Description of the methodology for spatial distribution of diffuse emissions from the agriculture sector

Description of the methodology for spatial distribution of diffuse emissions from the agriculture sector

Sonia Orlikova1, Jochen Theloke1, Balendra Thiruchittampalam1, Melinda Uzbasich1, Thomas Gauger2

1) Institute for Energy Economics and the Rational Use of Energy, Universität Stuttgart (IER)

2) Institute of Navigation, Universität Stuttgart (INS)

Title / Diffuse Air emission in E-PRTR
Customer / European Commission
Customer reference / 070307/2009/548773/SER/C4
Confidentiality, copyright and reproduction / Unrestricted

Content

1Gridding methodology for the agriculture sector

1.1Sector description

1.2Emission data input

1.3Applied methodology

1.4Emissions and proxy data sets

1.5Existing gridding methods and applied proxy data sets

1.6Applied models

1.7Conclusion and result

1.8Reference

1Gridding methodology for the agriculture sector

1.1Sector description

Sector definition

This sector include emission caused by animal husbandry and manure management (NFR 4B)andby crop production and agricultural soils (NFR 4D). For the purpose of this study, the main pollution sources are animal waste digestion, fertilization, livestock farms and field operation.

A partial overlap with E-PRTR Annex 1 activity 7 can take place in some countries. In such cases the subtraction method detailed in document (Subtraction Methodology) is applied.

Pollutants covered

  • The considered pollutants for this sector are: NH3 released by:
  • animal waste digestion (grazing animals)
  • N-fertilizer application in crop production.
  • PM10 released by:
  • livestock facilities
  • field operation.
  • Emission data input

The input emission data used for the spatial distribution of agriculture sector are provided by following dataset:

  • National emissions reported to “Convention on Long-Range Transboundary Air Pollution (CLRTAP)”[1]

There are no additional preparation processes regarding the diffuse emission releases from agriculture sources, except of the subtraction process regarding the point sources of swine (NFR-4B8) and poultry (NFR-4B9). Details on the subtraction method are given in the document Subtraction Methodology.

1.3Applied methodology

The agriculture emissions are assumed to be highly associated with the density of agricultural land cover. The distribution of animal census is in relation to the land cover and therefore the emissions from agriculture are spatially disaggregated according to the distribution of agricultural land use categories. The emissions of NH3 and PM10 by crop production and agricultural soil are a function of fertilizer–N and of handling and transport of agriculture product and are also spatially disaggregated according to the distribution of agricultural land areas.

The general method of emissions distribution from this sector based on the following formula[EEA 2009]:

where:

  • i: is a specific geographic feature;

here: e.g. a 5km x 5km grid, or administrative unit at NUTS 2 or NUTS 3 level

  • emission i,x: is the emissions attributed to a specific geographical feature i (e.g. a grid cell or administrative unit) within the spatial surrogate dataset x;

here: emission of agricultural activities in determined administrative unit or grid cell (5km x 5km) using the animal numbers as surrogate dataset x

  • emission t: is the total national emission for a sector t to be distributed across the national area using the (x) surrogate spatial dataset;

here:the national total emission from agriculture for each pollutant; reported by CLRTAP

  • value i,x: are the surrogate data value of each of the specific geographical features within the spatial surrogate dataset x;

here: the surrogate data are for example the animal number within administrative unit and animal number at national level.

With this formula, it is possible to allocate the diffuse sources using proxy data (animal number) into administrative units. This so-called Top-Down Method, starting with emission at national level (the top) and works downward by decomposing the emission into 5 km x 5 km grid cell(the bottom) is used.

Once the allocation for animal husbandry is done, the re-distributing of emissions for each pollutant and each animal species to a higher resolution (e. g. 5 km x 5 km grid cell) is implemented by using the CORINE Land Cover data set. The most probable location within the administrative unit for the agriculture activity has to be defined upfront.

The methodology of the spatial distribution of the diffuse part of agriculture emissions caused by animal husbandry and manure management (NFR 4B)contains therefore two main steps:

  1. Regionalization of national emission releases: calculation of the emission values based on statistical proxy data for each administrative area (e.g. NUTS 3); i.e. distribution of the national total for each pollutant onto each region
  2. Gridding: spatial distribution of the regionalized emission values on grid cell level based on different proxy data

These two main steps are divided into following sub-steps:

  1. Regionalization of national emission releases
  • This activity aims at allocating the share of the national total for each pollutant to each NUTS 3 levelusing the following underlying parameters and steps:
  • Animal number (NUTS 2,NUTS 3level);
  • Regionalization of the diffuse releases from industrial activities on NUTS 3 level based on the number of animal numbers
  1. Gridding
  • This activity aims at allocating the emission releases NUTS 3 values to each polygon/grid cell according to the defined 5km x5km grid cell resolution, using the following underlying parameters and steps:
  • Land cover types (CORINE 2000/2006)
  • Calculation of land use area (GIS - Area calculation)
  • Calculation of the share of different land cover types in each grid cell/polygon (e.g. "pastures" - CORINE 2000/2006);
  • Calculation of emission values for each grid cell/polygon based on the combination of the applied proxy data.

