Comparison of CH4 emission inventory data and emission estimates from atmospheric transport models and concentration measurements
Introduction
The Annex 1 countries to the Framework Convention of Climate Change agreed during the Third Conference of Parties (CoP) in 1997 in Kyoto to reduce their aggregate anthropogenic carbon dioxide equivalent emissions of greenhouse gases by 5% in the 2008–2012 period with respect to 1990 levels. Before the meeting in Kyoto, an editorial review in Nature stated: “How can emissions of greenhouse gases be contained within some limit yet to be determined if nobody knows what these emissions are?” (Nature, 1995). Reliable knowledge on emissions at a national level is therefore becoming increasingly important. A common methodology for reporting national emissions and sinks: “The Guidelines for National Greenhouse Gas Inventories”, developed and published by the IPCC IPCC/OECD/IEA, 1995, IPCC/OECD/IEA, 1997, were adopted during the First Conference of Parties in 1995. The Guidelines contain formats and methods to calculate and report the national emissions, supplemented with default emission factors and activity data. Countries are invited to use the IPCC calculation method but are also allowed to use their own emission factors and activity data if available. In 1997, the CoP adopted the `revised IPCC guidelines' for reporting country emissions, the so-called `National Communications' (NCs). Countries submitted their first National Communications in 1994 and a second in 1997. These national emission inventories, kept by the Climate Secretariat in Berlin, are public domain.
The following issues are given below:
- •
Do the reported national emission data indeed reflect the scientifically established atmospheric budget of greenhouse gases on a global, regional and even national scale?
- •
Can the National Communications be monitored and independently verified to receive the credibility needed to ascertain compliance with the Protocol?
In this paper two methods are introduced and discussed to address these questions:
- •
Comparison of NCs with a reference data set such as the EDGAR/GEIA database, both databases (NC and EDGAR) being based mainly on statistical information. This comparison is further elaborated in: Analysis of differences between national inventories and EDGAR by van Amstel et al. (1999)
- •
Comparison of NCs and EDGAR with results of a combination of atmospheric measurements, and global and regional transport models for methane, which is the main subject of this paper.
The Emission Database for Global Atmospheric Research (EDGAR) has been developed within the international scientific IGBP/GEIA program (Olivier et al., 1996). This database, maintained by RIVM and TNO, comprises statistics on energy use, agricultural and industrial activities, demographic data, social and economic factors, land use distributions and data sets of emission factors. Statistical data come from such international databases as IEA, FAO and UN; emission factors are mainly generated in the GEIA/IGAC program. This database contains sectoral emissions of all greenhouse gases on a 1°×1° grid and on a country basis. This special feature of the EDGAR system enables a comparison of aggregated country emission data as from National Communications with the gridded emissions, as used in climate modelling. NCs and EDGAR have in common that they both use statistical data as primary input.
The first signs that human activities could lead to climate change arose from the theoretical reasoning by S. Arrhenius (1896). These were confirmed by measurements of rapidly rising concentrations of greenhouse gases in the atmosphere (Keeling, 1960). Atmospheric transport models are used to investigate the relation between concentrations and emissions by comparing results of measurements and model calculations. In their scientific assessment reports the IPCC established global budgets of greenhouse gases based on an analysis of atmospheric measurements and climate models (IPCC, 1990, IPCC, 1994, IPCC, 1995). In this paper we will present results of several methods to estimate emission inputs based on atmospheric measurements and transport models: (a) forward modelling i.e. emissions are used as input and atmospheric concentrations are calculated and (b) inverse modelling i.e. measured concentrations are used as an input and the emissions are calculated. There are a number of methods to construct an inverse model Enting, 1998, Tarantola, 1987. Generally inverse models require a large amount of computer power because an optimum emission estimate is calculated out of a large number of possible distributions using statistical theory. Inverse models have therefore only recently become available on a broader scale now that computer power has increased.
There are various methods to complete an emission estimate. Bottom-up emission inventories are produced by using process information and applying an emission estimation algorithm (most frequently `activity rate×emission factor'; however, more complicated algorithms are also used for some source categories). The most important characteristic of bottom-up emission inventories is that the figures are based on the upscaling of process information to a higher level of aggregation i.e. to figures which are generally representative for that emission process on a larger (global, regional or national) scale. In upscaling, the `best available' average emission factor appropriate to the available statistics, is used to aggregate on a global, regional or national scale. An example is the aggregation of the CH4 emission from different rice cultivars in various regions of China to the CH4 emission from rice in a specific year. Statistical data can be used for upscaling. Top-down emission estimates are produced by using appropriate proxies to derive higher resolution (in space, time or source category) inventories from aggregated estimates. An example is the disaggregation of the national total emission of pesticides, determined from the sales of the particular pesticide, by using satellite information on crop fields. Another method to complete top-down emission estimates is downscaling the results of a limited number of atmospheric measuring points in a large (continental) area to an emission estimate on a much smaller (national) scale. Atmospheric transport models are used for downscaling. This paper will concentrate on an analysis of emission estimates derived from atmospheric modelling.
