Enhanced Air–Sea Exchange of Heat and Carbon Dioxide Over a High Arctic Fjord During Unstable Very-Close-to-Neutral Conditions
Eddy-covariance measurements made in the marine atmospheric boundary layer above a high Arctic fjord (Adventfjorden, Svalbard) are analyzed. When conditions are unstable, but close to neutral −0.1 < z/L < 0, where z is the height, and L is the Obukhov length, the exchange coefficient for sensible heat CH is significantly enhanced compared with that expected from classical surface-layer theory. Cospectra of the vertical velocity component (w) and temperature (T) reveal that a high-frequency peak develops at f ≈ 1 Hz for z/L > − 0.15. A quadrant analysis reveals that the contribution from downdrafts to the vertical heat flux increases as conditions become close to neutral. These findings are the signature of the evolving unstable very-close-to-neutral (UVCN) regime previously shown to enhance the magnitude of sensible and latent heat fluxes in the marine surface layer over the Baltic Sea. Our data reveal the significance of the UVCN regime for the vertical flux of the carbon dioxide (CO2) concentration (C). The cospectrum of w and C clearly shows how the high-frequency peak grows in magnitude for z/L > − 0.15, while the high-frequency peak dominates for z/L > − 0.02. As found for the heat flux, the quadrant analysis of the CO2 flux shows a connection between the additional small-scale turbulence and downdrafts from above. In contrast to the vertical fluxes of sensible and latent heat, which are primarily enhanced by the very different properties of the air from aloft (colder and drier) during UVCN conditions, the increase in the air–sea transfer of CO2 is possibly a result of the additional small-scale turbulence causing an increase in the water-side turbulence. The data indicate an increase in the gas-transfer velocity for CO2 for z/L > − 0.15 but with a large scatter. During the nearly 2 months of continuous measurements (March–April 2013), as much as 36% of all data are associated with the stability range −0.15 < z/L < 0, suggesting that the UVCN regime is of significance in the wintertime Arctic for the air–sea transfer of heat and possibly also CO2.
KeywordsAir–sea exchange Arctic CO2 Gas-transfer velocity Unstable very-close-to-neutral
2.1 Formation of the Convective Boundary Layer
2.2 Gas-Transfer Velocity
Here, B is the buoyancy flux at the sea surface, and zml denotes the mixed-layer depth in water. We follow the scaling concept presented in Andersson et al. (2017), and use Eqs. 6 and 7 to calculate the contribution from water-side convection to the value of k660.
2.3 Transfer Coefficient
2.4 Spectral Analysis
2.5 The Unstable Very-Close-to-Neutral Regime
During moderate instability, the boundary layer is characterized by organized longitudinal eddies of roll-type structure (e.g. Mason and Sykes 1982), giving a normalized spectral peak located at f ≈ 10−2 Hz, and the MOST approach is valid. As the boundary layer becomes closer to neutral (−L > 150 m) due to an increase in wind speed or a decrease in the air–sea temperature difference (ΔT), the small-scale turbulence dominates the spectrum of temperature (see Smedman et al. 2007a, b). This new regime is denoted as the unstable very-close-to-neutral (UVCN) regime, and is characterized by a development of a secondary peak at higher frequencies in the spectrum of T and in the cospectrum of wT. Simultaneously, the value of CH increases with increasing wind speed, and MOST becomes no longer valid. From a quadrant analysis of the value of wT, the additional small-scale turbulence coincides with an increase in the contribution to the heat flux from downdrafts of cold air from layers aloft. In the transition between the more unstable branch characterized by large-eddy structures and the UVCN regime, an intermediate range is found at the intersection of these two regimes characterized by “camel-shaped” spectra and cospectra with two distinct maxima: a large-scale maximum in the energy-containing range related to local production, and a maximum at higher frequency related to the UVCN regime. Smedman et al. (2007a) argued that this small-scale turbulence is explained by detached eddies created by shear in the upper part of the boundary layer. The concept of detached eddies for high Reynolds numbers, which was explained in Hunt and Morrison (2000) and Hunt and Carlotti (2001), and later confirmed by measurements (Högström et al. 2002), says that, as detached eddies move downwards, they are deformed due to blocking and stretching by the surface-layer wind shear. While the UVCN regime is found to occur for approximately −L > 150 m, Smedman et al. (2007a) speculated that the value of h/L, with h the boundary-layer height, would be the controlling parameter, but did not possess any measurements of h. In Sahlée et al. (2008a), the additional turbulence typical of the transition regime was found to significantly increase the vertical turbulent flux of humidity.
