Supplementary material for TCCON Nicosia site description manuscript

1 S1 Detector chamber – InGaAs & InSb setup Figure S1: Detector chamber of the FTIR. Yellow path shows the light source entering the chamber, split 50/50 by the CaF2 beamsplitter (BS) and directed to the InGaAs and InSb detectors. A five-place filter wheel is exchanging the filter in front of the InSb according to the experiment. This five-place filter wheel holds four filters with one place open. An eight-place filter wheel 5 holding two filters, with six open places is located before the detector chamber entrance. The filters’ band frequencies (wavenumbers) are presented in Table 1 in the main paper. 2 S2 The Cyprus AirCores S2.1 The AirCores sampling 10 Figure S2: Photos from Nicosia, Cyprus AirCore campaign. A cross section of our AC sampling system (lower left) and the AC just launched (upper left). Here, we use a low-resolution AC device from LSCE built as a double stainless-steel, coated tubing of 35 m long, consisting of a 12 m long 8 mm in diameter tube and a 23 m long 4 mm in diameter tube. The vertical air samples were analysed with a cavity ring-down spectrometry (CRDS) gas analyzer by Picarro, model G2401. Right: photo from the AC launch of 30 June 2020. 15 20 25 3 Table S1: AirCore flights info Flight Date Launch time (UTC) Launch location Lat (°N), Lon (°E) Recovery time (valve closing) (UTC) Landing location Lat (oN), Lon (oE) Landing location characterization Balloon cutoff altitude (km) Profile ceiling (km) Profile floor (km) Landing distance from FTS (km) 1 19-6-2020 09:05 35.01, 32.45 11:52 35.11, 33.34 Residential 33 22.86 0.83 5.5 2 29-6-2020 07:46 34.84, 32.87 10:30 35.05, 33.53 Rural 25 21.83 1.43 17 3 30-6-2020 08:40 34.84, 33.39 11:15 34.97, 33.39 Rural 30 23.10 1.11 20 Figure S3: AirCore (AC) flight trajectories and landing locations during the June 2020 campaign. The colored markers indicate the landing sites of each flight: blue for 19 June (Flight 1), orange for 29 June (Flight 2), and green for 30 June (Flight 3). The solid 30 trajectories are colored by altitude (see color bar). The dashed lines of corresponding color show the azimuthal range of the TCCON FTS line of sight during each flight. For each color pair, the rightmost dashed line corresponds to the FTS viewing direction at the beginning of the AirCore descent, and the leftmost dashed line corresponds to the viewing direction at the time of landing. Map data from © OpenStreetMap contributors 2024. Distributed under the Open Data Commons Open Database License (ODbL) v1.0. . 35 4 Figure S4: Hysplit airmasses backtrajectories for the AirCore flight of 29 June, running 96 h backwards. Black star represents the location of TCCON Nicosia (Cyprus), where the airmasses (in red, blue and green) arrive on 29 June, at 1.5 km, 5.5 km and 13 km altitude respectively. The green airmass appears to be part of the Asian Summer Monsoon Anticyclone (ASMA), and it was above 40 India 96 h before crossing over Cyprus. 5 S2.2 FTS Xluft data during AC flights Figure S5: Xluft time series during AC campaign (grey markers). Xluft flight medians in orange markers and red horizontal lines show the TCCON nominal values for Xluft median. This is a check that the Xluft flight median is within the median nominal values of 45 0.996 – 1.002. S2.3 Constructing the lower and upper bounds of the AirCore profiles Figure S6: Python snippet that constructs the lower and upper bounds of the main AirCore. It takes into account the in situ measurement uncertainty per gas specie (unc_values[specie]), which is 0.05 ppm, 0.7 ppb and 7 ppb for CO2, CH4 and CO 50 respectively, and the altitude uncertainty limits (‘lower_alt_unc_lim_gas_specie’ and ‘upper_alt_unc_lim_gas_specie’), as provided in the AirCore data files (https://zenodo.org/records/13132338, last access: 5 December 2024) for this type of AC sampling device. For each gas specie, two types for each of lower and upper bounds are constructed. The ‘specie_lo’ and ‘specie_lo_bo’ for the lower bounds, and the ‘specie_up’ and ‘specie_up_bo’ for the upper bounds. Then, these bounds are treated as extra AirCore profiles. We get the smallest value amongst the ‘specie_lo’ and ‘specie_lo_bo’ per altitude to construct the gas lower bound (‘gas_lower’) and the 55 largest amongst the ‘specie_up’ and ‘specie_up_bo’ to construct the gas upper bound (‘gas_upper’). See example in Fig. S7. 