The atmosphere is a fast but incomplete mixer and integrator of spatially and temporally varying surface fluxes of GHGs. Through inverse modelling the distribution and temporal evolution of CO2 and other GHGs in the atmosphere can be used to quantify surface fluxes, using numerical models of atmospheric transport. To make progress in the objectives described in the previous two sections a number of problems must be overcome, as large-scale inversion based sink/source estimates of GHGs suffer from a number of errors (see e.g. Gerbig et al., 2003, for a discussion with respect to CO2 but largely valid also for other GHGs):

1. •Solving for continental fluxes is an ill-constrained inverse problem given uneven distributed and sparse stations under continental influence.
   

2. •There is a large variability in both atmospheric transport and surface fluxes over vegetated areas, which yields a peak in the CO2 variance near the ground, so that the signal of mean continental fluxes is difficult to characterize from the noise (variability) when only discrete (flask) sampling is available.
   

3. •Vertical mixing and biogenic surface fluxes (e.g., NEE) - but also anthropogenic fluxes (e.g. traffic) - vary together on diurnal and seasonal scales to yield rectification gradients in mixing ratios that are especially difficult to capture in large scale transport models [Denning et al., 1996].
   

4. •Diurnal and seasonal atmospheric transport processes (e.g., boundary-layer mixing and ventilation, orographic effects, frontal uplift of tracers etc.) are usually poorly represented in large-scale transport models.
   

5. •Given the flux heterogeneity, measurements from a single location are not immediately representative of larger regions or grid cells thus causing representation errors, and preventing the information obtained at these sites to be simply aggregated up to the scale of a region.
   

6. •Solving for aggregate fluxes that do not evenly influence the overall mixing ratio may causes aggregation errors [Kaminski et al., 2001].
   

These errors can be substantially reduced when at the regional level a good link between the measurements obtained at the surface flux stations and those from continental scale inversions can be established. To achieve this, a region needs to be monitored equally well in spatial and temporal terms, while the precision and representativeness of the measurements should match that of the involved major source and sink processes. Therefore, continuous monitoring of GHG-concentrations in the PBL, of PBL-dynamics and its drivers, and of temporal and spatial variability of surface fluxes are one side of that link. Using atmospheric transport models of high resolution, with good parameterizations of PBL processes, of surface fluxes of heat and GHGs and resolving mesoscale transport (e.g., see breezes, orographic flows, convection) are the other side of this link. Combining the two, inversions then provide objective uncertainties of the estimated net surface fluxes. Using various combinations of atmospheric observations, the inverse modelling framework also permits studies of optimal sampling strategies for adding new stations to the network and for filtering the concentration "signal" representative of regional sources and sinks, from the "noise" induced by local fluxes and transport patterns.

In order to produce a best estimate of carbon dioxide uptake and its uncertainty, we need to make use of all of the constraints provided/imposed by the different data streams, as well as the physiological and ecological constraints embodied in behaviour of the land surface and the process-based Terrestrial Ecosystem Models (TEMs). Thus, we need to simultaneously use the observations to constrain the internal parameters of the TEMs, whilst using the TEMs to interpolate the plot scale observations to produce useful large-scale estimates of the carbon sink and its causes. In the frame work of the EU-FP5 funded RECAB-project [Hutjes, et al. 2003] a first attempt has been made to quantify CO2 emissions at regional scales. With analyses still continuing (e.g., under the EU-FP6 CarboEuropeIP Regional Component) a number of (tentative) lessons can be drawn from that project, and others addressing similar scales (e.g., CHIOTTO). Airborne flux measurements can provide estimates of regional fluxes that are directly comparable to tower based flux estimates [Gioli et al., 2004] and that can be used to validate fluxes simulated by distributed land surface models. Continuous, tall-tower based concentration measurements exhibit large signals (relaxing the need for sensor accuracy somewhat), that provide valuable information on diurnal flux dynamics representative of areas of ~104km2, provided that also the dynamics of PBL growth are well known. The footprint of such observations may be known through inverse lagrangian approaches, e.g. back trajectories, etc. CBL budgetting approaches are sometimes successful, but basically are limited in their application by the same factors: general lack of information on important parameters in PBL-dynamics (e.g., entrainment, advection) and imprecise knowledge of the footprint. On the modelling side it has been demonstrated that both forward and inverse approaches can be "downscaled" to resolutions of a few kilometers and may work in principle. But, where ecosystem fluxes are relatively well known for small space (plot) and time (diurnal) scales, anthropogenic fluxes are generally known only at national and annual levels respectively (from census data), requiring development of proper downscaling techniques.

The present proposal will build on this heritage from RECAB and others, while closely collaborating with ongoing initiatives (CarboEurope IP) to advance on the development of both observational and modelling components that together will constitute an information system that will eventually be able to quantify - with known uncertainties - magnitudes and trends of CO2 and other GHG emissions at scales of 103–105km2. Most partners involved in the present project maintain active research lines well embedded in the mentioned past and present international frameworks. Proper embedding, close collaboration, and sharing of data and tools will provide essential input for the proposed project to achieve its ambitious goals, and minimize the risk of failure by provision of a certain level of redundancy of tasks to be completed in both this project and its European equivalent.

It is a project structured around five work packages, as outlined in section Methodology. The aim is to establish an intensive observational programme both at the ground and in the atmosphere, in order to quantify with the highest possible accuracy the balance of the most important greenhouse for the Netherlands for at least a full year. Improved, high resolution atmospheric transport models -forward and inverse- will help interpret and analyze the data. Multiple components will be measured and modelled as together they may constrain the atmospheric budgets better than. E.g. the use of 222Rn to test the model behaviour for continental influence on the air masses. The forward modelled concentrations of components in the order of growing source complexity (222Rn, SF6, HFC, N2O, CH4, CO2) will help to evaluate and compare the performance of the different models. The emission budget of the other most important greenhouse gases will be quantified using the same models and inverse methods as for CO2 For socio-economic relevance we refer to the original proposal, which you can download here.

A complete PDF-version of the project proposal (excl. Work Package Six) can be downloaded here.

next >

< previous