and Plüss, 2010;Schroth et al., 2012) at or above the cover-soil surface, as well as below-ground fluxes. Results indicated that the stud-Introduction sions close to the gas-collection outlets (daily mean fluxes ranging between 0.05 and 1.5 g CH 4 m −2 d −1 ) (Henneberger et al., 2012;Schroth et al., 2012). A net flux of up to −0.002 g CH 4 m −2 d −1 (uptake) was usually observed away from the gas-collection outlets. CH 4 efflux from the waste body was highly variable over short distances and time. CH 4 oxidation activity in the cover soil was generally high, mitigating most of the produced CH 4 , but also exhibiting substantial spatial variability (estimated to −1.92 to −64 g CH 4 m −2 d −1 in 2010) being strongest where efflux from the waste body was highest. Mitigation of landfill-derived CH 4 in the Lindenstock cover soil is mediated by a highly diverse, abundant methanotrophic community (Henneberger et al., 2012(Henneberger et al., , 2013. Similar experiments during winter indicated stronger net CH 4 emissions (up to 2.5 g CH 4 m −2 d −1 ) from the studied landfill section (Ugolini et al., 2009). This was primarily attributed to a decrease in oxygen availability within the cover soil as a result of increased soil water content at shallow depths, but also to a decrease in temperature, which both adversely affected the CH 4 oxidation activity.
These results from Lindenstock are not sufficiently representative for estimating total 15 landfill emissions in Switzerland, but they broadly agree with previous studies on landfills in Europe and the USA, reporting oxidation activities in a similar but higher range. Hence, cover soils in general provide an effective buffer for landfill-derived CH 4 , mitigating emissions to the atmosphere as a result of the activity of methanotrophs (Whalen et al., 1990;Boeckx et al., 1996;Börjesson et al., 1998;Chanton et al., 2009

Wastewater treatment
CH 4 is produced in the sewage system as well as in the anoxic part of the wastewater treatment plant and the upgrading of sewage gas. To our knowledge, no direct wastewater CH 4 measurements exist for Switzerland; however, the CH 4 emissions can be estimated from the organic load in the wastewater. The chemical oxy-5 gen demand (COD) ranges from 100 to 110 g COD person −1 d −1 , with one third each being aerobically respired, converted to CH 4 , and remaining in the sewage sludge. A large part of the CH 4 produced is used for power and heat supply of the wastewater plant, and only about 10 % is directly emitted to the environment. The resulting EF of 0.9 g CH 4 person −1 day −1 lies within the range of reported conversion rates by Deal- CH 4 also depends on the sewage system (higher with long pipes at low inclination) and the plant type (higher with uncovered anoxic post-digester). Using an average EF of 0.9 g CH 4 person −1 day −1 and a 12 million population equivalent (Swiss population plus industrial wastewater load converted to additional population), annual emissions 15 from wastewater collection and treatment result in about 4 Gg CH 4 yr −1 . However, a recent publication proposes a higher EF of 1.5 g CH 4 person −1 day −1 , arguing that CH 4 production in the sewage system was underestimated (Wunderlin et al., 2013). The resulting CH 4 emissions would increase by 50 % to about 6 Gg CH 4 yr −1 . In contrast, the SGHGI is based on a completely different method reporting only 0.48 Gg CH 4 yr −1 , be-20 cause the emissions are estimated from loss rates within the individual plant units and the total CH 4 used for energy or biogas production. Hence, emissions from tanks that are not connected to the gas system and emissions in the sewage are not included. To be consistent with the SGHGI, we proportionally distributed the 0.48 Gg CH 4 yr −1 to the 854 plants in Switzerland based (Foen, 2012) on their capacity expressed in popu- 25 lation equivalents. However, emissions might turn out up to a factor twelve higher using alternative estimation approaches. 10,2013 Anthropogenic and natural CH 4 fluxes in Switzerland Gas; FOEN, 2013). In Switzerland, 12.2 % of the total energy consumption was covered by natural gas in 2011 (SFOE, 2012b). Gas distribution in Switzerland includes ≈ 19 000 km of pipelines, from which 12 % are operated at pressures > 5 bar, 23 % between 1 and 5 bar, and 65 % < 1 bar. Another ≈ 6000 km of pipes guarantee the final distribution to the end user (SGWA, 2012). A large proportion of the transported gas transits Switzerland on the way from the production sites in northern Europe to Italy (Xinmin, 2004).

