the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 07 Apr 2025
Lagrangian aerosol particle trajectories in a cloud free marine atmospheric boundary layer: Implications for sampling
Abstract. Meteorological processes such as gust fronts, roll structures, internal boundary layer development, and smaller scale turbulence complicate the physical interpretation of measured aerosol particle properties, fluxes, and transport in the marine atmospheric boundary layer (MABL). To better decipher maritime aerosol measurements by aircraft, ships, and towers we describe an ensemble of particle trajectories using high resolution large eddy simulations (LES) of surface-emitted aerosol particle within a Lagrangian framework. We identified two clusters of particle trajectory types from which we created probabilistic distributions of particle histories: a) short lived particles that do not exit the surface layer and are subsequently deposited back to the ocean; and b) much older particles that are able to exit the surface layer into the mixed layer and subsequently oscillate up and down through convective roll structures. After emission in a neutral atmosphere, particles slowly disperse through the MABL requiring, on average, up to 100 minutes to mix to the ~570 m deep mixed layer inversion. However, for even slightly unstable conditions, particles are rapidly transported to the top of the MABL in roll structure updrafts, where they then more slowly diffuse downwards, with some similarities to a looping plume rise to the stable inversion followed by fumigation. Consequently, particles can exhibit a bimodal lifetime distribution that results in different particle ages by altitude. Further, based on wind speed and stability, the initial looping behavior following an emission event spans 15 to 30 minutes and may result in sampling “blind spots” up to 15 km downwind. Overall, our findings suggest that there should be a consideration of the representativeness of particle ages, even in what is often assumed to be a well-mixed MABL. This representativeness is related to how long particles have been suspended and whether they were sourced locally, which is critical for situations such as for measuring wind generated emissions or ship track plumes. Further, the Lagrangian technique for treating the particle transport captures the inherently random motion of the MABL turbulence and does not exhibit artificial numerical diffusion. As such, it produces differences when compared to a traditional, column-based eddy-diffusivity approach used in mesoscale to global scale models. We used the LES to drive a 1D column model to approximate single grid point physics. The results were starkly different near the surface, with the 1 D column model missing the looping behavior and showing a slow upward dispersion. This difference in the 1D and LES frameworks is an excellent example of sub-grid problems and may explain some of the differences between observations and global and meso-scale model simulations of marine particle vertical distribution and dry scavenging.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(4166 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(4166 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-576', Steven Howell, 16 May 2025
- AC1: 'Reply on RC1', Jeffrey Reid, 03 Jul 2025
-
RC2: 'Comment on egusphere-2025-576', Anonymous Referee #2, 19 May 2025
Overall: This is an interesting study with valuable insights into particle motion throughout the MABL. It points out many relevant and meaningful gaps in existing SSA research related to the transport of these giant particles from the surface throughout the MABL. I think it represents novel research that deserves to be published. Below, I've posted a few questions to consider that I think could be meaningful discussions/alterations to the manuscript.ÂÂMethods:ÂI understand the method releases a singular aerosol plume event to observe the evolution of these particles throughout the LES, however, I wonder if you continually released particles and allowed them to loop through the periodic boundary conditions if the mixing eventually reaches some sort of homogeneity? In the intro, you preface that this research could impact/provide insights into proposed in situ sampling strategies and that something like cloud streets might have some impact on where whitecaps are being generated. While SSA are produced by these plumes, they exist within a background concentration of previously produced aerosol. If you were to continually release these particles throughout the whole domain (or say, at a length scale of every whole number of T_eddy/T_neut to represent increased production at these eddy lengths), how prevalent are these "near-absence of particles near the surface" (line 387)? Is the focus on the lowest surface layer in this statement? If so, I believe this can be made more clear.ÂÂAn additional question I have is what percentage of these particles exist within the primary (trapped at the surface) vs. the secondary (entrained into the MABL) modes? If you have statistics on this, those are really novel and particularly interesting in understanding how many of these giant particles are functionally being entrained.