The ICOS atmosphere station network currently consists of 25 officially labelled stations (with 12 stations still to come), located in 12 countries, and covering Europe from Scandinavia to Italy and from Great Britain to the Czech Republic (see Fig. 1). The preferred station types are tall tower sites, allowing vertical profile sampling at a minimum of three height levels up to at least 100 m above ground level (a.g.l.). Tall tower stations cover footprints of several tens to hundreds of kilometres of distance from the sites (Gloor et al., 2001; Gerbig et al., 2006). Although their representation in state-of-the-art regional atmospheric transport models is more difficult than in the case of tower observations, due to their often long history of GHG measurements, a number of mountain and coastal stations are also part of the ICOS network. However, the flask sampling strategy developed here was designed specifically for the standard ICOS tall tower stations.

All ICOS atmosphere stations are equipped with commercially available instruments measuring CO2, CH4, and CO continuously at high temporal resolutions. Instruments are tested at the Atmosphere Thematic Centre (ATC), an ICOS Central Facility hosted by the Laboratoire des sciences du climat et de l'environnement (LSCE) in Gif-sur-Yvette, France, before they are installed at the sites (Yver Kwok et al., 2015). The calibration gases for the in situ measurements are prepared and calibrated at the Flask and Calibration Laboratory (FCL), which has been established at the Max Planck Institute for Biogeochemistry in Jena, Germany, as part of the ICOS Central Analytical Laboratories (CAL). This procedure guarantees the best possible compatibility of observations within the ICOS atmosphere network and maintains the link to the internationally accepted World Meteorological Organization (WMO) calibration scales. In addition, the FCL analyses the flasks with a focus on QC and additional species. Precise 14CO2 analysis of integrated samples and selected flasks is conducted in the second part of ICOS CAL at the Heidelberg University Institute of Environmental Physics in the Karl Otto Münnich Central Radiocarbon Laboratory (CRL).

All raw data (level 0) are automatically transferred, on a daily basis, from the measurement sites to the ATC, where they are converted to calibrated (level 1) concentration values (Hazan et al., 2016) based on regular on-site calibrations and FCL-assigned calibration values. For ongoing automatic data quality assurance of all measurements, the ATC has developed automatic procedures. Further software tools are made available by the ATC for mandatory validation of all raw data by the station PIs. These quality-assessed data form the basis of the hourly mean concentrations, which are finally released as level 2 data and made available to the user community on the ICOS Carbon Portal hosted by Lund University, Sweden. For the latest data release, see ICOS RI (2020a).

Two station types are currently implemented in the ICOS atmosphere station network, namely Class 1 and Class 2. Class 1 stations are equipped with the complete instrumentation, including integrated 14CO2 and flask sampling. Class 2 stations perform only in situ continuous measurements of CO2, CH4, and CO (currently not mandatory) but with the same instrumentation and demand for data quality as Class 1 stations. A detailed description of the specifications of the instrumentation is given in the ICOS Atmosphere Station Specification document (ICOS RI, 2020b), which is regularly updated. To become an official part of the ICOS atmosphere station network, stations have to undergo a two-step labelling process, which warrants their conformance with the ICOS station specifications, including smooth data transfer to the ATC and meeting ICOS data quality requirements.

To develop and test our flask sampling strategy, we selected five ICOS Class 1 tall tower stations in four different countries. A short description of these stations is given in the following.