6 Figure S7: Left: An example of the bounds that result from the python code in Fig. S6 for CO for flight 1. Light blue marker indicates 60 ‘specie_lo’ and blue marker indicates ‘specie_lo_bo’. Pink marker indicates ‘specie_up’ and dark red indicates ‘specie_up_bo’. Right: In order to have just one set of bounds, we selected the smallest value in ‘specie_lo’ and ‘specie_lo_bo’ at each altitude to create the gas lower bound (‘CO_lower’, blue), and similarly, to create the gas upper bound (‘CO_upper’, red), we selected the largest value in ‘specie_up’ and ‘specie_up_bo’ at each altitude. 65 7 S2.4 Assembling full profiles Figure S8: This figure shows a schematic of the approach followed to construct a full vertical profile. For comparability with the FTS prior, the in situ profile needs to extend from 0 m (a.s.l.) to 70 km. Both left and right panels show the same assembled profile (CO2, 30 June flight); however, the left profile is in wet mole fractions and the right profile in dry mole fractions. Left (wet profile): 70 Example of an assembled AirCore (AC) profile for the flight on 30 June, 2020, showing CO2 in wet mole fractions. This vertical resolution profile, was used for the GGG2020 custom retrievals (see Sect. S2.5). Red stars represent the re-gridded AC profile, while a flat extrapolation of the lowest AC measurement to near-ground levels is shown as red crosses in the inset. The grey shading around the main AC profile indicates the uncertainty bounds; which is very small for CO2. The FTS prior profile is depicted as grey circles connected by a grey line, with the prior used to extend the in situ profile upwards shown as grey circles connected by a red 75 line. The horizontal dashed line marks the altitude of the last AC grid level. Near-surface in situ measurements are represented by the median (orange 'x') and the mean ± standard deviation (green triangle). The inset focuses on the lower 3 km, showing nearsurface variability and the comparability between the prior and in situ measurements. A complete profile is constructed by assembling 1) the re-gridded AC profile (red stars), 2) the FTS prior above the highest AC measurement (gray circles with red line), 3) flat extrapolation of the lowest AC measurement to near-ground levels (red cross at 0.88 km), and 4) the in situ surface median 80 (orange 'x') for the lowest two levels (0 and 0.42 km). Right (dry profile): Assembled AirCore profile in full resolution (in dry molefractions). This profile was used as the true profile, x, in Eq. 2, main paper. The ~8 ppm difference near the ground of the leftand 8 right-panel profiles is due to the difference between wet and dry mole fraction; which is largest at the surface due to the concentration of water there. 85 S2.5 Running custom GGG2020 retrievals We follow the method of Laughner et al. (2024) and run custom GGG2020 retrievals by replacing the GGG priors with the assembled AirCore profiles (see Fig. S8, left). Our method of re-gridding the AC profile to the FTS levels is simple (Fig. S8, left), unlike Laughner et al. (2024) that compute weighted averages of the in situ measurements around the adjacent FTS levels 90 (see Sect. C3 and Fig. C1 of Laughner et al. (2024)). Another difference is that when running the GGG2020 custom retrievals, we do apply the in situ correction (AICF) on the custom Xgas data as post-processing. In addition, we use retrieved FTS data of spectra recorded within ±1h around the AirCore central flight time instead of the AC