Component 4: Integration of Scales and Carbon Data Assimilation Methods (”Continental Scale Integration”)

Component Leader: M. Heimann, MPI-BGC


To assess the present European carbon balance , its component processes and its variability on a continuum of spatial scales going from local (10 km) to continental (5000 km), by combining the data streams of flux measurements, concentration measurements, forest and soil carbon inventories and merge them with additional information from remote sensing, process understanding and modelling into a continental Carbon Data Assimilation System (CDAS).


4.1 To develop advanced modelling tools for estimating the spatially explicit continental carbon balance and its variability at a resolution of 10 to 50 km for at least the length of a Commitment Period (MO1, 2, 4, 7, 8).

4.2 To test a multiple constraint approach by applying top down and bottom up methodologies to achieve the best possible estimate of the European carbon balance and to determine the variation in biospheric and anthropogenic fluxes over Europe (MO2).


The primary data streams generated in CarboEurope cover very different spatio-temporal domains. These different scales of the observational data streams have to be bridged by means of a numerical modelling framework. Remote sensing data, which do cover the scales from the individual plot to the entire continent and from weeks to several years, do not record directly the carbon balance, but only certain features of the vegetation, such as the fraction of absorbed photosynthetically active radiation, albedo and, with limitations, aboveground biomass. However remote sensing information can be used to drive numerical models and thus help bridge the scales between the different observational data streams. Using this approach, several independent bottom-up and top-down methods will be applied and compared against each other for consistency and finally merged into a European scale Carbon Data Assimilation System to determine the European carbon balance over the past and present decade. In order to cover the past, present and future evolution of the European carbon balance on longer time scales (100-200 years), however, requires the use of prognostic numerical terrestrial ecosystem models, which have to be tested against the rich observation data sets compiled in CarboEurope.

The common denominator spatial scale will be the ”Eurogrid”, i.e. a spatial scale with grid elements of 20-50 (100) km (to be decided), on which the European carbon balance can be estimated independently by the different methods. This is a grid size that an atmospheric mesoscale model, covering the whole continent and nested in a global weather/climate model, can handle without taking recourse to sophisticated downscaling techniques. Hence terrestrial ecosystem carbon models (TEMs) developed for this grid resolution (E.g. LPJ, ORCHIDEE, BETHY, Biome-BGC, Triffid, ED) have or can all be coupled relatively straightforward with the land-surface model of an atmospheric mesoscale model. On the other hand, the Eurogrid is the grid size which is also accessible by means of upscaling with site-specific models using high-resolution georeferenced surface characteristics (topography, hydrography, edaphic conditions, ecosystem and agriculture distribution, land management information, etc.) from remote sensing and statistical data (see Regional Experiment, Component 3) and information from flux site clusters.

In the top-down approach spatio-temporal variations of the atmospheric CO2 concentration observed by the Atmosphere Component of the CarboEurope-IP (Component 2) are used to infer the net surface exchange fluxes by means of inverse atmospheric modelling. CO2 inverse modelling will be complemented by multitracer analyses. Thereby, measurements of atmospheric O2/N2 and d 13C in CO2 ratios will help to separate oceanic from terrestrial contributions, while radiocarbon ( 14CO2) and CO will independently constrain the fossil emissions within Europe . Transport tracers such as SF 6 and CFCs measured at some stations and 222Rn measured at all stations will be used to check on the performances of atmospheric transport models, and if necessary, used to improve them. As an added value, and although the focus of the project is CO2, the sources of CH4 and N2O will also be analysed using the same inverse modelling method, which will provide a better quantification of the European sources of these species.

The inversion modelling work will be based on the methods established in the Aerocarb project within FP5. Several high-resolution global models of atmospheric transport based on the observed meteorology from weather forecast models, or nested limited area mesoscale atmospheric circulation models will be used to model the atmospheric transport: LMDZ, REMO, DEHM, TM3/5, MM5, RAMS. The problem of atmospheric inversions being mathematically underdetermined is addressed using a Bayesian approach by means of careful inclusion of a priori information on magnitude, location and uncertainty of the various surface-atmosphere CO2 fluxes.

Bottom-up approaches proceed by extrapolating surface in situ process information (e.g. ecosystem functioning, weather and climate, land use, etc.), net ecosystem fluxes observed at individual flux towers, or by extrapolating local or county level carbon inventory data. In the Continental Integration Component we will use four types of bottom-up modelling approaches based on different and complementary concepts:

The bottom up approach will also include an enhanced mapping of spatial and temporal patterns of fossil fuel emissions at the Eurogrid scale because fossil fuel burning is the largest flux constituent of the European carbon balance. Building on the work of the CarboEurope-GHG Concerted Action in FP5, where 50 km resolution maps of fossil CO2 will be created fossil emission maps possibly will have to be improved over targeted areas, and they will have to be extended to cover also the mapping of industrial emissions of key tracers measured in the Atmospheric Component (CO, SF 6, CH4).

Ultimately, the bottom-up and top-down approaches will be merged into a CarbonData Assimilation System (CDAS). In this approach, coupled land surface – terrestrial ecosystem models (LSM-TEM) run coupled in an atmospheric high-resolution global or nested limited area mesoscale model (M-AGCM). In this system, data streams of different quality, temporal and spatial characteristics are merged in an optimal way which is mathematically consistent with the dynamics that govern the evolution of the system. This approach is similar to the techniques employed in numerical weather forecasting. This activity will build on the work of the Camels FP5 project. In CarboEurope-IP the very much enhanced observational datasets over the European domain will be used for constraining surface fluxes on much finer spatial and temporal scales (Eurogrid, daily-weekly) using the same methodology as in Camels.

A major goal of the integration activity consists in rigorous, quantitative consistency checks addressing fully each of the inherent uncertainties of the different modelling and extrapolation approaches. This will be performed by a series of benchmarking exercises in which modelled carbon flux estimates for defined target areas will be inter-compared and evaluated, where possible, against independent observations. This activity will be closely co-ordinated with the modelling activity in the Regional Experiment Component that provides the high resolution test bed for this. Additional target areas will be defined in several of the major ecosystems in Europe and selected on the basis of data availability and existence of high-resolution mesoscale model analyses.

In order to reconcile the top-down and bottom-up approach, several additional, minor1 carbon flows have to be addressed as well. These include carbon flows through trade products, VOC emissions from the vegetation, CH4 from natural and anthropogenic sources, CO from incomplete fossil fuel burning, carbon transport by rivers, weathering and erosion fluxes, carbon stored in reservoirs and lakes, and carbon fluxes from marginal seas and continental shelves.

The organisation of the Continental Integration Component through different Activities is shown in the figure below. This figure also shows the interaction between the activities in Component 4 and the other CarboEurope-IP Activities. Component 4 is the core link between all Components with regard to data input (Components 1, 2, 3) and modelling and data assimilation (Component 3).

1 we use the term minor in comparison to the main biospheric flows of NPP and fossil fuel emissions, realising that these minor flows can on occasion by relatively large in terms of carbon amounts

Expected Results

The Continental Integration Component will provide several synthesis products:

 Organisation of the Continental Integration Component

Figure : The organisation of the Continental Integration Component through different Activities. The arrows indicate the relation to other Activities in the CarboEurope-IP

Activity 4.1 Auxiliary Datasets and Remote Sensing
Activity 4.2 Land Carbon Inventories
Activity 4.3 Determination of the Surface Carbon Balance by Inverse Atmospheric Modelling
Activity 4.4 Bottom-up determination of the surface carbon balance
Activity 4.5 Development of a carbon data assimilation system for Europe


updated by Yvonne Hofmann,