Hyltemossa (HTM) is located a few kilometres south of Perstorp in northwestern Skåne, Sweden (56.098∘ N, 13.418∘ E; 115 m above sea level – a.s.l.). It hosts a combined atmosphere and ecosystem station labelled, respectively, as Class 1 and Class 2 sites in its respective networks. The site was established in 2014 in a 30-year-old managed Norwegian spruce forest. More than 600 m away from the tower there is a mosaic consisting of forests, clear-cuts, and farm fields. Within a radius of 100 km, the elevation changes from 0 to 200 m a.s.l., while in the near vicinity of the tower the elevation gently changes by only 35 m. In the larger footprint, the site is surrounded by cities; i.e. Halmstad to the north (70 km; 58 000 inhabitants), Kristianstad to the east (45 km; 36 000 inhabitants), Lund (45 km; 111 000 inhabitants), Malmö (60 km; 318 000 inhabitants), and Copenhagen (in Denmark; 70 km; 1 990 000 inhabitants) to the southwest, and Helsingborg (45 km; 124 000 inhabitants) and Helsingør (in Denmark; 55 km; 61 000 inhabitants) to the west. The station is equipped with a Picarro, Inc. G2401 cavity ring-down spectroscopy (CRDS) gas analyser that measures CO2, CH4, and CO. Air inlets are located at 30, 70, and 150 m a.g.l. Air is sampled for 5 min from each level, where the data for the first minute after switching to the new level are discarded. Subsampling lines have installed 8 L mixing volumes that are continuously flushed with a flow rate of 2.1 L min-1, resulting in a residence time of 270 s in each line. In addition, at the height of each air inlet, air temperature, relative humidity, and wind speed and direction are being measured.

The ICOS tall tower station Gartow (GAT; 53.066∘ N, 11.443∘ E; 70 m a.s.l.) is situated in the easternmost region of Lower Saxony, Germany, close to the river Elbe, approximately at the midpoint between Hamburg and Berlin. The surrounding area is very flat, with elevations ranging from less than 9 m a.s.l. (Elbe Valley) up to 124 m a.s.l. (at the Hoher Mechtin hill 35 km west of GAT). The land use in this area is dominated by forests and agricultural fields. The station hosts a lattice television tower operated and managed by the Deutsche Funkturm GmbH (DFMG). The closest cities are Schwerin (65 km north of the station; ca. 100 000 inhabitants), Wolfsburg (80 km south of the station; ca. 120 000 inhabitants), and Lüneburg (70 km northwest of the station; ca. 70 000 inhabitants). Air inlets are at 30, 60, 132, 216, and 341 m. A Picarro, Inc. G2301 cavity ring-down spectroscopy (CRDS) gas analyser, measuring CO2, CH4, and CO, and, since the beginning of 2019, a Los Gatos Research, Inc. (part no. 913-0015; Enhanced Performance – EP) off-axis integrated cavity output spectroscopy (OA-ICOS) analyser, measuring CO and N2O, have been installed in a container next to the tower. Air is sampled for 5 min from each level, where data for the first minute after switching to the new level are discarded. All inlet lines are continuously flushed with approximately 5 L min-1. Meteorological sensors for air temperature, relative humidity, and wind speed and direction have been installed at every sampling height. For historical reasons, Gartow modelling was conducted for 344 m a.g.l. (and not for the highest sampling level at 341 m); this difference between the measured and modelled level is, however, not relevant for the comparisons presented in the context of this study.