ÂÂSection 2.2: What is the equivalent dry particle diameter/radius for this 10 micron aerodynamic diameter particle? Is it dry? If it's dry, a small discussion on how these giant SSA may act when their particles are deliquesced (say at different altitudes, with differing RH, within the MABL) would be interesting. These particles will grow and shrink through these many eddy cycles. How applicable are your PDFs of residence times at different regions in the MABL (Fig. 7) when these particles experience dynamic fall velocities as a function of changing size due to RH changes? Also, if your particles are dry, what is the equivalent sized particle in situ that this 10 um aerodynamic diameter size represent for an SSA particle at formation diameter (d_99), which is ~4x the diameter with a mass that is mostly water (~half the density of salt)? This has huge implications for in situ scientists trying to gauge which size this is relevant for. A quick back of the envelope estimate is that the particles in this study are likely representative of a < 3 um particle at formation diameter, which after undergoing evaporation outside the surface layer, become exceptionally well mixed throughout the MABL and aren't subject to much removal through dry deposition.ÂÂLines 346-350: Yes, the particles in the unstable case go through more cyclic motions in the 6000 s time, but isn't this just a function of the eddy roll lengths determined by the MABL stability? Say, if you were to generate the same graph as Figure 4 but with a normalized X-axis to T_eddy/T_neut, with Time (s) varying on the upper X-axis, would these graphs look much different from each other? Time is a valid means of presenting the research, but I wonder how different these results would look if presented from an eddy roll length perspective. For example, Figure 5 demonstrates the same " aerosol at approximately 2.5 T_eddy/T_neut for all three conditions (neutral, slightly unstable, and unstable) (highlighted).ÂÂOverall Writing and Formatting: The writing is really great, which is interesting, because it's in direct contradiction to all of the copy-editing that needs to be done on this manuscript. The lack of attention to minor details in the writing as well as figures generates detracts from the paper. The formatting of the many unit notations is inconsistent across the paper (italicized, non-italicized, inconsistent spacing). I will try to provide some proposed copy-edits and inconsistencies I saw in the attached PDF, but ultimately this manuscript should be copy-edited professionally before any future submission. Additionally, the citation formatting is all over the place - I can't tell if these have been manually input or why there's such inconsistencies, but these need to be non-italicized and formatted properly, linking to their citations at the end of the paper. There's many things that need to be improved to comply with Copernicus publishing standards, including colorblind friendly figures and increasing the dpi of some figures (why does Figure 5 have a distorted aspect ratio?) - https://publications.copernicus.org/for_authors/manuscript_preparation.html#figurestablesCitation: https://doi.org/
10.5194/egusphere-2025-576-RC2 - AC2: 'Reply on RC2', Jeffrey Reid, 03 Jul 2025
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-576', Steven Howell, 16 May 2025
This paper was a pleasure to read. The results are not particularly profound or unexpected, but are a very nice presentation of how to think about particle trajectories and histories. The writing is clear and straightforward and the graphics are generally well chosen to get the points across. Most of my thoughts while reading were about how nice it would have been to extend the work farther rather than about whether I trusted the reasoning. I don't actually have a lot of substantive comments.
I was a bit confused by figure 6. The legend describes panels b through d as "the time-height evolution of the column count normalized deposited particle concentration"  whereas the text described those plots as "the probability of a deposited particle’s maximum altitude reached, given a particular lifetime". I don't understand the first phrasing at all, but the second makes some sense to me.
I was surprised not to see explicit mention of an apparent pattern that the slightly unstable plots would look a lot like the unstable ones if the x-axis were Teddy. Â Yes, Figure 8 shows that some details would differ, but the overall impression is that characterizing mixing in the MBL simply with Teddy would be useful.
I'm trying to think of to whom this information would be really valuable. After all, a sudden release of low-altitude 10 µm particles is not a common occurrence! I wind up thinking about time scales of mixing: how long after a front passage or scavenging event (a rain storm) do I have to wait before I can assume that aerosol in the mixed layer is well-mixed? How often is a a particle likely to have encountered clouds in stratocumulus or trade wind cumulus regimes? What do I need to know to make those estimations? These questions would be relevant to sampling expeditions or modeling.
Line 163--4: "The particle sizes are set to 10μm in aerodynamic diameter to represent coarse mode particles, which is much smaller than the smallest turbulence scales of the flow" Well, yes, but any realistic particle size is smaller than the turbulence scale. I expect you're referring to stopping distance for the particle being much smaller than turbulence scales or terminal velocity much less than typical vertical winds.
Equations 2 and 3: Inconsistent use of boldface to indicate vector quantities
Line 210: Is Q* supposed to be Q0?
Line 283: "There is a slight crossover in wind speeds at the top of the mixed layer, with neutral having a lower wind speed at 9.2 m/s, and unstable at 10.7 m/s" That crossover is at something like 30 meters, hardly the top of the mixed layer. Seems to be more at the transition between a surface layer and the bottom of the central mixed layer.