The emission caused by crop production and agricultural soils (NFR 4D sector) is allocated directly into 5km x 5km grid cell, through specific land use categories.

The Corine Land Cover data set offers 44 differentland use categories. For the agriculture sector the following categories were taken into account:

- class 2: Agricultural areas, except the sub-classes 2.1.3. (rice fields) and

- class 3: Forests and semi-natural areas, except the sub-classes 3.1. (forest)

For instance, to distribute the emissions for cow and sheep livestock, the distribution of grazing land is used. For distributing pigs and poultry numbers, the distribution of all non-urban land is used (Dragosits, 1998). Therefore it is necessary to calculate the amount of agricultural land, including pastures, natural grassland, non-irrigated arable land etc. within the administrative units, categorized on NUTS 2 or NUTS 3-Level.

1.4Emissions and proxy data sets

Table 1 shows the European emissions for both agricultural sectors (NFR 4Band NFR 4BD) officially reported by the countries to CLRTAP for the year 2007. The amount of all emissions from releases to air in the EU27 + EFTA countries is also presented.

Table 1: Emissions reported for the agricultural sector officially reported by the EU27 + EFTA countries

Sector / Sector_Code
[NFR_08[2]/CRF[3]] aggregated / Sector_Name
[NFR08/CRF] aggregated / CLRTAP Emissions in 2007 [kt]
NH3 [kt] / PM10 [kt]
Manure Management and animal husbandry / 4B1a / Cattle Dairy / 737 / 7
4B1b / Cattle Non-Dairy / 779 / 8
4B2 / Buffalo / 17 / 0
4B3 / Sheep / 77 / 1
4B4 / Goats / 10 / 0
4B6 / Horses / 49 / 1
4B7 / Mules and Asses / 2 / 0
4B8 / Swine / 623 / 44
4B9a / Laying Hens / 140 / 16
4B9b / Broilers / 90 / 14
4B9c / Turkeys / 15 / 0
4B9d / Other Poultry / 183 / 27
4B13 / Other / 47 / 0
Crop production and agricultural soils / 4D1a / Synthetic N-fertilizers / 878 / 3
4D2a / Farm-level agricultural operations incl. storage, handling and transport of agr. products / 0 / 122
4D2b / Off-farm storage, handling and transport of bulk and agricultural products / 0 / 0
4D2c / N-excretion on pasture range and paddock Unspecified / 24 / 5
Sum / 3670 / 247
All emission releases to air in EU 27 and EFTA countries / 3936 / 1915

As mentioned above, the spatial distribution and the obtaining of grid cells for each pollutant are based on the emissions presented in 2 and the following proxy data sets:

  • animal numbers
  • land use data

Table 2 lists the main proxy variables used for spatial distribution of the diffuse part of agricultural sector.

Table 2: Proxy data sets used for the spatial distribution of the emissions from agricultural sector

Sector / Proxy Dataset / Data Source / Release Year / Extend
Manure Management and animal husbandry / animal density / EUROSTAT / 2008 / EU 27 +EFTA
Land use / Corine land cover (CLC90) Switzerland, CLC 2000, CLC 2006 / 1990, 2000, 2006 / EU 27 + EFTA
Crop production and agricultural soils / Land use / Corine land cover (CLC90) Switzerland, CLC 2000, CLC 2006 / 1990, 2000, 2006 / EU 27 + EFTA

1.5Existing gridding methods and applied proxy data sets

The main selection criteria for the chosen proxy data for the spatial distribution agriculture sectors wereavailability and coherence of data.To date, there are different gridding methodologies using different distribution parameters. The main methodologies are described and compared below.

UK approach:

Emissions of PM10 from agricultural livestock and poultry sources were distributed at 1 km resolution using a combination of animal census data and land use. For England very detailed farm level data was obtained, for Scotland, Wales and Northern Ireland agricultural census data were only available for larger spatial units and therefore the land use data were used to generate a distribution of emissions within these spatial units. The resulting distributions for England, Scotland Wales and Northern Ireland were combined, weighted according to the relevant regional statistics on the number of livestock or poultry in these regions[NAEI Emission Mapping Methodology 2006]. The emission distribution using this method can leads to inconsistency.