Station Křešín u Pacova (KRE; 49.572∘ N, 15.080∘ E; 534 m a.s.l.) is located in the central Czech Republic, about 100 km southeast of Prague. The site was established in 2013 close to the Košetice Observatory, a station of the Czech Hydrometeorological Institute with 30 years of practice in meteorology and air quality monitoring. Today, these two stations form the National Atmospheric Observatory in the Czech Republic. Since the site is designed as a background station, the area is not significantly influenced by human activity. The tower is surrounded by fields and, at a greater distance, forests and small villages (the closest is 1 km away). There is a highway running northeast of the tower at an approximate distance of 6 km; however, the wind frequencies from the north and east are 9 % and 5 %, respectively. The closest towns, namely Pelhřimov, Vlašim, and Humpolec, with 10 000 to 17 000 inhabitants, are located approximately 20 km away from the station. As for industrial activity, a small wood-processing company is located 20 km to the west (which is the prevailing wind direction). The town of Havlíčkův Brod, with ca. 20 000 inhabitants, is located about 30 km from the site; larger towns (up to 50 000 inhabitants) are about 40 km away (i.e. Jihlava and Tábor). Further still, there are only towns with populations of, at most, 35 000 inhabitants, except for Prague (80 km; 1 million inhabitants), Pardubice (80 km; 90 000 inhabitants), and České Budějovice (90 km; 90 000 inhabitants). The terrain around the tower is relatively flat within a few kilometres' distance, with only small hills around. The Bohemian-Moravian Highlands, where the site is located, have an average altitude of 500–600 m a.s.l., with rare spots of 800 m a.s.l. The highest hills, namely Javořice (837 m a.s.l.) and Devět skal (836 m a.s.l.), are located 43 m and 69 km away. The station is equipped with the ICOS atmosphere-recommended instrumentation for CO2 and CH4 (Picarro, Inc. G2301 CRDS) and for N2O and CO (Los Gatos Research, Inc.; part no. 913-0015; EP). The air is sampled at 10, 50, 125, and 250 m levels of the tower. Sampling period is 10 min per height, where the highest level is sampled in between all other levels. This results in a complete vertical profile measured within 1 h, with a preference for the 250 m level. After switching to a new height, 3 min measurements are always excluded (known as the stabilisation period). All sampling heights of the tower are equipped with meteorological sensors (wind speed and direction, air pressure and temperature, and relative humidity).

The Observatoire Pérenne de l'Environnement (OPE; 48.563∘ N, 5.506∘ E; 395 m a.s.l.) is located on the eastern edge of the Paris basin in the northeastern part of France. The station is located in a rural area with large crop fields, some pastures, and forest patches. A local village and small roads are about 1 km away. The closest large towns are between 30 and 40 km away, and a major road is found at a distance of about 15 km. The station hosts a complete set of in situ measurements of meteorological parameters, trace gases (CO2, CH4, N2O, CO, O3, NOx, and SO2), and particle characteristics. The station is part of the French aerosol in situ network, contributing to Aerosol, Clouds and Trace Gases Research Infrastructure (ACTRIS; https://www.actris.eu/, last access: 20 September 2020) and the Institut de Radioprotection et de Sûreté Nucléaire (IRSN) network for ambient air radioactivity monitoring. It also contributes to the French air quality monitoring network and to the European Monitoring and Evaluation Programme (EMEP). The infrastructure, including a 120 m tall tower, was built in 2009–2010, and the various measurements started between 2011 and 2013. Ambient air is sampled at three levels, namely 10, 50 and 120 m a.g.l., of the tower and is analysed by Picarro, Inc. cavity ring-down spectrometers (CRDSs; series G1000 and G2000) for CO2, CH4, H2O, and CO as well as Los Gatos Research, Inc. off-axis-ICOS spectrometers for N2O and CO (Conil et al., 2019). The sampling period for each level is 20 min, including an automatic rejection of the first 5 min. Meteorological parameters are measured at all air sampling levels.

The ICOS station Hohenpeißenberg (HPB; 47.801∘ N, 11.010∘ E; 934 m a.s.l.) is located on top of a solitary hill that rises approximately 300 m above the almost flat to rolling landscape, 30 km north of the Alps and approximately 60 km southwest of Munich. The main land uses are forests and meadows. The station hosts a concrete television tower operated and managed by the DFMG. Cities closest to the station are Weilheim (10 km east of the station; 20 000 inhabitants), Landsberg (30 km north of the station; 30 000 inhabitants), Augsburg (60 km north of the station; 270 000 inhabitants), Munich (60 km northeast of the station; 1 million inhabitants), and Innsbruck (in Austria; 65 km south of the station; 127 000 inhabitants). Air inlets are at 50, 93, and 131 m. A Picarro, Inc. G2401 CRDS analyser, measuring CO2, CH4, and CO, and a Los Gatos Research, Inc. (part no. 913-0015; EP) OA-ICOS analyser, measuring CO and N2O, are installed in the basement of the tower. Air is sampled for 5 min from each level, where data for the first minute after switching to the new level are discarded. All inlet lines are continuously flushed with approximately 5 L min-1. Meteorological sensors (air temperature, relative humidity, and wind speed and direction) are installed at every sampling height.

