Articles | Volume 11, issue 12
20 Jun 2011
 | 20 Jun 2011

Variability and budget of CO2 in Europe: analysis of the CAATER airborne campaigns – Part 1: Observed variability

I. Xueref-Remy, C. Messager, D. Filippi, M. Pastel, P. Nedelec, M. Ramonet, J. D. Paris, and P. Ciais

Abstract. Atmospheric airborne measurements of CO2 are very well suited for estimating the time-varying distribution of carbon sources and sinks at the regional scale due to the large geographical area covered over a short time. We present here an analysis of two cross-European airborne campaigns carried out on 23–26 May 2001 (CAATER-1) and 2–3 October 2002 (CAATER-2) over Western Europe. The area covered during CAATER-1 and CAATER-2 was 4° W to 14° E long; 44° N to 52° N lat and 1° E to 17° E long; 46° N to 52° N lat respectively. High precision in situ CO2, CO and Radon 222 measurements were recorded. Flask samples were collected during both campaigns to cross-validate the in situ data. During CAATER-1 and CAATER-2, the mean CO2 concentration was 370.1 ± 4.0 (1-σ standard deviation) ppm and 371.7 ± 5.0 (1-σ) ppm respectively. A HYSPLIT back-trajectories analysis shows that during CAATER 1, northwesterly winds prevailed. In the planetary boundary layer (PBL) air masses became contaminated over Benelux and Western Germany by emissions from these highly urbanized areas, reaching about 380 ppm. Air masses passing over rural areas were depleted in CO2 because of the photosynthesis activity of the vegetation, with observations as low as 355 ppm. During CAATER-2, the back-trajectory analysis showed that air masses were distributed among the 4 sectors. Air masses were enriched in CO2 and CO over anthropogenic emission spots in Germany but also in Poland, as these countries have part of the most CO2-emitting coal-based plants in Europe. Simultaneous measurements of in situ CO2 and CO combined with back-trajectories helped us to distinguish between fossil fuel emissions and other CO2 sources. The ΔCO/ΔCO2 ratios (R2 = 0.33 to 0.88, slopes = 2.42 to 10.37), calculated for anthropogenic-influenced air masses over different countries/regions matched national inventories quite well, showing that airborne measurements can help to identify the origin of fossil fuel emissions in the PBL even when distanced by several days/hundreds of kms from their sources. We have compared airborne CO2 observations to nearby ground station measurements and thereby, confirmed that measurements taken in the lower few meters of the PBL (low-level ground stations) are representative of the local scale, while those located in the free troposphere (FT) (moutain stations) are representative of atmospheric CO2 regionally on a scale of a few hundred kilometers. Stations located several 100 km away from each other differ from a few ppm in their measurements indicating the existence of a gradient within the free troposphere. Observations at stations located on top of small mountains may match the airborne data if the sampled air comes from the FT rather than coming up from the valley. Finally, the analysis of the CO2 vertical variability conducted on the 14 profiles recorded in each campaign shows a variability at least 5 to 8 times higher in the PBL (the 1-σ standard deviation associated to the CO2 mean of all profiles within the PBL is 4.0 ppm and 5.7 ppm for CAATER-1 and CAATER-2, respectively) than in the FT (within the FT, 1-σ is 0.5 ppm and 1.1 ppm for CAATER-1 and CAATER-2, respectively). The CO2 jump between the PBL and the FT equals 3.7 ppm for the first campaign and −0.3 ppm for the second campaign. A very striking zonal CO2 gradient of about 11 ppm was observed in the mid-PBL during CAATER-2, with higher concentrations in the west than in the east. This gradient may originate from differences in atmospheric mixing, ground emission rates or Autumn's earlier start in the west. More airborne campaigns are currently under analysis in the framework of the CARBOEUROPE-IP project to better assess the likelihood of these different hypotheses. In a companion paper (Xueref-Remy et al., 2011, Part 2), a comparison of vertical profiles from observations and several modeling frameworks was conducted for both campaigns.

Final-revised paper