Emission estimation following the IPCC guidelines

The amount of a gas released follows from four quantities: the burnt area, the fuel mass per unit area, the share of it that actually burns (the combustion factor), and the emission factor of the gas concerned in grams per kilogram of dry matter.

The area comes from the EFFIS polygons, the vegetation type from their land cover shares, the factors from the tables of the 2006 IPCC guidelines. The 2019 refinement marks the tables for fuel consumption and combustion factors as updated; the values relevant to the Balkans remained unchanged, and the emission factor table was not touched.

Combustion factors from table 2.6: 0.45 ± 0.16 for coniferous, broadleaved and mixed forest; 0.72 ± 0.25 for Mediterranean shrub and transitional vegetation; 0.77 ± 0.26 for grassland. For crop residues the table carries no factor of its own, samoEKO substitutes the grassland value there and states that it does so.

The carbon fraction from table 4.3 is 0.47 by default, 0.48 for broadleaved and 0.51 for coniferous trees; carbon is converted to carbon dioxide with the factor 44/12.

For the carbon dioxide equivalent the hundred-year values of the sixth assessment report apply: 27.0 for biogenic methane and 273 for nitrous oxide.

Four points at which something is computed rather than measured, they appear in every method line that is output.

First, the IPCC tables know no category “Mediterranean”. For pine forest on the Adriatic one must choose between “other temperate forests” (0.45), boreal crown fire (0.43) and surface fire (0.15). samoEKO consistently takes the first value, because the table's footnote explicitly assigns the remaining forest types to that category. That is a model assumption, not a measurement.

Second, table 2.4 already gives the product of fuel mass and combustion factor, not the mass alone. Anyone taking those values and multiplying again by the combustion factor computes more than half too low.

Third, two routes to carbon dioxide are mutually exclusive. The emission factor of 1,569 grams per kilogram corresponds arithmetically to a carbon fraction of about 0.428, because part of the carbon escapes as carbon monoxide, methane and soot. Adding the emission factor route and the carbon route counts the same carbon twice.

Fourth, at this tier the calculation can overestimate in the fire year, because carbon in biomass that is killed but not burnt counts as released immediately. Non-methane volatile organic compounds are left out entirely because of excessive uncertainty, as the IPCC guidelines themselves prescribe.

The uncertainties of the input quantities are known and large: roughly 36 per cent on the combustion factor in temperate forest, roughly 35 per cent in shrubland, roughly 8 per cent on the carbon dioxide emission factor, 30 to 50 per cent on biomass and 20 to 30 per cent on burnt area from satellite detection.

Propagated in quadrature this yields 55 to 70 per cent overall uncertainty on the carbon dioxide figure a factor of 1.6 to 2.

From this follows the output convention: samoEKO always states carbon dioxide as a range, rounded to two significant figures. Every function of the balance returns a value, a lower bound, an upper bound, a unit and a method, and never a bare number. For nitrous oxide the range is wider still: 27 per cent from the emission factor alone, multiplied by a global warming potential of 273. Anyone publishing a carbon dioxide equivalent must publish the range with it.

Above-ground biomass comes from the ESA CCI Biomass dataset at one hundred metres resolution, with its uncertainty carried alongside.

Its limit must be stated: this value contains neither litter nor dead wood and is therefore systematically lower than the fuel mass of the IPCC definition, which includes both pools. In a tier 1 calculation without land use change the IPCC sets those pools to zero anyway, which is consistent, but has to appear in the output, otherwise the emission is underestimated and nobody notices.