A footprint simulation tool based on the regional atmospheric transport model, STILT (Lin et al., 2003; Gerbig et al., 2006), was implemented at the ICOS Carbon Portal (https://www.icos-cp.eu/about-stilt, last access: 20 September 2020) as a service for ICOS station PIs and data users. The STILT model simulates atmospheric transport by following a particle ensemble, released at the measurement site, backwards in time and calculating footprints that represent the sensitivity of tracer concentrations at this site to surface fluxes upstream. The footprints are mapped on a 1/12∘ latitude ×1/8∘ longitude grid and are coupled to the Emission Database for Global Atmospheric Research (EDGAR) version 4.3.2 emission inventory (Janssens-Meanhout et al., 2019) and the biosphere model, Vegetation Photosynthesis and Respiration Model (VPRM; Mahadevan et al., 2008), to simulate atmospheric CO2 and CO concentrations. These regional concentration components represent the influence from surface fluxes inside the model domain (covering the greater part of Europe). For CO2, the contributions from global fluxes are accounted for by using initial and lateral boundary conditions from the Jena CarboScope globally analysed CO2 concentration fields (http://www.bgc-jena.mpg.de/CarboScope/s/s04oc_v4.3.3D.html, last access: 20 September 2020), while for CO only regional contributions are evaluated in our study. Note that STILT does not account for the stack emission height of point sources. This may cause biases when estimating ffCO2 contributions from close-by emissions of, for example, power plants. However, as this model deficiency becomes less important with increasing distance from the source, it seems of minor relevance for the ICOS stations studied here as they are located far away from major emitters.

The automated ICOS flask sampler was designed and constructed at the Max Planck Institute for Biogeochemistry (MPI-BGC), Jena, Germany, by the Flask and Calibration Laboratory (FCL) of the CAL to allow automated air sampling under highly standardised conditions. The sampler can hold up to 24 individual glass flasks (four drawers with six flasks each) for separate air sampling events (Fig. 2, upper panel). The glass flasks can be individually replaced and sent to the CAL for analysis. The glass flasks used within ICOS (3 L volume; product no. ICOS3000; Pfaudler Normag Systems GmbH, Germany) were developed according to ICOS' specific requirements based on well-proven designs (Sturm et al., 2004). Each flask has two valves, one at each end, that allow air exchange by flushing sample air through the flask. The flasks are attached with 1/2 in. clamp ring connectors to the flask sampler. The flask valves, with polychlorotrifluoroethylene (PCTFE) sealed end caps, can be opened and closed by a motor.

A sample is taken by flushing air through a flask at a constant overpressure of 1.6 bar (absolute). Sampling at overpressure increases the amount of available sample air for analysis and allows for the detection of flasks with leak problems. Flasks are prefilled with 1.6 bar of dry ambient air with a well-known composition at the FCL to avoid concentration changes due to wall adsorption effects. The schematic sampler layout is depicted in the flow diagram in Fig. 2. Incoming air is dried to a dew point of approximately -40 ∘C by passing through a cooled glass vessel where the exceeding humidity is frozen out. The glass vessel is placed in a silicon oil heat bath that is cooled for drying and heated for flushing out the collected water to regenerate the trap. The drying unit is automated and consists of two independent inter-switchable drying branches that complement each other and allow a near interruption-free drying. The dryer design is inspired by an already existing system from Neubert et al. (2004). The incoming sample air is compressed with a pump (J161-AF-HJ0; Air Dimensions, Inc.). A mass flow controller (MFC; F-201CV; Bronkhorst) between the compressor and flasks allows one to sample preset flow rates; i.e. with a decreasing flow rate over time so that the sample represents a real average, for example, over 1 h (Turnbull et al., 2012). The flask pressure during sampling (1.6 bar) is kept constant through a pressure regulator at the outlet of the flasks. An overpressure valve set at 2.0 bar behind the pump assures a constant flow rate through the intake line, independent of the flow rate through the mass flow controller.