All methods have their own uncertainties. Errors are made in up- and downscaling; the availability and accuracy of the data may differ. If only results of atmospheric modelling are used, only total emissions (added natural and anthropogenic emissions summed over a number of source categories) can, in principle, be calculated. Additional information on isotopes, temporal and spatial variations in emission strengths, correlation with other species (fingerprints) and `a priori' emission estimates etc is needed to derive information about the locations and strengths of specific sources or source categories. An accurate estimation of emission inputs from atmospheric measurements and transport models is most of the time severely hampered because of lack of sufficient measurements. The system then has degrees of freedom (the number of variables is larger then the number of equations) and offers several solutions, all of which satisfy the boundary conditions. The estimation is then considered as an `ill-conditioned' or `underdetermined' problem. Next, comparing bottom-up and top-down emission estimates is not trivial. This is due to the different ways in which emissions are averaged in time and space and in which emissions are subdivided into sectors and source categories. The differences between the methods decrease the accuracy of the comparison. Next, some expert judgement is often involved in choosing the `best estimate'. An inventory expert uses various sources of information to select the best emission factor; the atmospheric modeller can use different approaches to adjust the emissions assumed to the available observations. Because of all these degrees of freedom, the question arises as to whether a comparison of emission inventories compiled by different methods can be used to independently verify emission data. We are obviously dealing with a comparison process. At this stage of scientific knowledge we should not consider the results of this cross-examining as a method to arrive at a `true result' but as a way to reduce uncertainties and reinforce the scientific credibility of the various emission estimates.
The way in which emissions are averaged in time and space using the various methods differs greatly, as does the subdivision of emissions into sectors and source categories. To support the development of a comparison procedure we will present here the different levels at which data can be compared and some common formats. This will be followed by the results of the comparisons used to rank the uncertainty of source strengths inputting the range of the emission strengths reported and the location of emission sources as indicators.
At the moment emission inventories based on statistical data such as the NCs and EDGAR, and emission estimates based on atmospheric measurements and models are developed rather independently. The FCCC calls for emission inventories which can be monitored and independently verified. This paper discusses the results of a comparison of different methods to calculate CH4 emissions and the use of comparisons as a tool for establishing compliance with the Kyoto Protocol.
Section snippets
The data sets
The bottom-up inventories used were the National Communications database (van Amstel et al., 1997) and the EDGAR/GEIA global emission database (Olivier et al., 1996). Top-down estimates are based on results of six global atmospheric transport models for CH4 Fung et al., 1991, Taylor et al., 1991, The and Beck, 1995, Hein et al., 1997, Saeki et al., 1997, Lelieveld et al., 1998 and three regional models (Vermeulen et al., 1997a, Vermeulen et al., 1997b, Stijnen et al., 1998, Berdowski et al.,
A comparison of the global totals
Table 4 presents a comprehensive overview of the global CH4 budget as compiled by several authors. Ranges of emission estimates are also included.
The lowest global total CH4 emission was estimated by Saeki et al. (445–463 Tg CH4/yr); the global total estimated by Taylor et al. (1991) (611–623 Tg CH4/yr) is found between the highest emissions. The assumed lifetime of CH4 in the atmosphere is a very important factor in the determination of the global total. The adapted lifetime for CH4, ranging
A comparison of the zonal distribution
The comparison can be expanded in more detail by analyzing the total and sectoral inventory data (EDGAR) and modelled CH4 emissions in zonal bands of 10°. N.B. A zonal band of x° in the text extents from x−5° to x+5°.
A comparison of results of emission inventories and transport models on regional and national scales
If measuring data and models with high spatial and temporal resolution are available a much more detailed analysis will be possible, even on a national scale. The results of a comparison of emission inventories (EDGAR and NCs) and emission inputs of transport models for some countries in NW Europe using high resolution measurements and inverse modelling will be presented in this section.
General conclusions
The conclusions from the comparison of emission inventories and emission estimates derived from atmospheric measurements and transport models of CH4 can be summarized as:
- •
There is a considerable range in the estimates of the global CH4 emission as used by various atmospheric modellers as input in their global transport models.