3 Site and Measurements
3.1 The Adventpynten Site
During the period from 7 March to 22 April 2013, a field campaign was conducted in the area of Adventfjorden close to Longyearbyen, Svalbard in Norway (Fig. 1). Adventfjorden is a typical, high Arctic fjord where the valley opens out into the water, giving a fjord surrounded by steep mountains. However, the terrain at the Adventpynten site (see Fig. 1a) is relatively flat, resulting in a smooth transition from land to water. As a side fjord to the larger Isfjorden, Adventfjorden is about 7 km long, and the distance across the fjord from Adventpynten to the other side of the fjord is about 3.5 km. The site contains (Fig. 1b) one tower equipped with an eddy-covariance system installed at 3-m height above mean sea level, and a second tower equipped with slow-response instruments measuring the wind speed, temperature, and humidity at two heights (0.5 m and 4 m above ground). The eddy-covariance system consists of one CSAT3 sonic anemometer (Campbell Scientific, North Logan, Utah, USA) measuring the three velocity components and temperature, and a LICOR-7500A gas analyzer (LI-COR Inc., Lincoln, Nebraska, USA) measuring the humidity, CO2 concentration, and pressure. On five occasions during the period 14–25 March, measurements from a boat were taken within the flux footprint. A net radiometer (CNR-1, Kipp & Zonen, Delft, The Netherlands) installed on a boom at the front of the boat measured the net radiation over water. Measurements of the sea-surface temperature, partial pressure of CO2, and the water salinity were also recorded, and vertical profiles of water temperature were taken every 15 min using a conductivity, temperature, and depth sensor (CTD, SeaBird SBE 19plus V2 SeaCat, Seabird Electronics Inc., Bellevue, Washington, USA).
3.2 Data Analysis
Based on a footprint analysis and spectral/cospectral evaluations (Andersson et al. 2017), data not associated with a wind direction in the range of 080°–150° were discarded to minimize the influence of land. A double rotation was performed on the eddy-covariance velocity data, with vectors first rotated into the mean wind direction and then tilt corrected to yield a zero mean vertical velocity component, giving a velocity vector aligned with the mean wind direction. The gas-analyzer data were screened with a filter using the mean concentrations of humidity and CO2 to exclude data with icing on the device. Data were then despiked and divided into 30-min blocks, with data not fulfilling the skewness and flatness criteria of Vickers and Mahrt (1997) discarded. Every individual block of data was linearly detrended and corrected for time lags caused by the separation distance between the sonic anemometer and the gas analyzer. Corrections for density fluctuations due to heat and moisture fluxes were made based on Webb et al. (1980). For the spectral and cospectral analysis of CO2, the density correction was performed directly on the raw signal following the method presented in Sahlée et al. (2008c), where the output signal was transformed into a mixing ratio to avoid the influence from temperature and humidity variations on the shape of the wC cospectrum. Here, cospectra consist of 21 points, with each point representing a normalized mean cospectral estimate for the respective frequency. The data and site are described in detail in Andersson et al. (2017).
4.1 Heat Transfer
4.2 Spectral Analysis
4.3 Quadrant Analysis
Figure 7 shows a quadrant plot for a moderately unstable case with L = − 5 m (red) and a UVCN situation with L = − 300 m (black), which is also shown in Fig. 5 (green). For the moderately unstable case, the majority of the flux is constrained to quadrant I, implying that the flux originates from an upwards transfer of warm air. The case related to L = − 300 m displays a different behaviour, with a large part of the positive contribution to the heat flux originating from quadrant III, corresponding to a downwards transfer of cold air from above. A closer inspection of the case related to L = − 300 m shows that for H = 4, the sum of the flux fraction is 0.66 and S3,4 = 0.25, meaning that 66% of the flux occurs in events where the flux is more than four times the average value, and where the flux fraction from quadrant III is 25%. For the more unstable case related to L = − 5 m with H = 4, the sum of the flux fractions is 0.49, and the flux fraction from quadrant III is below 3%.