In ICOS we strive to sample real 1 h mean concentrations in 3 L flasks. The 1/t filling approach requires, for this specific case, a theoretical dynamic flow rate between 80 mL min-1 and infinity. In reality, the maximum flow rate of the selected flow controller is limited to 2 L min-1. An almost constant weighting of the sample concentration over the 1 h sampling time is achieved by the temporal modulation of the sample flow f(t) (standard litre per minute – SLPM) passing a flask, which acts at the same time as mixing volume V given in litre standard temperature and pressure (STP). The flow rate f is changed over time t according to f(t)=V/(t-t0). Since the flow rate at the start time t0 in a 1/t function would be infinite, a 30 min flushing phase at maximum flow rate precedes the averaging period to ensure a complete air exchange in the flask with ambient air before the sampling starts.

The concentration cF(t) in the flask is determined by the ambient air concentration cA(t) and can be described as a time series using sufficiently small time steps Δt as follows: 1cFt+Δt=cFt⋅V-f(t)⋅Δt+cAt⋅f(t)⋅ΔtV.

The resulting weight of the ambient air concentration wcA at time step tn in the flask depends on the following two factors: 2wcAtn∼cAtn⋅ftn⋅ΔtV⋅∏i=n+1E1-fti⋅ΔtV, namely the weight at the moment when the ambient air portion enters the flask, and a weight-reduction factor caused by dilution with sampled air entering the flask at later times. The reduction is calculated by multiplication of the respective dilution steps from tn to the sampling end time tE. This weighting function has to be applied to the ambient air measurements so that the flask concentrations can be compared with the in situ data. Average in situ minus flask concentration differences with the aimed uncertainty can only be reached under sufficiently stable concentration conditions during sampling.

With the current design of the flask sampler, technical restrictions do not allow parallel sampling of flask duplicates or triplicates as a means for quality control, for example, based on flask pair agreement. The technical effort for allowing exact parallel hourly averaged sampling is very high; it would, for example, require flow controllers for all individual flasks sampled in parallel. Therefore, the ICOS Atmosphere Monitoring Station Assembly (MSA) decided to sample only single flasks. This seems appropriate because in the ICOS network the flask sampler is always collecting flasks in parallel to continuous measurements, and erroneously collected flasks, or errors due to flask leakages, can be detected when comparing results with the continuous data. Therefore, in contrast to the general practice of duplicate flask sampling, in our network single flask sampling seems to be sufficient for meeting ICOS objectives. This has the additional advantage that single flask sampling allows more frequent sampling and, thus, a more representative coverage of the footprint of the stations. If true duplicate samples are required in the future, the flask sampler is designed to accommodate an additional mass flow controller to fulfil this task. The sampler is controlled by an embedded PC offering a broad range of interaction possibilities satisfying the emerging needs within ICOS. Sampling event time schemes can be preprogrammed, and communication with external devices (i.e. data loggers) is possible with analogue or digital signals. Flask-to-port attributions are completely barcode controlled. Sampling and sensor data are automatically stored, and all necessary sampling-related data can be automatically transferred to the CAL. Various automated internet-assisted approaches, like remote programming of sampling times and preselection of samples, are possible.

The ICOS atmosphere station network, supported by the ICOS Central Facilities (ATC and CAL), has been designed and implemented to achieve the highest possible accuracy, precision, and compatibility of atmospheric GHG measurements. For ICOS CO2 observations, a compatibility goal of 0.1 ppm or better is compulsory. Similarly, ICOS needs to meet the WMO compatibility goals for CH4 and CO, which are 2 parts per billion (ppb) for both gases (WMO, 2020). First evaluations of ICOS CO2 measurements indeed yield monthly mean afternoon differences between stations in the free troposphere above 100 m of typically very few parts per million (Ramonet et al., 2020), underlining the importance of the excellent precision and compatibility of these observations.