BGD
In the SGHGI, fugitive emissions of natural gas are estimated based on the amount of transported gas as well as on the infrastructure, namely the length, type and pressure of the gas pipelines (FOEN, 2013). Most emissions are assumed to occur during final 15 distribution and consumption, while emissions from welded high-pressure pipes are assumed to be low (Xinmin, 2004). Therefore, emissions reported in the SGHGI were distributed close to the gas consumers for our spatially-explicit inventory. Based on the national buildings and dwellings survey (FSO, 2010), the national emissions were proportionally distributed to those areas where natural gas is used for heating, i.e., 20 to each 1 ha grid cell where at least two houses are heated with natural gas. The emissions were subsequently aggregated onto the 500 m × 500 m grid.

Natural and semi-natural CH 4 sources and sinks
The SGHGI only reports anthropogenic CH 4 emissions while natural and semi-natural fluxes are omitted, except for wildfires in the Land Use, Land-Use Change and Forestry BGD 10,2013 Anthropogenic and natural CH 4 fluxes in Switzerland dated based on new EF and compiled into our spatially-explicit inventory as described below.

Lakes and reservoirs
Approximately 3.5 % of Switzerland (1450 km 2 ) is covered by lakes and reservoirs (FSO GEOSTAT, 2009), which can emit significant amounts of CH 4 (Bastviken et al., . These CH 4 emissions can occur via four main pathways: (1) standard gas exchange at the air-water interface; (2) ebullition (bubbling) from aquatic sediments; (3) turnover of a stratified water column with storage of CH 4 in (anoxic) bottom water; and (4) transport by plants in the shallow littoral zones (Chanton and Whiting, 1995;Bastviken et al., 2004). Hydropower reservoirs have an additional fifth emission path-10 way as they release water for energy production. Often the turbine intakes of a hydropower dam are located in the CH 4 -rich bottom water of a stratified reservoir, thus CH 4 can be emitted via degassing at the turbine or along the downstream river to which the water is released (Kemenes et al., 2007). The most important sink for CH 4 in aquatic environments is oxidation, which occurs mostly at oxic/anoxic boundaries in 15 the sediment (e.g. Frenzel et al., 1990) or water column (e.g. Schubert et al., 2010) and can account for a significant reduction of total CH 4 produced by decomposition of organic material in a lake before the CH 4 reaches the atmosphere. Measuring all of these CH 4 transport pathways and their spatiotemporal variability in a single lake requires immense effort. Thus, often only a subset of all possible path-20 ways is directly measured while others are either neglected or estimated from literature data. Truly accurate and validated models for estimating CH 4 emissions via all these pathways do not exist. An approach that can be used when attempting to estimate CH 4 emissions from a large amount of lakes without direct measurements is to use the equations proposed by Bastviken et al. (2004), which estimate diffusion, ebullition, and 25 storage emissions based on comprehensive measurements from a collection of North American and European lakes (see Supplement for details). Bastviken et al. (2004) found significant relationships between the CH 4 emission estimates and measurable 15193 Introduction variables, such as lake area, dissolved organic carbon and phosphorus concentrations, water depth, and volume of the anoxic fraction of the water column. Following Bastviken et al. (2004), we estimated diffusion, ebullition and storage emissions of CH 4 from the lake areas of all major Swiss water bodies, but made the following modifications. (1) We tripled CH 4 emissions from lakes and reservoirs shallower than 5 30 m based on direct measurements of emissions from a small, shallow lake in the low alpine region which indicated high rates of ebullition (Schubert et al., 2012). (2) For the hydroelectric reservoir Lake Wohlen on the Swiss Plateau, we directly used the emission estimate of DelSontro et al. (2010) that was based on a year-long measurement study and results in a value ten times higher than that obtained with the 10 method of Bastviken et al. (2004). No adjustments to any other reservoirs were made as Lake Wohlen may not be a representative system within Switzerland. (3) We provided a rough temporal variability of ebullition emissions by assuming that ebullition occurs only during the warmest half of the year as DelSontro et al. (2010) found a strong correlation between emissions and seasonal water temperatures. (4) Finally, we as- 15 sumed ebullition not to be a relevant process in high alpine lakes at altitudes above 1500 m a.s.l. since they receive only little organic input, have low water temperature, and quite low CH 4 concentrations in the water column (Diem et al., 2012). The resulting emission factors are summarized in Table S1. The above presented modifications suggest that other factors in addition to those proposed by Bastviken et al. (2004) may 20 need to be considered for estimating CH 4 emissions from lakes and reservoirs in the future.
The locations and areas of Swiss lakes and reservoirs were taken from the primary surfaces of the digital version of the Swiss topographical map at 1 : 25 000 scale in vector format (VECTOR25; Swisstopo, 2004), while the depths for lakes > 0.1 km 2 were 25 obtained from FOEN (2007a). We included all lakes that are contained in the Swiss water bodies information system (GEWISS; FOWG, 2000). Water depth data could not be found for 652 out of 798 lakes. Therefore, we assumed that the depth of lakes < 0.2 km 2 (678 lakes) is less than 30 m. Lake altitude was taken from the digital ele-BGD 10,2013 Anthropogenic and natural CH 4 fluxes in Switzerland  (FSO GEOSTAT, 2006). In total, we found that lakes > 0.1 km 2 are emitting 2.1 Gg CH 4 yr −1 , with a 21 % (0.4 Gg CH 4 yr −1 ) share from hydroelectric reservoirs.

