Comment on the preprint

“Particle flux-gradient relationships in the high Arctic: Emission and deposition patterns across three surface types”

by Piotr Markuszewski^1,2 and Monica Mårtensson²

1. Institute of Oceanology, Polish Academy of Sciences
2. Stockholm University
3. Uppsala University

This manuscript presents a well-executed field study investigating near-surface aerosol particle fluxes and sensible heat fluxes over different Arctic surface types using a novel gradient-based measurement system during the ARTofMELT expedition. The comparison with eddy covariance data enhances the credibility of the measurements. The study provides valuable insight into particle source/sink behavior over wide leads, narrow leads, and closed ice, contributing to our understanding of aerosol-cloud-sea-ice interactions in the central Arctic.

The paper is timely, methodologically sound, and clearly structured, but there are several areas where the manuscript could be improved for clarity, completeness, and scientific robustness.

The manuscript would benefit from a significantly broader and more critical engagement with the prior literature on gradient-based flux measurements, particularly over marine and ice surfaces. While the authors cite several key studies related to Arctic fluxes (e.g., Nilsson et al. 2001, Held et al. 2011a,b), the discussion omits earlier foundational work using gradient methods to estimate aerosol and heat fluxes in polar and marine environments.

For instance, the first gradient-based aerosol flux measurements conducted by Petelski (2003), Petelski et al. (2005), and further refined in Petelski and Piskozub (2006), including the later comment given by Andreas (2007), are highly relevant and should be referenced. Additionally, Savelyev et al. (2014) addressed fluxes under low-turbulence regimes using similar profile techniques, which is particularly relevant for the stable and weakly turbulent conditions encountered over closed ice. Authors may also find recent publication dedicated to the gradient method (Markuszewski et al., 2024).

In its current form, the manuscript gives the impression that the gradient method is underexplored in this field, which is not accurate. A proper contextualization would strengthen the justification for the study, allow a more meaningful comparison of uncertainties and assumptions, and acknowledge the methodological evolution within the field of flux-gradient applications.

The manuscript lacks of presentation of the raw or processed aerosol concentration profiles that underpin the flux-gradient calculations. Since the fluxes are derived from linear regressions across vertical gradients of particle number concentrations, it is essential to show representative examples of these vertical profiles to assess the validity of the method. Including a few example profiles—either in the main text or as supplementary material—would serve as a critical demonstration that the gradient system performs as intended. For instance, plots showing aerosol number concentrations at each height, along with fitted regression lines and R² values, would give readers confidence in the robustness of the derived fluxes. It is also unclear what the range of correlation coefficients (e.g., R² of the linear fit) was across the dataset, or how often profiles were rejected due to poor fits. Providing a histogram or table of regression diagnostics (e.g., slope, R², residuals) would help clarify the quality and reliability of the profiles used in flux calculations. Furthermore, it would be useful to include a discussion of profile curvature, measurement noise, or transient concentration spikes, and how these were addressed during preprocessing and averaging. Without this level of transparency, the central assumption—that vertical particle concentration gradients are well-defined and resolved—is insufficiently supported.

Another methodological weakness stems from the limited footprint characterization and the difference between systems used for intercomparison. Yet, the paper compares their fluxes as though they were co-located. While some differences are acknowledged qualitatively, there is no attempt to assess or estimate the footprints, nor to indicate whether the observed discrepancies fall within expected spatial variability. The authors should include either a footprint model (even a simplified 1D footprint estimate based on surface roughness and stability) or a discussion of fetch dependence to justify the comparability of the datasets. This is particularly important for interpreting disagreements in flux direction and magnitude over mixed or narrow lead surfaces.

The uncertainty estimation methodology, though commendable in its use of Monte Carlo simulations, underrepresents potential systematic errors. For example, the correction for tubing losses is based on size distribution measurements from a ship-based DMPS located several meters above the sea surface, which may not accurately represent the near-surface environment sampled by the gradient system. Variability in vertical gradients of particle size, especially under stratified conditions or during local emissions, could lead to an incorrect penetration fraction estimate and a biased flux. A sensitivity analysis showing how variations in assumed size distributions affect the loss correction would greatly improve transparency. Additionally, more details on the impact of inlet length, orientation, and isokinetic sampling conditions should be included to assess sampling biases under varying wind regimes.