With regular and frequent comparisons of flask and in situ measurements, ICOS aims to independently monitor their compatibility and provide respective alerts if, for example, the average difference of CO2 exceeds 0.1 ppm over a few weeks of comparisons. Using flasks sampled from a dedicated intake line to crosscheck the in situ measurements is an important part of the ICOS quality management. It allows an independent end-to-end QC of the entire in situ measurement system consisting of inlet system, drier, analyser, and calibration. As mentioned above, for logistical reasons, about once per month, or every 5 weeks, a box with 12 flasks is scheduled to be shipped from a remote station to the FCL. After analysis, the flask results covering about 1 month of time will be compared with the corresponding in situ data. In the following paragraph, we elaborate on the minimum number of comparison flasks and the corresponding time delay for detecting a significant CO2 bias between flask and in situ measurements larger than 0.1 ppm. Therefore, we tested the envisaged flask sampling procedure experimentally at the ICOS pilot station in Heidelberg and present here its first application at an ICOS field station.

In the preceding section we showed that low ambient variability situations would be best suited for meeting aim 1 as synchronisation and exact weighting of flask filling and in situ measurements are not so important at low ambient variability. Moreover, a potential bias between flask and in situ measurements could be detected with better confidence and with an increased number of comparisons. However, to meet aim 2, a scheme for collecting flasks only during low variability situations may cause a significant bias in the sampled footprint. We have tested if such a sampling bias would be visible in the European ICOS network and calculated, with STILT, all midday (13:00 LT) footprints of the five selected stations for the year 2017, using the Jupyter Notebook package of Karstens (2020). Figure 4 shows the respective aggregated footprints for October 2017. A time of 13:00 LT was chosen as an example throughout the paper, but other afternoon hours could also have been chosen, leading to similar results. The left column in Fig. 4 shows the aggregations if every afternoon hour (13:00 LT) was sampled, the middle column shows the aggregated footprints for every third day, and the right column shows the 10 footprints with the lowest variability during October 2017. As expected, the regional coverage of the entire station footprint is generally better when sampling randomly, every third day, than when sampling on the 10 d with the lowest variability.

In addition to the footprint analysis, which gives a visual, qualitative idea of the effect of different flask sampling schemes, we evaluated the first 3 years of continuous CO2 measurements from the five ICOS stations to quantify the effect of random sampling every 3 d versus only sampling low variability situations. Figure 5a–e show, in the upper panels, for each station, all available hourly atmospheric CO2 data as grey dots, while the blue lines, each shifted by 1 d, connect the 13:00 LT data every 3 d. The red dots in the upper panels highlight the 10 lowest variability afternoon values in each month. As expected, all summer afternoon concentrations generally fall into the lower concentration range of the bulk of data. At all stations, the variability changes from a diurnal shape during the summer months to a more synoptic variability in the winter (for more details, also see Figs. 7 and 8). This synoptic variability is also represented in the afternoon sampling. In the middle panels of Fig. 5a–e we have plotted, as black dots, monthly means calculated from all afternoon hours between 11:00 and 15:00 LT and their standard deviations. The blue dots show the monthly mean values obtained from sampling every third day (the three different 3 d patterns are shown in individual shifted blue dots), while the red dots represent the monthly means calculated from the 10 samples with the lowest variability (the coloured dots were shifted by 1 d each for better visibility). It is obvious that regular sampling provides better representative monthly means, deviating in only a few cases from the all-afternoon means in CO2 by more than 2 ppm (Fig. 5a–e, bottom panels). If samples were collected at low variability only, they would often underestimate monthly mean values, in some cases by more than 4 ppm (red lines in Fig. 5a–e, bottom panels). Although regular sampling every third day also introduces some variable deviations from the correct afternoon means, sampling only at low variability may introduce rather large biases – mainly towards lower CO2 concentrations. Note that inversion models also select measured data for their inversion runs only for the time of the day, and not for low variability data, to estimate fluxes (Rödenbeck, 2005).