Wetlands
Wetlands are the largest natural source of CH 4 globally, where it is produced by mi-5 crobial decomposition of organic material under anoxic conditions. However, wetlands have become rare in Switzerland (0.5 %, 200 km 2 , of the land area today compared to 6 % in 1800; FOEN, 2007b). In our study, we also considered wetland areas of a mixed ecosystem type and hence a ten times larger area (see Table S2). They are classified as wetland on the basis of their high biodiversity, protected by the Swiss legislation on the protection of mires, rather than by their hydrogeological properties that would better reflect their characteristics in terms of CH 4 fluxes. Most information on wetland CH 4 fluxes originate from the arctic, boreal, and tropical zones, and it is not trivial to translate those results to Swiss wetlands. An important complication is the fact that even in moist environments the vegetated surface may act 15 as a net sink for atmospheric CH 4 when water saturation in the soil is limited to deeper layers or when drainage ditches lower the average water table (e.g. Moore and Roulet, 1993). In most cases it can be expected that periods where wetlands are a sink for CH 4 are restricted to a few warm and dry weeks a year, which reduces the overall annual CH 4 emissions from such ecosystems compared to permanently waterlogged 20 wetlands.
In Swiss fens, CH 4 emission rates ranging from 100 to 330 mg CH 4 m −2 d −1 have been reported for the summer months (alpine fen at Göschener Alp, Liebner et al., 2012). Constant emissions between 0.12 and 31 mg CH 4 m −2 d −1 were also found from glacier forefields with calcareous bedrock (Nauer et al., 2012) to the atmosphere is substantially lower than what microorganisms produce (e.g. King et al., 1998). For our spatially-explicit inventory, CH 4 emissions from Swiss wetlands were estimated from the wetland areas in Switzerland and literature-based EFs available for different wetland types as summarized in Table S2. Different types of wetland areas 10 were determined from the national inventories of raised bogs, fens and mires (FOEN, 2008b(FOEN, , 2010 as well as of riparian landscapes (FOEN, 2008a). Additionally, the wetland core and sprawl areas reported in the national ecological network (FOEN, 2011b) were included in the analysis since these contain additional wetlands of regional and local importance. For the emission estimate, polygons were mapped to a 100 m×100 m 15 raster. Grid cells classified by one of the different types of wetlands were subsequently multiplied with the corresponding EF. If a grid cell belonged to more than one wetland type, the one with highest priority was selected. The priority refers to the level of detail of the data set (e.g. specification of different zones within a wetland) and the importance of a wetland type for CH 4 emissions. As a final step, the data were averaged to 20 the 500 m × 500 m grid. In total, Swiss wetlands are estimated to emit approximately 2.3 Gg CH 4 yr −1 .