The use of Monte Carlo simulations to estimate uncertainty in particle and sensible heat fluxes is a commendable choice; however, the manuscript lacks a sufficiently detailed explanation of how this method was applied specifically in the context of aerosol flux measurements. In aerosol science, uncertainty propagation is particularly sensitive to input data variability, non-linearity in particle losses, and the low signal-to-noise ratios often encountered in Arctic conditions. The authors state that 10,000 random values were generated per profile, but it is unclear whether the distributions used were strictly Gaussian, whether input variances were assumed to be independent across heights, or how temporal autocorrelation in measurement noise was handled. Furthermore, the choice to summarize daily uncertainties using the 90th percentile of simulated fluxes lacks justification—was this percentile empirically chosen, or based on prior studies? Overall, a clearer explanation of the statistical assumptions, potential bias sources, and limits of the method’s sensitivity is needed to allow readers to evaluate how well the simulation captures real-world variability and uncertainty in the derived fluxes.

The comparison of friction velocity estimates between the gradient system and eddy covariance (Figure 2) shows substantial scatter, yet the potential causes of these discrepancies are not sufficiently explored. In particular, it is unclear whether inertial motion correction was applied to the eddy covariance measurements on the ship mast. Motion-induced errors are known to bias vertical velocity estimates on moving platforms, and the absence of an inertial measurement unit (IMU) or equivalent correction could lead to systematic over- or underestimation of u*, especially under low turbulence. I recommend that the authors clarify whether motion correction was applied and, if not, discuss this as a potential limitation. In addition, the eddy covariance post-processing methods (e.g., coordinate rotation, averaging strategy) should be described more explicitly, and their influence on the u* estimates should be assessed. Without such information, the comparability of the methods remains questionable, particularly over complex surfaces like narrow leads where the surface heterogeneity is likely to exacerbate footprint mismatches.

In terms of data analysis, the manuscript occasionally uses ambiguous terminology when referring to “net emission” and “net deposition”. It is not always clear whether these terms refer to statistically significant fluxes (i.e., exceeding uncertainty bounds) or to the algebraic sign of the estimated flux. In several figures, flux values that lie within the uncertainty range are still color-coded as deposition or emission. I suggest the authors define a threshold for meaningful flux detection (e.g., based on the daily maximum error estimate) and clearly distinguish between statistically significant fluxes and those near the noise level. This would reduce the risk of overinterpreting marginal cases.

Throughout the manuscript, the authors report both absolute particle fluxes in m⁻²·s⁻¹ and normalized fluxes in cm·s⁻¹, which is a standard approach in aerosol micrometeorology. However, the manuscript lacks a clear explanation of why and when each form is used. The definition of normalized flux (V_D = –P/C) is provided, but the rationale for expressing it in [cm/s] rather than [m/s] is not discussed. This unit switch can be confusing, particularly since figure axes and captions do not always indicate units explicitly. I recommend that the authors (1) clearly define both flux formats early in the Methods or Data Analysis section, (2) justify the choice of units for normalized flux, and (3) ensure that all figure axes and table entries explicitly label the flux units used. A brief explanation of the comparability advantage of normalized fluxes (especially under variable concentration regimes) would also strengthen the interpretation of results.

While the authors interpret many of the observed concentration and flux patterns in terms of local surface type (wide lead, narrow lead, closed ice), they do not apply any trajectory analysis to objectively evaluate the origin of the sampled air masses. This is a significant limitation. In polar environments, long-range transport can dramatically influence background particle concentrations, size distributions, and chemical composition. The manuscript refers to air mass changes (e.g., "southerly winds bringing humid air") and suggests terrestrial influence from Greenland or Svalbard, but these assertions are speculative without the support of backward trajectory modeling (e.g., HYSPLIT, FLEXPART). Including at least a qualitative or cluster-based trajectory analysis would strengthen the interpretation of aerosol variability and help distinguish between local emission/deposition processes and transported signals. I recommend the authors include such an analysis or, at minimum, acknowledge this limitation in the discussion section.