4.4 Implications for the Air–Sea Gas Transfer of CO2
Section 4.2 shows that the vertical turbulent flux of temperature is enhanced as conditions become neutral from the unstable side. As mentioned earlier, the enhanced fluxes of temperature and humidity during UVCN conditions reported in Smedman et al. (2007a, b) and Sahlée et al. (2008a, b) are the result of the different properties (colder and drier) of the air aloft, which is transported downwards by detached eddies. The quadrant analysis of the heat flux presented in Fig. 7 also shows an enhanced contribution from downdrafts of cold air for the UVCN regime, which possibly enhance the magnitude of the flux wT, and cause the increase in the value of CH for this regime. For CO2, while we do not expect as large differences in concentration between the upper and lower part of the atmospheric surface layer as for temperature, it is possible that the imprints of the additional small-scale turbulence increase the magnitude of the water-side turbulence, and thereby the transfer velocity of the air–sea transfer of CO2. In Fig. 8, a quadrant analysis is presented for three cases with L ≈ − 6 m (red dot), L ≈ − 30 m (black stars) and L = − 150 m (black dots), with the positive contributions to the measured downwards flux Fc from quadrants II and IV. Similar to the quadrant analysis of the cospectrum wT (Fig. 7), the increasing significance of downdrafts when approaching the UVCN regime are also observed for the flux Fc (here quadrant IV) when L = − 150 m.
As conditions became more neutral and z/L > − 0.1 (see the black dashed line in Fig. 10), a different behaviour is also observed for the gas-transfer velocities of CO2; the scatter decreases and as much as 80% of the data attain values k660 − (kwsc + kW09) > 10 cm h−1. The data within this regime are characterized by larger values of u*, resulting in enhanced water-side turbulence, and an enhanced contribution of the CO2 flux from downdrafts of air with higher concentrations of CO2, such as the case L = − 150 m (green circle), which is also presented in Fig. 8 (black dots) and Fig. 9 (green curve).
5 Summary and Conclusions
In winter, a CBL develops over the high Arctic fjord where the height of the boundary layer increases with increasing distance from land. The combination of a large air–sea temperature gradient and relatively high wind speed results in a well-mixed boundary layer with stability −1 < z/L < − 0.03. During conditions with stability −0.2 < z/L < 0, measurements show enhanced vertical fluxes of heat not associated with the local gradient. From being relatively constant for z/L < − 0.15, the bulk transfer coefficient increases from a typical mean value of CH = 0.0015 to CH = 0.0018 for z/L = − 0.075. The increase of the bulk transfer as a function of wind speed is not as prominent, which is probably a result of the large variation in temperature gradient. The cospectra of wT show that the enhanced heat transfer is associated with small-scale turbulence with a peak at around f = 1 Hz, which gradually grows in magnitude as conditions became more neutral, and the relative contribution from the low-frequency peak at f = 0.003 Hz decreases in magnitude. For L = − 300 m, the high-frequency maximum dominates the normalized cospectrum of wT. Simultaneously, the contribution from downdrafts to the total heat flux increases from 3% for L = − 5 m to 25% of the flux for L = − 300 m. This is the signature of the UVCN regime presented in a series of papers.
Small-scale turbulence, which is potentially caused by detached eddies, also affects the air–sea exchange of CO2. A quadrant analysis of the CO2 flux shows that, as conditions approach neutral and the Obukhov length decreases from L = − 30 m to L = − 150 m, the contribution to the CO2 flux from downdrafts of air with higher concentrations of CO2 increases. Similar to the flux wT, the increased contribution from downdrafts to the air–sea CO2 flux is found to coincide with the formation and increase in a high-frequency peak in the cospectra for wC. After removing the effect from the two dominant processes affecting the gas-transfer velocity using published parametrizations of water-side convection (kwsc) and shear-induced turbulence from the mean wind speed (kW09), the gas-transfer velocity was examined, showing the values of k660 − (kW09 + kwsc) scattered around zero for moderately unstable conditions. However, when approaching the UVCN regime (z/L > − 0.1 and U > 6.5 m s−1), the values of k660 − (kW09 + kwsc) are found to increase, where as much as 80% of the data take values k660 − (kW09 + kwsc) > 0.1 m h−1. These results are somewhat surprising since one does not expect the vertical gradient of CO2 in the atmosphere to be as large as the vertical gradient for temperature. We speculate that a part of the enhanced CO2 flux is due to imprints on the water surface by the additional small-scale turbulence resulting in increased levels of water-side turbulence.