Wild animals
Red and roe deer, alpine chamois and alpine ibex are the most abundant wild living ruminants in Switzerland. CH 4 emissions from these wild animals were estimated from  (FOEN, 2011a). We multiplied these animal populations by the respective species dependent EF in SAEFL (1996) (see Table S3 spatial distribution depends on the habitat of the animals. While red deer prefer dense and open forest, row deer prefer dense forest. Alpine chamois prefer unproductive vegetation as well as rocks and scree, while Alpine ibex mainly thrive on rocks and scree. The respective land cover types were selected from the Swiss land-use statistic (FSO GEOSTAT, 2009). Additionally, the locations of these alpine habitats were restricted to 5 altitudes above 1500 m a.s.l. (FSO GEOSTAT, 2006). The number of large wild animals (260 000 red and roe dear, alpine chamois and alpine ibex) in Switzerland (FOEN, 2011a) is substantially less than the 1 580 000 cattle (FSO, 2011) in the agricultural sector. Moreover, wild animals are smaller in size and show a smaller energy uptake than cattle. This results in a comparatively low emission

Agricultural soils
Two counteracting processes -methanogenesis and methanotrophy -drive the net exchange of CH 4 between agricultural soils and the atmosphere. In Switzerland, the agricultural sector comprises typical crop production on arable land (18 %) and comparatively large areas of grasslands (49 %) and alpine summer pastures (33 %), adding 25 up to 15 000 km 2 , corresponding to more than one third of the total area of Switzerland (FSO GEOSTAT, 2009;FSO, 2011). Several studies conducted at managed grasslands and alpine pastures in Switzerland have reported small CH 4 uptake rates by soils (Hartmann et al., 2010;Stiehl-Braun et al., 2011;Imer et al., 2013;Merbold et al., 2013), which is also supported by other studies (Mosier et al., 1991;Flessa et al., 1998;Ineson et al., 1998;van den Pol-van Dasselaar et al., 1999;Kammann et al., 2001). The CH 4 fluxes depend on multiple drivers such as water filled pore space, soil and air temper- . To the best of our knowledge, these are the only two year-round data sets that exist in Switzerland, leading to large uncertainties when up-scaling to the total area of managed agroecosystems. We expect the annual net CH 4 flux for Switzerland to range 15 between 0 and −1.5 Gg CH 4 yr −1 (EF: −0.14 ± 0.14 mg CH 4 m −2 d −1 ; Freibauer, 2003) for the 10 500 km 2 agricultural land excluding Alpine pastures, being a small sink. This is comparable in magnitude to other natural fluxes, but does not significantly contribute to the total CH 4 budget of Switzerland. Since CH 4 uptake across the agricultural areas is highly spatially variable, we did not attempt to spatially distribute this small CH 4 sink 20 across Switzerland in our study.