The lower x-axis in Figure 2b is paradoxical. A log scale can't go to zero. Is it linear between -10-4 and +10-4? That would explain the kinks in the blue and orange lines and the smooth passage through 0. Makes it hard to imagine dividing the blue lines by 10. Not sure I know of a better way to present the data though.
Figure 3: It is gorgeous, but since the w' color scale is biased, it looks like there are net downdrafts since a 0.4 m/s downdraft looks just as saturated as a 0.8 m/s updraft. Does it not work with a symmetric color scale, leaving the strongest downdrafts unsaturated?
Figure 6a: It appears that the most probable lifetime is much shorter than the 1000 s you mention. If the data are saved every 5 s, you could have shown even shorter periods at the beginning of the run. Does that not work? I'd be interested to see something like 6a, but with fraction of original particles remaining. It wouldn't have the nice dips in the unstable cases that you point out, but a flattening of the curve, so it wouldn't be as striking a plot, but would be easier to understand.
Line 593: "spatially" is missing the y
Line 627: "Even for neutral conditions, it can take over 90 minutes" implies that unstable conditions take longer! Perhaps just ditch the "Even"
Citation: https://doi.org/10.5194/egusphere-2025-576-RC1 - AC1: 'Reply on RC1', Jeffrey Reid, 03 Jul 2025
-
RC2: 'Comment on egusphere-2025-576', Anonymous Referee #2, 19 May 2025
Overall: This is an interesting study with valuable insights into particle motion throughout the MABL. It points out many relevant and meaningful gaps in existing SSA research related to the transport of these giant particles from the surface throughout the MABL. I think it represents novel research that deserves to be published. Below, I've posted a few questions to consider that I think could be meaningful discussions/alterations to the manuscript.ÂÂMethods:ÂI understand the method releases a singular aerosol plume event to observe the evolution of these particles throughout the LES, however, I wonder if you continually released particles and allowed them to loop through the periodic boundary conditions if the mixing eventually reaches some sort of homogeneity? In the intro, you preface that this research could impact/provide insights into proposed in situ sampling strategies and that something like cloud streets might have some impact on where whitecaps are being generated. While SSA are produced by these plumes, they exist within a background concentration of previously produced aerosol. If you were to continually release these particles throughout the whole domain (or say, at a length scale of every whole number of T_eddy/T_neut to represent increased production at these eddy lengths), how prevalent are these "near-absence of particles near the surface" (line 387)? Is the focus on the lowest surface layer in this statement? If so, I believe this can be made more clear.ÂÂAn additional question I have is what percentage of these particles exist within the primary (trapped at the surface) vs. the secondary (entrained into the MABL) modes? If you have statistics on this, those are really novel and particularly interesting in understanding how many of these giant particles are functionally being entrained.ÂÂSection 2.2: What is the equivalent dry particle diameter/radius for this 10 micron aerodynamic diameter particle? Is it dry? If it's dry, a small discussion on how these giant SSA may act when their particles are deliquesced (say at different altitudes, with differing RH, within the MABL) would be interesting. These particles will grow and shrink through these many eddy cycles. How applicable are your PDFs of residence times at different regions in the MABL (Fig. 7) when these particles experience dynamic fall velocities as a function of changing size due to RH changes? Also, if your particles are dry, what is the equivalent sized particle in situ that this 10 um aerodynamic diameter size represent for an SSA particle at formation diameter (d_99), which is ~4x the diameter with a mass that is mostly water (~half the density of salt)? This has huge implications for in situ scientists trying to gauge which size this is relevant for. A quick back of the envelope estimate is that the particles in this study are likely representative of a < 3 um particle at formation diameter, which after undergoing evaporation outside the surface layer, become exceptionally well mixed throughout the MABL and aren't subject to much removal through dry deposition.ÂÂLines 346-350: Yes, the particles in the unstable case go through more cyclic motions in the 6000 s time, but isn't this just a function of the eddy roll lengths determined by the MABL stability? Say, if you were to generate the same graph as Figure 4 but with a normalized X-axis to T_eddy/T_neut, with Time (s) varying on the upper X-axis, would these graphs look much different from each other? Time is a valid means of presenting the research, but I wonder how different these results would look if presented from an eddy roll length perspective. For example, Figure 5 demonstrates the same " aerosol at approximately 2.5 T_eddy/T_neut for all three conditions (neutral, slightly unstable, and unstable) (highlighted).