landing time. Table S2 summarizes the retrieval fitting errors of custom versus public data. 95 Table S2: Mean fitting errors resulting from the fitting residuals (“xgas_error” variable in the public data), in “custom” vs. standard (public) data retrievals. Values are rounded to the nearest decimal. Flight number 1 2 Fitting error mean ± STD 3 custom.XCO2 (ppm) public.XCO2 0.48 ± 0.01 0.45 ± 0.01 0.54 ± 0.01 0.50 ± 0.01 0.57 ± 0.02 0.51 ± 0.01 custom.XCH4 (ppb) public.XCH4 2.17 ± 0.04 2.02 ± 0.08 2.20 ± 0.06 2.52 ± 0.06 2.45 ± 0.08 2.33 ± 0.05 custom.XCO (ppb) public.XCO 1.16 ± 0.04 1.22 ± 0.03 0.96 ± 0.01 0.99 ± 0.02 1.13 ± 0.04 1.13 ± 0.04 100 S2.6 Xgas and AC.Xgas uncertainty calculation All types of measurements are susceptible to two main types of effects causing measurement uncertainty: random and systematic. Below we list uncertainty sources for both FTS-derived Xgas (valid for standard and custom retrieval) and AirCore (AC)-derived Xgas measurements. [Xgas]: The FTS-derived dry-air mole fractions have two known sources of uncertainty described below. 105 9 1. Random effects; represented by the measurement variability within the selected flight window of ±1 hour around the AirCore (AC) central time, quantified as the standard deviation around the mean (ϵXgas.std). 2. Systematic instrument effects; deviations in Xluft from the nominal network value (0.999) introduce bias in Xgas values. To account for this, we include an Xluft-derived bias (∈Xluft) using Eq. C11 and values in Table C6 from Laughner et al. (2024). The total uncertainty is calculated by combining standard uncertainties (from random effects) in quadrature. Known systematic 110 uncertainties (i.e. ∈Xluft) are added separately, following Eq. 13 in Laughner et al. (2024), here Eq. S1: ϵXgas =√ϵXgas.std 2+|∈Xluft|, (S1) Unknown systematic effects are those identified through comparisons with independent measurements, such as WMOreferenced in situ AC profiles. These effects are corrected network-wide using the in situ correction outlined in Sec. 8.3 of Laughner et al. (2024). 115 [AC.Xgas]: For the integrated AC-derived total-column quantity (see Eq. 2, main paper), the primary sources of uncertainty arise from: 1. Unmeasured atmospheric sections (Geibel et al., 2012; Laughner et al., 2024; Messerschmidt et al., 2011; Wunch et al., 2010). 2. The AirCore measurements uncertainty. 120 The magnitude of these uncertainties is largely gas dependent. For point 1) we account for the unmeasured stratospheric and surface levels, and this calculation aims in assessing how the in situ integrated AC.Xgas will be affected by the assumptions made to extend it. Given that our in situ profiles extend up to the lower stratosphere, the most significant unmeasured part is the upper stratosphere. While Messerschmidt et al. (2011) accounted for ~2.02 ppm stratospheric uncertainty for AC.XCO2 by not 125 measuring any part of the stratosphere, our stratospheric uncertainty should be considerably less. We quantify stratospheric uncertainty (∈strat) following Sec. C6 of Laughner et al. (2024), by calculating the difference between a total column-integrated perturbed and unperturbed profile (see Eq. C10 in Laughner et al. (2024)). The perturbed profile is constructed by adding twice the difference between the top AC profile gas value measurement and the corresponding FTS prior level (Eq. C9 in Laughner et al. (2024)). This difference between the AC and the prior value is shown in Fig. S8 130 (left) at the “cutoff altitude”. For the lowest unmeasured section (typically the lowest 0-1500 m, or 1-3 FTS grid levels), we flat-extrapolate the last AC measurement down to 880 m (the 3rd GGG grid level). From the 880 m to the Nicosia FTS level (180 m) we assume the surface gas value has a range from the surface in situ (Picarro, 185 m ASL) mean ± standard deviation, to the last AC measured value. The total column-uncertainty due to this possible range (∈ground) is calculated by integrating profiles where the lowest levels 135 are filled once with the minimum and once with the maximum of this possible range. 