Forest soils
The net CH 4 flux of forest soils is again dominated by the two counteracting processes, methanogenesis and methanotrophy. The available literature suggests that forests soils generally are a larger CH 4 sink than agricultural soils due to higher soil gas diffusivity 25 in these systems (Smith et al., 2000). were therefore not considered in our spatially-explicit inventory.
To estimate the CH 4 uptake by Swiss forests, we followed a method developed by Hobi et al. (2011). Forest cover was derived from the land-use statistics (FSO GEO-STAT, 2009), and forest type information was taken from the 25 m × 25 m forest mixture data set (FSO GEOSTAT, 2004) and thereafter aggregated to 100 m × 100 m. CH 4 uptake rates differ significantly between evergreen (−0.46±0.27 mg CH 4 m −2 d −1 ) and deciduous forest soils (−1.12 ± 68 mg CH 4 m −2 d −1 ), according to the literature reviewed by Hobi et al. (2011). Forest areas were therefore multiplied with the uptake rate appropriate for the type of forest at a 1 ha resolution. For mixed forests an average rate was used. Finally, the data were averaged to a 500 m × 500 m grid. Overall, our estimate of 15 CH 4 net flux of forest soils is −2.8 Gg CH 4 yr −1 for 2011.

Spatially-explicit CH 4 inventory
Anthropogenic emissions are strongly dominating total Swiss CH 4 emissions and mostly originate from agriculture (see Table 2). Hence, the highest emissions are ob-20 served in the southern part of the Swiss Plateau, an area dominated by livestock farming between the pre-Alps to the south and the Jura mountains to the north, covering approx. 30 % of Switzerland ( Figs. 1 and 2a). Due to the proximity to the Alps, this region receives more precipitation than the rest of the Swiss Plateau and is therefore less suited for production of vegetable and cereal, which are mainly cultivated in the the Swiss Plateau are densely populated and consequently less land is dedicated to agriculture, which corresponds to relatively low emissions in this region. In the Alps, agricultural activity is concentrated on the valley floors. During the summer months, part of the livestock is moved to Alpine pastures for grazing to save the resources in the valley for the winter. This practice is part of the traditional Swiss three-stage farm- 5 ing system (Bätzing, 2003) and therefore CH 4 emissions can also be found in relatively remote areas of the Alps. CH 4 emissions from waste management (Fig. 2b) are more abundant in regions with high population density. This also applies to the energy sector (Fig. 2c), where highest CH 4 emissions occur in urban areas because natural gas is distributed to private 10 households for cooking and heating.
Natural and semi-natural CH 4 emissions from lakes, wetlands, and wild animals ( Fig. 2d-f) as well as the uptake by forest soils (Fig. 2f) are considerably lower than anthropogenic emissions (Table 2). Natural lakes in Switzerland are remnants from previous glaciations. The largest lakes are located in the lowlands, while many small 15 lakes are found throughout the country. Reservoirs are mainly situated in Alpine areas to exploit the descent for hydropower generation (Fig. 2f). Since the large wetlands in the floodplains were drained for agricultural use in the 19th and 20th centuries, highest emissions from this ecosystem type are limited today to shore areas and to hilly landscapes where agriculture is less favorable (Fig. 2d). Wild animals are more abundant 20 in rural areas with continuous forests, the preferred habitat for many species. Alpine ibex and chamois also populate remote and sparsely vegetated mountainous areas (Fig. 2e). Forests cover mountain slopes up to the timberline, protecting from natural hazards. At lower elevations, forests were often converted into agricultural land during the last centuries, but some remained and are protected today by law. While wild ani-25 mals living in the forest are a source of CH 4 , the forest soil acts as a sink. Deciduous forests are limited to lower elevations whereas evergreen forests dominate in higher elevations. Due to the lower uptake rate of evergreen forests, CH 4 uptake by forest soil tends to decrease with elevation.