In summary, this manuscript presents an important contribution to the field of Arctic micrometeorology and aerosol flux measurements. The deployment of a mobile gradient flux system under challenging field conditions, combined with extensive observational data across varied surface types, offers valuable insights into particle exchange processes in the central Arctic. However, to ensure scientific rigor and broader relevance, the manuscript requires substantial revision. Key areas for improvement include deeper engagement with prior literature on gradient methods, clarification of methodological details (particularly around uncertainty estimation and regression quality), and more cautious interpretation of flux patterns in the absence of chemical or trajectory data. Provided that these issues are thoroughly addressed, I believe the paper will make a strong and meaningful contribution and would ultimately be worthy of publication.

We fully understand that this is a comment in the discussion and that further development of the manuscript may already be underway. Nevertheless, we would be very grateful if the authors would consider addressing at least some of the issues and suggestions raised here. We believe that doing so would significantly enhance the clarity, robustness, and scientific value of this already promising contribution.

Piotr Markuszewski and Monica Mårtensson

References

Andreas, E. L.: Comment on “Vertical coarse aerosol fluxes in the atmospheric surface layer over the North Polar Waters of the Atlantic” by Tomasz Petelski and Jacek Piskozub, J. Geophys. Res.-Oceans, 112, C11, https://doi.org/10.1029/2007JC004191, 2007.

Held, A., Brooks, I. M., Leck, C., and Tjernström, M.: On the potential contribution of open lead particle emissions to the central Arctic aerosol concentration, Atmos. Chem. Phys., 11, 3093–3105, https://doi.org/10.5194/acp-11-3093-2011, 2011a.

Held, A., Orsini, D. A., Vaattovaara, P., Tjernström, M., and Leck, C.: Near-surface profiles of aerosol number concentration and temperature over the Arctic Ocean, Atmos. Meas. Tech., 4, 1603–1616, https://doi.org/10.5194/amt-4-1603-2011, 2011b.

Markuszewski, P., Nilsson, E. D., Zinke, J., Mårtensson, E. M., Salter, M., Makuch, P., and Piskozub, J.: Multi-year gradient measurements of sea spray fluxes over the Baltic Sea and the North Atlantic Ocean, Atmos. Chem. Phys., 24, 11227–11253, https://doi.org/10.5194/acp-24-11227-2024, 2024.

Nilsson, E. D., Rannik, Ü., Swietlicki, E., Leck, C., Aalto, P. P., Zhou, J., and Norman, M.: Turbulent aerosol fluxes over the Arctic Ocean: 2. Wind-driven sources from the sea, J. Geophys. Res.-Atmos., 106, 32139–32154, https://doi.org/10.1029/2000JD900747, 2001.

Petelski, T.: Marine aerosol fluxes over open sea calculated from vertical concentration gradients, J. Aerosol Sci., 34, 359–371, https://doi.org/10.1016/S0021-8502(02)00191-9, 2003.

Petelski, T., and Piskozub, J.: Vertical coarse aerosol fluxes in the atmospheric surface layer over the North Polar Waters of the Atlantic, J. Geophys. Res.-Oceans, 111, C06039, https://doi.org/10.1029/2005JC003295, 2006.

Petelski, T., Piskozub, J., and Paplińska‐Swerpel, B.: Sea spray emission from the surface of the open Baltic Sea, J. Geophys. Res.-Oceans, 110, C10023, https://doi.org/10.1029/2004JC002800, 2005.

Savelyev, I. B., Anguelova, M. D., Frick, G. M., Dowgiallo, D. J., Hwang, P. A., Caffrey, P. F., and Bobak, J. P.: On direct passive microwave remote sensing of sea spray aerosol production, Atmos. Chem. Phys., 14, 11611–11631, https://doi.org/10.5194/acp-14-11611-2014, 2014.