16  Partial profile replacement: we replace only three trace-gas profiles (CO2, CH4, and CO) while numerous other profiles – including H2O, temperature, and pressure – remain unchanged 235  Re-gridding smoothing: re-gridding the high-resolution AirCore profiles to the coarser GGG grid levels reduces some of the profile variability  Spatial mismatch propagation: If the FTS and AirCore sampled air masses with genuinely different profiles, using the AirCore instead of the prior does not eliminate error from an incorrect prior shape 240 Conclusion – Implications for comparison: given these considerations, we believe the disagreement on June 29 reflects genuine atmospheric heterogeneity and spatial sampling differences rather than systematic instrumental biases. The fact that: 1) flight 1 shows good agreement for XCO2 while flights 2 and 3 present similarly complex cases (Table 2), 2) the disagreement on flight 2 falls within the combined uncertainties when the individual uncertainties are properly accounted for (all larger in flight 2 compared to flights 1 and 3, see Table S4 for XCO2), 245 3) the custom retrieval does not improve agreement (supporting the spatial mismatch hypothesis); support the interpretation that this case demonstrates the challenges of comparing column and in situ measurements in spatially heterogeneous conditions, rather than indicating a systematic problem with the TCCON Nicosia data. References Geibel, M. C., Messerschmidt, J., Gerbig, C., Blumenstock, T., Chen, H., Hase, F., Kolle, O., Lavrič, J. V., Notholt, J., Palm, 250 M., Rettinger, M., Schmidt, M., Sussmann, R., Warneke, T., and Feist, D. G.: Calibration of column-averaged CH4 over European TCCON FTS sites with airborne in-situ measurements, Atmos. Chem. Phys., 12, 8763–8775, https://doi.org/10.5194/acp-12-8763-2012, 2012. Laughner, J. L., Toon, G. C., Mendonca, J., Petri, C., Roche, S., Wunch, D., Blavier, J.-F., Griffith, D. W. T., Heikkinen, P., Keeling, R. F., Kiel, M., Kivi, R., Roehl, C. M., Stephens, B. B., Baier, B. C., Chen, H., Choi, Y., Deutscher, N. M., DiGangi, 255 J. P., Gross, J., Herkommer, B., Jeseck, P., Laemmel, T., Lan, X., McGee, E., McKain, K., Miller, J., Morino, I., Notholt, J., Ohyama, H., Pollard, D. F., Rettinger, M., Riris, H., Rousogenous, C., Sha, M. K., Shiomi, K., Strong, K., Sussmann, R., Té, Y., Velazco, V. A., Wofsy, S. C., Zhou, M., and Wennberg, P. O.: The Total Carbon Column Observing Network’s GGG2020 data version, Earth Syst. Sci. Data, 16, 2197–2260, https://doi.org/10.5194/essd-16-2197-2024, 2024. Messerschmidt, J., Geibel, M. C., Blumenstock, T., Chen, H., Deutscher, N. M., Engel, A., Feist, D. G., Gerbig, C., Gisi, M., 260 Hase, F., Katrynski, K., Kolle, O., Lavrič, J. V., Notholt, J., Palm, M., Ramonet, M., Rettinger, M., Schmidt, M., Sussmann, R., Toon, G. C., Truong, F., Warneke, T., Wennberg, P. O., Wunch, D., and Xueref-Remy, I.: Calibration of TCCON columnaveraged CO2: the first aircraft campaign over European TCCON sites, Atmos. Chem. Phys., 11, 10765–10777, https://doi.org/10.5194/acp-11-10765-2011, 2011. Wunch, D., Toon, G. C., Wennberg, P. O., Wofsy, S. C., Stephens, B. B., Fischer, M. L., Uchino, O., Abshire, J. B., Bernath, 265 P., Biraud, S. C., Blavier, J.-F. L., Boone, C., Bowman, K. P., Browell, E. V., Campos, T., Connor, B. J., Daube, B. C., Deutscher, N. M., Diao, M., Elkins, J. W., Gerbig, C., Gottlieb, E., Griffith, D. W. T., Hurst, D. F., Jiménez, R., Keppel-Aleks, G., Kort, E. A., Macatangay, R., Machida, T., Matsueda, H., Moore, F., Morino, I., Park, S., Robinson, J., Roehl, C. M., Sawa, 17 Y., Sherlock, V., Sweeney, C., Tanaka, T., and Zondlo, M. A.: Calibration of the Total Carbon Column Observing Network using aircraft profile data, Atmos. Meas. Tech., 3, 1351–1362, https://doi.org/10.5194/amt-3-1351-2010, 2010. 270