We have investigated only potential sampling effects on CO2 concentrations here; however, other tracer concentrations are also expected to be affected in a similar way. For the ICOS atmosphere network we, therefore, choose the simpler sampling scheme of one flask every third day. This sampling scheme is expected to serve aims 1 and 2, where those flasks with low within-hour variability (on average one flask per week; see Sect. 4.1) could be used for the quality control aim, while all flask samples would deliver as much representative data as possible for all additional trace components analysed in the FCL solely based on flasks.

The first 14C analyses on integrated CO2 samples at ICOS stations showed rather low average fossil fuel CO2 (ffCO2) concentrations, therewith confirming that ICOS stations primarily monitor the terrestrial biospheric signals. Figure 6a–d (upper panels of the graphs for the individual stations) shows our first 14CO2 results from the 2-week integrated CO2 sampling at Hohenpeißenberg, Observatoire Pérenne de l'Environnement, Hyltemossa, and Křešín. Particularly during summer, the monthly mean regional fossil fuel CO2 offsets, if compared to a background level calculated from the composite of 2-week integrated 14CO2 measurements at Jungfraujoch in the Swiss Alps and Mace Head on the Irish coast, are often lower than a few parts per million (Fig. 6a–d, lower panels). Only during winter can regional ffCO2 offsets reach 2-week mean concentrations of more than 5 ppm. These signals, although providing good mean ffCO2 results for the average footprints of the stations, are often too small to provide a solid top-down constraint of regional fossil fuel CO2 emission inventories and its changes when evaluated in regional model inversions (Levin and Rödenbeck, 2008; Wang et al., 2018). One of the aims of flask sampling in ICOS is, therefore, to explicitly sample air which has passed over fossil fuel CO2 emission areas. Ideally we would like to obtain signals and analyse flasks for 14CO2 only in cases when the expected fossil fuel CO2 component is larger than 4–5 ppm. This would allow us to obtain an uncertainty of the estimated ffCO2 component below 30 % (Levin et al., 2003; Turnbull et al., 2006). Furthermore, as sample preparation for 14C analysis is very laborious and the capacity of the CRL is limited to about 25 flask samples per station per year, one should know beforehand if a sample potentially contains a significant regional fossil fuel CO2 component. This could either be found out with near real-time transport model simulations or directly using the in situ observations at the station.

A good indicator of the potential regional fossil fuel CO2 concentration at a station is the ambient CO concentration (Levin and Karstens, 2007), a trace gas that is monitored continuously at all ICOS Class 1 sites. It would then depend on the average CO:ffCO2 ratio of fossil fuel emissions in the footprint of the stations to estimate, from the measured CO, the expected ffCO2 concentration. Mean CO:ffCO2 emission ratios can be very different in different countries; they mainly depend on the energy production processes and on domestic heating systems (Gamnitzer et al., 2006; Turnbull et al., 2006, 2011; Levin and Karstens, 2007; Vogel et al., 2010). In this respect, the share of biofuel use may also be relevant. In our study we first analysed our selected ICOS stations for regional fossil fuel CO2 signals larger than 4 ppm and determined the frequencies of those events. Note that in order for the flask results to be used in transport model investigations, similar to all other flask samples, 14CO2 flasks should also be collected during early afternoon when atmospheric mixing can be modelled with good confidence. During these situations, however, any ffCO2 signals will be highly diluted. Similar to the approach in the previous section, we investigated the potential ffCO2 levels for the five stations of Hyltemossa, Gartow, Křešín, Observatoire Pérenne de l'Environnement, and Hohenpeißenberg; this was first done theoretically with STILT model simulations transporting EDGAR version 4.3.2 emissions to the five measurement sites. As a second step, we evaluated the real continuous CO2 and CO observations from 2017 and 2018 (see Table 1).