Comparison with other inventories
The EDGAR v4.2 inventory and the TNO/MACC inventory in Fig. 3 show spatial distributions of total anthropogenic CH 4 emissions over Switzerland for the latest available years of 2008 and 2009, respectively. These maps can be qualitatively compared with the total emissions of our inventory presented in Fig. 1 (FOEN, 2013). The difference between EDGAR and TNO/MACC likely reflects the fact that EDGAR is an independent inventory applying its own methodologies for the collection of activity data, application of emission factors, and spatial allocation. The TNO/MACC inventory, in contrast, is scaled to total emissions reported by the individual countries. 15 In both inventories, the spatial allocation of the emissions is based on different and less detailed geostatistical information than available in our study. Emissions in the EDGAR inventory are higher in densely populated regions but lower in agriculturallydominated regions compared to our inventory (Fig. 3c), suggesting that the EDGAR inventory is too dependent on population density. Differences are less pronounced be-20 tween the TNO/MACC inventory and our inventory (Fig. 3d). The TNO/MACC inventory correctly identifies the regions of farming in the southern parts of the Swiss Plateau but the emissions tend to be higher in these areas and lower in the mountains compared to our inventory. The spatial differences are further assessed in Sect. 3.3.1 to obtain a rough estimate of the uncertainty associated with the spatial disaggregation. 25 For natural CH 4 emissions, we only compare our country totals with numbers reported in an earlier study for Switzerland (SAEFL, 1996) SAEFL (1996), but equates to 3 % of the total CH 4 emissions in Switzerland.

Uncertainties of the inventory
For many purposes, and in particular for inverse modeling studies in which emission inventories are used as a priori estimates, it is important to quantify not only the distribution of the emissions but also its uncertainty. In the SGHGI (FOEN, 2013), an uncertainty is determined for each emission category based on errors associated with 20 the activity data and the EFs. The combined uncertainties are listed in Table 2 together with the mean emissions for 2011. The uncertainty of the annual total emissions can then be computed as the square root of the sum of squares of the individual uncertainties, assuming uncorrelated errors. The uncertainty of the total Swiss anthropogenic CH 4 emissions is estimated to only 16 % (see Table 2), which is largely due to the low Fermentation. It is interesting to note that this uncertainty is smaller than the difference between the SGHGI and the EDGAR v4.2 inventory. For the uncertainty of emissions of a given grid cell at a given time, additional errors need to be considered, including errors associated with the spatial disaggregation and with the temporal variability as described in the following.

Spatial uncertainty
Uncertainties associated with the spatial disaggregation are difficult to assess. They depend on the accuracy of the spatial data sets, on quantization errors due to the use of discrete classes, and on the often crude assumptions made for spatial disaggregation. Therefore, we adopted a pragmatic approach by comparing the spatial distribution in 10 our inventory with those of the EDGAR v4.2 and TNO/MACC inventories shown in Fig. 3 (see also Table 1). All inventories were first linearly scaled to the same country total. To determine a representative error correlation length scale, we analyzed the variogram of the residuals R, where R = E (this study) −E (REF), the difference between emissions in our spatially-explicit inventory E (this study) and emissions in a reference 15 inventory E (REF) (EDGAR or TNO/MACC) (Fig. 4). A variogram describes the variance of the difference of a spatial variable R, i.e. var(R(x) − R(x + h)) as a function of the distance h (Cressie, 1993). The standard deviation of the residuals is larger for EDGAR than for TNO/MACC, which also results in a higher sill (see Fig. 4). As described in Lin and Gerbig (2005), a correlation length scale L can then be derived by fitting an 20 exponential variogram model to the raw variogram. The length scale L obtained in this way was 13.0 km for EDGAR v4.2 and 8.0 km for TNO/MACC, which is close to the grid sizes of the two inventories, suggesting that their limited resolution is a constraining factor and that the true correlation length may be even smaller.
For each grid cell, the uncertainty was assumed to be a fraction f of the absolute 25 emission in that cell. The fraction f was then determined based on the requirement that the uncertainties add up, following Gaussian error propagation, to the 16 % uncertainty of the total country emissions. Thereby, the correlation length scale L was 15203 Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | used to compute the error covariances. Using this approach, the uncertainty of the individual 500 m × 500 m grid cells in our inventory was estimated as 110 % (mean of 85 % and 130 % determined for the two different length scales derived for EDGAR and TNO/MACC). Emissions in individual grid cells thus have a large uncertainty and could well be double or half as large as estimated.