The distributions of NH3, from agricultural sources have been mapped at a 5km resolution. Data from the agricultural census for all UK were combined with emission factors for livestock and fertilizer use and LandCover Map 2000 data within the Atmospheric Emissions for National Environmental Impacts Determinationmodel (AENEID) to calculate emissions maps.

To map other agricultural emission of PM10 and NH3, the Land Cover Map 2000 data were used and the emissions were distributed evenly across the arable land cover map for the UK.

NL approach:

To allocate an emission spatially, the Emission Register has a spatial allocation available for each emission source.

The PM10and NH3allocation to grid cell is based on numbers of livestock and land use data using the GIAB database (geographical information system which records the location of every farm). Various farm data are linked to these locations, originating from the agricultural survey carried out by the Ministry of Agriculture, Nature and Food Quality, and livestock data from the data files of the Animal Health Service. Besides numbers of livestock, the emission factors for PM10and NH3per animal category are also used. These emission factors are in accordance with the factors used by the Netherlands Environmental Assessment Agency (PBL), in the Netherlands Pollutant Release & Transfer Register [PRTR Netherlands, 2010].

Additionally, the protection of privacy of individual farms has been taken intoaccount. This was done by not separately presenting the emissions in cells with fewer than five farms, but by distributing the emissions equally across all these cells.

The following table shows in short form the main characteristics of the presented gridding methodologies.

Table 3: Availablemethodologies for the spatial distribution of diffuse industrial emissions

Methodology / Advantage / Disadvantage / Uncertainties
NAEI Emission Mapping Methodology 2006 (UK) / PM10 emission distribution at high resolution (1km x 1km grid cell) based on detailed agriculture census and land use data.
NH3 emission distribution at resolution (5km x 5km) using emission factors for livestock and fertilizer use and Land Cover Map 2000 data within the AENEID Model / No consistency in proxy data on regional level for all relevant countries. / Proxy Data
Spatial allocation of diffuse agricultural emissions (PRTR Netherlands - The Emission Register project, 2004) / PM10 and NH3 distribution at high resolution (1km x 1km and 5km x 5 km) based on farm data linked to specific locations, emission factors for PM10 per animal category and land use data / No consistency in proxy data on regional level for all relevant countries. / Proxy Data
E-PRTR Methodology / Available land cover data;
Available animal data
High resolution (5 km x 5 km grid cell) emission distribution maps. / No central database with proxy data for agriculture sector, all pollutants and all countries;
Large amounts of data;
Requires reliable EU-wide datasets / Proxy Data

1.6Applied models

For spatial distribution of the diffuse part for agricultural sector no existing models are to be used.

1.7Conclusion and result

For the purpose of this study, the emission distribution caused by animal husbandry and manure management (NFR 4B) is based on animal census and Corine Land Cover. The major emissions of NH3 and of PM10 from crop production and from agricultural soils (NFR 4D)are generally proportional to the amount of the N-fertiliser applied and to the agricultural activities. In order to gain a consistent result at EU scale, the Corine Land Cover data set and the Global land cover data setare applied for the emission distribution from crop production and agricultural soils.

Figure 1 shows an example of NH3 emissions for Europe distributed for 0.05 degree x 0.05 degree grid including all sectors. The main source for NH3 emission is agricultural sector with about 95% and therefore the emission map is more and less the same across Europe.

Figure 1: Example for gridding of the NH3 emissions for Europe [IER - CARBOEUROPE 2008]

1.8References

Dore C. J., et al (2007): APMOSHERE (Air Pollution Modelling for Support to Policy on Health and Environmental Risk in Europe). ImperialCollege of Science, Technology and Medicine, London.

Bush, T. et al., 2008: NAEI UK Emission Mapping Methodology 2006 [URL: reports.php, 08.07.2010]

DragositsU., et al (1998): Modelling the spatial distribution of agricultural ammonia esmissions in the UK. Environmental Pollution 102, pages 195 -203

EMEP/EEA CLC2000 (Corine Land Cover 2000) 100 m: [URL: 02.07.2010]

EMEP/EEA CLC2006 (Corine Land Cover 2006) Version 02/2010 [URL: 02.07.2010]

EMEP/EEA air pollutant emission inventory guidebook 2009: Chapter 4.B. Manure Management [URL:

PRTR Netherlands 2010 - The Emission Register project: Spatial allocation of emissions [URL: 10.08.2010]

1

[1]CLRTAP_NFR09_GF_v10 (Convention on Long-range Transboundary Air Pollution) [URL: 05.08.2010]

[2]The NFR_08 code (New Reporting Format) is based on the sector systematic which is applied by the countries for the CLRTAP reporting submissions

[3]The CRF code (Common Reporting Format) is based on the sector systematic which is applied by the countries for the UNFCCC reporting submissions