ÂÂOverall Writing and Formatting: The writing is really great, which is interesting, because it's in direct contradiction to all of the copy-editing that needs to be done on this manuscript. The lack of attention to minor details in the writing as well as figures generates detracts from the paper. The formatting of the many unit notations is inconsistent across the paper (italicized, non-italicized, inconsistent spacing). I will try to provide some proposed copy-edits and inconsistencies I saw in the attached PDF, but ultimately this manuscript should be copy-edited professionally before any future submission. Additionally, the citation formatting is all over the place - I can't tell if these have been manually input or why there's such inconsistencies, but these need to be non-italicized and formatted properly, linking to their citations at the end of the paper. There's many things that need to be improved to comply with Copernicus publishing standards, including colorblind friendly figures and increasing the dpi of some figures (why does Figure 5 have a distorted aspect ratio?) - https://publications.copernicus.org/for_authors/manuscript_preparation.html#figurestablesCitation: https://doi.org/
10.5194/egusphere-2025-576-RC2 - AC2: 'Reply on RC2', Jeffrey Reid, 03 Jul 2025
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 753 | 61 | 20 | 834 | 19 | 37 |
- HTML: 753
- PDF: 61
- XML: 20
- Total: 834
- BibTeX: 19
- EndNote: 37
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Hyungwon John Park
Jeffrey S. Reid
Peter F. Caffrey
Maria J. Chinita
David H. Richter
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(4166 KB) - Metadata XML
This paper was a pleasure to read. The results are not particularly profound or unexpected, but are a very nice presentation of how to think about particle trajectories and histories. The writing is clear and straightforward and the graphics are generally well chosen to get the points across. Most of my thoughts while reading were about how nice it would have been to extend the work farther rather than about whether I trusted the reasoning. I don't actually have a lot of substantive comments.
I was a bit confused by figure 6. The legend describes panels b through d as "the time-height evolution of the column count normalized deposited particle concentration"  whereas the text described those plots as "the probability of a deposited particle’s maximum altitude reached, given a particular lifetime". I don't understand the first phrasing at all, but the second makes some sense to me.
I was surprised not to see explicit mention of an apparent pattern that the slightly unstable plots would look a lot like the unstable ones if the x-axis were Teddy. Â Yes, Figure 8 shows that some details would differ, but the overall impression is that characterizing mixing in the MBL simply with Teddy would be useful.
I'm trying to think of to whom this information would be really valuable. After all, a sudden release of low-altitude 10 µm particles is not a common occurrence! I wind up thinking about time scales of mixing: how long after a front passage or scavenging event (a rain storm) do I have to wait before I can assume that aerosol in the mixed layer is well-mixed? How often is a a particle likely to have encountered clouds in stratocumulus or trade wind cumulus regimes? What do I need to know to make those estimations? These questions would be relevant to sampling expeditions or modeling.
Line 163--4: "The particle sizes are set to 10μm in aerodynamic diameter to represent coarse mode particles, which is much smaller than the smallest turbulence scales of the flow" Well, yes, but any realistic particle size is smaller than the turbulence scale. I expect you're referring to stopping distance for the particle being much smaller than turbulence scales or terminal velocity much less than typical vertical winds.
Equations 2 and 3: Inconsistent use of boldface to indicate vector quantities
Line 210: Is Q* supposed to be Q0?
Line 283: "There is a slight crossover in wind speeds at the top of the mixed layer, with neutral having a lower wind speed at 9.2 m/s, and unstable at 10.7 m/s" That crossover is at something like 30 meters, hardly the top of the mixed layer. Seems to be more at the transition between a surface layer and the bottom of the central mixed layer.
The lower x-axis in Figure 2b is paradoxical. A log scale can't go to zero. Is it linear between -10-4 and +10-4? That would explain the kinks in the blue and orange lines and the smooth passage through 0. Makes it hard to imagine dividing the blue lines by 10. Not sure I know of a better way to present the data though.
Figure 3: It is gorgeous, but since the w' color scale is biased, it looks like there are net downdrafts since a 0.4 m/s downdraft looks just as saturated as a 0.8 m/s updraft. Does it not work with a symmetric color scale, leaving the strongest downdrafts unsaturated?
Figure 6a: It appears that the most probable lifetime is much shorter than the 1000 s you mention. If the data are saved every 5 s, you could have shown even shorter periods at the beginning of the run. Does that not work? I'd be interested to see something like 6a, but with fraction of original particles remaining. It wouldn't have the nice dips in the unstable cases that you point out, but a flattening of the curve, so it wouldn't be as striking a plot, but would be easier to understand.
Line 593: "spatially" is missing the y
Line 627: "Even for neutral conditions, it can take over 90 minutes" implies that unstable conditions take longer! Perhaps just ditch the "Even"