Temporal variability
Our spatially-explicit inventory only includes annual mean emissions, but no seasonal and diurnal cycles due to a lack of suitable data. Nevertheless, we will briefly discuss the available temporal information relevant to our inventory to estimate the importance of temporal variability.

10
In the agricultural sector, livestock numbers are reported once a year in April, and seasonal fluctuations are only on the order of ±3 %, with census data slightly above the annual mean (Bretscher, 2010). Within the traditional Swiss three-stage farming system, cattle is moved to Alpine meadows in summer to save the fertile valley floor for crop and winter fodder production. Hence, the spatial allocation of CH 4 emissions from 15 ruminants changes between summer and winter. In addition, CH 4 emissions from ruminants depend on animal metabolism; thus, emissions peak following feed intake with a delay of a few hours and therefore display a diurnal pattern. Kinsman et al. (1995), for example, reported an average 20 % higher emission during the day than during the night, while Gao et al. (2011) observed emissions peaks following the feeding rhythm. 20 Emissions from manure are lowest at low temperatures and increase with longer storage duration, peaking only after about two months (Hindrichsen et al., 2005(Hindrichsen et al., , 2006Klevenhusen et al., 2010). The storage period before application is typically longer in winter, but lower storage temperatures likely dominate the influence on CH 4 production. Manure storage practice further influences CH 4 emissions (Külling et al., 2001(Külling et al., , 25 2002(Külling et al., , 2003. Higher than average emissions from farmyard manure are compensated by lower emissions from urine-rich slurry, and on average do not differ significantly compared to complete slurry. Hence, these differences are of no relevance compared to the effects of storage period and temperature. Landfills in Switzerland are covered by a soil layer; therefore, no large seasonal temperature fluctuations are expected within the deposited waste. However, the cover soil, where CH 4 is oxidized by methanotrophs, undergoes seasonal temperature fluc-5 tuations. More importantly, moisture positively influences CH 4 production in the waste body and inhibits CH 4 uptake in the cover soil (Chanton and Liptay, 2000). Both factors are expected to lead to higher CH 4 emissions in winter than in summer (Klusman and Dick, 2000). Nevertheless, the available information is insufficient to quantify seasonal fluctuations of CH 4 emissions from Swiss landfills in general. 10 A large proportion of natural gas is used for heating and therefore consumption is more than four times higher in January than in July (VSG, 2012). CH 4 emissions from leaking pipelines are therefore expected to be higher in winter than in summer, but no reliable data are available for proposing a seasonal cycle of these emissions.
Lake CH 4 emissions may exhibit a strong seasonal cycle similar to that found in 15 a Swiss reservoir where emissions were positively correlated with water temperature (DelSontro et al., 2010). In addition, turnover of a seasonally stratified water column can also significantly contribute to the annual CH 4 emissions from lakes (Schubert et al., 2010), a quite common process in Swiss lakes. Thus, both processes lead to a pronounced seasonal cycle in CH 4 emissions from water bodies in temperate zones 20 with higher emissions during summer than during winter (Sect. 2.2.1). The above listed diurnal and seasonal cycles indicate that observed CH 4 fluxes on a single day at a given time may significantly differ from the annual mean fluxes reported in our spatially explicit inventory, but at present the available information is not sufficient to provide source-category specific time functions.

