WP 5.4: Network design and source and sink estimates
Objectives
- Assess the effect of surface measurement network density on reduction of uncertainty in continental NOx
emissions
- Perform Monte Carlo and other uncertainty analysis to assess errors in model input and
formulation
- Improve the design of measurement networks in order to improve estimation of sources and sinks of pollutants
- Improve the quality of 4D pollutant climatologies, including those long enough to derive trends
- Improve chemical weather and air quality forecast.
Networks concerned are surface measurement networks, ground-based vertical profiling and airborne measurements, which complement present and future satellite measurements. Long-lived climate active gases (CO2, CH4), the reactive gases (O3, NO2, NOx, CO) and aerosols at different spatial scales will be analysed. Results are either of prospective nature, i.e. they indicate the anticipated improvement when extending networks, or they document in a concrete manner the uncertainty reduction due to use of additional data bases.
Assess the effect of surface measurement network density
on reduction of uncertainty in continental NOx
emissions
A method has been developed in the framework of GEOmon that allows to use surface measurements (NO2, O3) to assess the improvement in optimised NOx emissions. Continental-scale NOx emissions (Europe including Russia and Mediterranean regions) have been inverted from satellite derived NO2 tropospheric columns (from the SCHIAMACHY instrument) and using the continental scale model CHIMERE. A specific inverse modelling technique was set-up to perform the inversion work. This technique allows the assessment of the improvement in optimised NOx emissions by using surface measurements (NO2, O3) as shown in the figure below for NOx. A report on the work is available on the GEOmon intranet (for partners only, sorry).
New top-down estimates of NOx emissions in Europe from inverse modelling confirm recent bottom-up EMEP estimates
Trends in anthropogenic NOx emissions averaged over several countries, shown here for Germany and Spain
(figure taken from Konovalov et al., 2008, ACP).
| This study | Emissions from inversion of IUP SCIAMACHY and GOME tropospheric NO2 columns with the CHIMERE regional CTM. |
| EMEP (new) | Emissions from new EMEP inventory http://webdab.emep.int in spring 2007 |
| EMEP (old) | Emissions from old EMEP inventory http://webdab.emep.int in autumn 2006 |
Using this
method, we are systematically exploring to which extend measurement networks of varying density allow
discerning between bottom-up emission cadastres (for example EMEP), and optimised emissions obtained by
inverse modelling from satellite measurements. By this means it will be possible to document the expected
reduction in uncertainty in optimised emissions by use of surface measurements. In a next step, also trends in
NOx emissions will be taken into account. These results have been published in Konovalov et al., 2008.
Perform Monte Carlo and other uncertainty analysis to assess
errors in model input and
formulation on the European scale
A Bayesian Monte Carlo (BMC) analysis framework has been set-up for evaluating model uncertainty in O3 production and in its sensitivity to emission changes. It allows the evaluation of uncertainty due to meteorological input parameters, to emissions, to reaction rates and photolysis frequencies, to dry deposition, and to vertical mixing. This approach has been applied as a first step for regional modelling with the CHIMERE model in the Ile-de-France region during two summer seasons. Measurements of urban NO and O3 concentrations and rural O3 over the Paris area from the AIRPARIF network has been used for constraining the Monte Carlo simulations. The major conclusion from this work was that the chemical regime over the Paris agglomeration and within the plumes exhibits a regime clearly VOC sensitive on the average over two summers, and that this statement is robust with respect to the BMC uncertainty analysis. These results were published (Deguillaume et al., 2008).
The next step was to extend uncertainty analysis from a regional to a continental European domain in order to derive regions with maximal uncertainty. In the regional analysis, the uncertainty has been set as constant over the model domain. This simplification is no longer valid over a larger and more heterogeneous continental domain. Thus, prestudies have been started on how to address the spatial variability and correlation of the different error sources. Once these issues will have been fixed in the next 6 months, an ensemble run with about 50 perturbed ensembles will be performed for a three month period in summer 2004 (a typical average summer) with the CHIMERE model to address uncertainty in photooxidant buid-up. Longer simulation periods would be prohibitive because of limited computer time.
A strategy has been designed to address the uncertainty in modelled aerosol (PM2.5) concentrations. The LOTOS-EUROS model will be used to perform ensemble runs with 12-15 perturbed ensembles. The ensembles will be created by either putting noise on the emissions or the deposition velocities or both. The investigated time period will be a three month period in the summer of 2003. To test the sensitivity for different meteorologies another ensemble run will be performed where we will use different members of an ensemble of ECMWF forecasts. Taking note of the availability of the meteorological input, this last ensemble run will be performed for a 1-month period in summer 2003.