- •
This range limits the possibilities to accurately and univocaly verify CH4 emission inventories by using atmospheric models and measurements.
- •
Additional model calculations
Unlinked References
Curriculum Vitae's Vitae's 1, 2 and 3 not found in text
Acknowledgements
We are grateful to A.W. Heemink, J.W. Stijnen and X.F. Zhang of the Technical University of Delft and Ben Farnham of Strathclide University for their major contributions to the Kalman filtering. We also gratefully acknowledge the European Consortium for Mathematics in Industry (ECMI) for their initiatives in facilitating the scientific cooperation between the collaborating institutes (Delft University of Technology, Technical University of Eindhoven, University of Strathclyde and RIVM). J.T.
Leon H.J.M. Janssen studied Physics and Mathematics at the University of Utrecht. At the Joint Laboratories of the Dutch Power Companies he did research into the dispersion of air pollutants and the measuring of concentrations of the main greenhouse gases carbon dioxide and methane. Next, as policy maker at the Ministry of Housing, Spatial Planning and the Environment he coordinated the First Netherlands Memorandum on Climate Change which laid out the Dutch climate policy. In 1993 he joined the
References (36)
- et al.
A computationally efficient Kalman smoother for the evaluation of the CH4 budget in Europe
Applied Mathematical Modelling
(1999) - van Amstel, A.R., Kroeze, C., Janssen, L.H.J.M., Olivier, J.G.J., van der Wal, J.T., 1997. Greenhouse Gas Emission...
- van Amstel, A.R., Janssen, L.H.J.M., Olivier, J.G.J., 1999. National greenhouse gas inventory improvement....
On the influence of carbonic acid in the air upon the temperature of the ground
Phil. Mag. S.5
(1896)- Berdowski, J.J.M., Draaijers, G.P.J., Janssen, L.H.J.M., Hollander, J.C.Th., van Loon, M., Roemer, M.G.M., Vermeulen,...
- et al.
Trends 1993
(1993) Green's function methods of tracer inversion
- Farnham, B., 1997. Testing and using the Kalman smoothing technique as a means to identify methane emissions in Europe....
- et al.
Three-dimensional model synthesis of the global methane cycle
J. Geophys. Res
(1991)
An inverse modeling approach to investigate the global atmospheric methane cycle
Global Biogeochemical Cycles
Estimation of methane emissions from rice fields using inverse modelling
IPCC Guidelines for National Inventories of Sources and Sinks of Greenhouse Gases
Cited by (5)
Regionalised inventory of biogenic greenhouse gas emissions from European agriculture
2003, European Journal of AgronomyEstimation of potential methane production through the mass balance equations from agricultural biomass in Korea
2016, Applied Biological ChemistryEstimation of atmospheric methane emissions between 1996 and 2001 using a three-dimensional global chemical transport model
2006, Journal of Geophysical Research AtmospheresAtmospheric modeling of high- and low-frequency methane observations: Importance of interannually varying transport
2005, Journal of Geophysical Research D: Atmospheres
Leon H.J.M. Janssen studied Physics and Mathematics at the University of Utrecht. At the Joint Laboratories of the Dutch Power Companies he did research into the dispersion of air pollutants and the measuring of concentrations of the main greenhouse gases carbon dioxide and methane. Next, as policy maker at the Ministry of Housing, Spatial Planning and the Environment he coordinated the First Netherlands Memorandum on Climate Change which laid out the Dutch climate policy. In 1993 he joined the National Institute of Public Health and the Environment where he worked on the use of Kalmanfiltering as a data-assimilation method to evaluate the methane budget of Western Europe and the Netherlands using high resolution concentration measurements and dispersion models. He was co-organizer of the IPCC Expert Meeting on Methods for the Assessment of Greenhouse Gas Inventory Quality in November 1997.
Jos Olivier trained at the Free University in Amsterdam in physics. He works at the Netherlands National Institute of Public Health and the Environment (RIVM) since 1990 as senior scientist and co-ordinator of international emission inventories, with a special interest in energy-related topics. Currently, he is co-convenor of the Global Emission Inventory Activity (GEIA) of IGAC/IGBP and also participates in Expert Groups on Fuel Combustion and Industrial Processes of the National Greenhouse Gas Inventory Programme of the IPCC.
Andre van Amstel is trained as a physical geographer at the University of Amsterdam. He worked at the National Institute of Public Health and the Environment from 1991–1996 where he was responsible for the national greenhouse gas emission inventories for the first National Communication from the Netherlands. Since 1997 he is working at the Wageningen Agricultural University, at the Environmental Systems Analysis Group. His main interest now is greenhouse gas emissions from agriculture.