Conclusions
A spatially-explicit high resolution CH 4 inventory was developed for Switzerland for the year 2011. This is the first comprehensive inventory at national level synthesizing most of the available Swiss datasets on anthropogenic as well as on natural and semi-natural fluxes. Anthropogenic emissions of 177 Gg yr −1 in 2011 are by far larger 5 than the emissions from all natural and semi-natural sources, which were estimated to only 5.7 Gg yr −1 , an order of magnitude less than an estimate reported in an earlier study on natural sources in Switzerland (SAEFL, 1996). Forest soils are estimated to be a net sink with a net flux of −2.8 Gg yr −1 and agricultural soils are estimated to be CH 4 neutral or a small sink with a net flux between −1.5 and 0 Gg yr −1 for agricultural 10 soils, partially offsetting the natural emissions. In total, Switzerland acted as a net CH 4 source of 180 Gg CH 4 yr −1 in 2011. With a share of nearly 85 %, agricultural emissions are by far the most important anthropogenic source in Switzerland, followed by the waste and energy sectors. The uncertainty of the total anthropogenic emissions is estimated to be only 16 %, which is largely a result of the low uncertainty assigned to 15 the largest single CH 4 source -enteric fermentation of ruminants. Detailed geospatial information is available for Switzerland, thereby allowing the spatial allocation of the individual emission sources. Information on temporal variability of CH 4 emissions, however, is very sparse and currently insufficient for prescribing diurnal and seasonal variations, an aspect that should be better addressed in future studies. 20 This inventory will provide invaluable input for regional-scale atmospheric modelling and inverse source estimation, which are urgently needed for independent validation of inventories based on atmospheric measurements. The spatial disaggregation of other CH 4 sources currently not covered by this inventory, especially from biogas production and composting, might become more critical in the future with the expected increase  München, 431 pp., 2003. Beauchemin, K. A., Kreuzer, M., O'Mara, F., and McAllister, T. A.: Nutritional management for enteric methane abatement: a review, Aust. J. Exp. Agr., 48, 21-27, 2008. Bergamaschi, P., Krol, M., Dentener, F., Vermeulen, A., Meinhardt, F., Graul, R., Ramonet, M., 5 Peters, W., and Dlugokencky, E. J.: Inverse modelling of national and European CH 4 emissions using the atmospheric zoom model TM5, Atmos. Chem. Phys., 5, 2431-2460, doi:10.5194/acp-5-2431 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Choularton, T. W., Gallagher, M. W., Bower, K. N., Fowler, D., Zahniser, M., Kaye, A., Monteith, J. L., and Harding, R. J.: Trace gas flux measurements at the landscape scale using boundary-layer budgets, Philos. T. Roy. Soc. A, 351, 357-369, doi:10.1098/rsta.1995.0039, 1995 Friedrich, R., Rivier, L., Houwelling, S., and Schulze, E. D.: The European carbon balance. Part 1: fossil fuel emissions, Glob. Change Biol., 16, 1395-1408, doi:10.1111/j.1365-2486.02098.x, 2010. Cressie, N. A. C.: Statistics for Spatial Data, John Wiley and Sons, New York, 900 pp., 1993 Agr. Forest Meteorol., 150, 841-851, doi:10.1016/j.agrformet.2009.12.008, 2010. Farquhar, G. J. and Rovers, F. A.: Gas production during refuse decomposition, Water Air Soil Poll., 2, 483-495, doi:10.1007/BF00585092, 1973 1970-2008 2000-2005 2003-2007 and 2009 1997, 2000, 2001 and 2003   The provided uncertainty estimates follow the Tier 1 methodology (IPCC, 2000), represent half of the 95 % confidence interval expressed in percent (IPCC, 1997), and accounts for uncertainties in emission factors (EFs) and activity data for the individual level or category. Total national emissions exclude the Land Use, Land-Use Change and Forestry sector (LULUCF in italics) as well as International Bunkers (not shown) in accordance with the reporting requirements under the UNFCCC. Methods applied and EFs used are indicated (D = IPCC Default, T1 = IPCC Tier 1, T2 = IPCC Tier 2, T3 = IPCC Tier 3, CR = CORINAIR, CS = Country-specific). All data for the anthropogenic sources are taken from the national Greenhouse Gas Inventory (FOEN, 2013), while CH 4 fluxes from the natural categories base on estimates presented in this study. The categories indicated with an asterisk are included in our spatially explicit inventory.  4. Semivariogram of the differences between the EDGAR v4.2 and TNO/MACC inventories and our inventory. Also shown are exponential fits to the data (see text for further details).