In summary, we show the possible importance of the UVCN regime for the air–sea exchange of heat in the Arctic during winter, and that the UVCN regime affects the air–sea exchange of CO2, becoming significant for stabilities z/L ≥ –0.1. During the nearly 2 months of continuous measurements, as much as 30% of the data were associated with conditions −0.1 ≤ z/L < 0. We believe that a larger dataset of gas-transfer velocities would emphasize the relevance of the UVCN regime for air–sea gas exchange.
The authors wish to thank Tor de Lange at the Geophysical Institute, University of Bergen, Norway for his useful support and technical assistance during the fieldwork. Data supporting the results will be provided upon request.
- Andersson A, Falck E, Sjöblom A, Kljun N, Sahlée E, Omar AM, Rutgersson A (2017) Air–sea gas transfer in high Arctic fjords. J Geophys Lett. https://doi.org/10.1002/2016gl072373
- Bell TG, Landwehr S, Miller SD, de Bruyn WJ, Callaghan AH, Scanlon B, Ward B, Yang M, Saltzman ES (2017) Estimation of bubble-mediated air–sea gas exchange from concurrent DMS and CO2 transfer velocities at intermediate–high wind speeds. Atmos Chem Phys 17:9019–9033. https://doi.org/10.5194/acp-17-9019-2017 CrossRefGoogle Scholar
- Else BGT, Papakyriakou TN, Galley RJ, Drennan WM, Miller LA, Thomas H (2011) Wintertime CO2 fluxes in an Arctic polynya using eddy covariance: Evidence for enhanced air–sea gas transfer during ice formation. J Geophys Res Oceans 116(C9)Google Scholar
- Garratt JR (1992) The atmospheric boundary layer. Cambridge University Press, CambridgeGoogle Scholar
- Le Quéré C, Moriarty R, Andrew RM, Peters GP, Ciais P, Friedlingstein P, Jones SD, Sitch S, Tans P, Arneth A, Boden TA, Bopp L, Bozec Y, Canadell JG, Chini LP, Chevallier F, Cosca CE, Harris I, Hoppema M, Houghton RA, House JI, Jain AK, Johannessen T, Kato E, Keeling RF, Kitidis V, Klein-Goldewijk K, Koven C, Landa CS, Landschützer P, Lenton A, Lima ID, Marland G, Mathis JT, Metzl N, Nojiri Y, Olsen A, Ono T, Peng S, Peters W, Pfeil B, Poulter B, Raupach MR, Regnier P, Rödenbeck C, Saito S, Salisbury JE, Schuster U, Schwinger J, Séférian R, Segschneider J, Steinhoff T, Stocker BD, Sutton AJ, Takahashi T, Tilbrook B, van der Werf GR, Viovy N, Wang Y-P, Wanninkhof R, Wiltshire A, Zeng N (2015) Global carbon budget 2014. Earth Syst Sci Data 7(1):47–85. https://doi.org/10.5194/essd-7-47-2015 CrossRefGoogle Scholar
- MacIntyre S, Eugster W, Kling GW (2002) The critical importance of buoyancy flux for gas flux across the air–water interface, in: Donelan MA, Drennan WM, Saltzman ES, Wanninkhof R (eds) Gas Transfer at Water Surfaces, Geophysical Monograph, American Geophysical Union, Washington DC, vol 127, pp 13–28Google Scholar
- Smedman A, Högström U, Sahlée E, Johansson C (2007b) Critical reevaluation of the bulk transfer coefficient for sensible and latent heat over the ocean during unstable and neutral conditions. Q J R Meteorol Soc 1:227–250Google Scholar
- Willmarth WW, Lu SS (1974) Structure of the Reynolds stress and the occurrence of bursts in the turbulent boundary layer. Adv Geophys 18A:287–314Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.