the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
The importance of detection thresholds for the quantification of source and timing of high-latitude dust emission using remote sensing
Rosemary Alice Huck
Robert G. Bryant
James King
Abstract. The observation and quantification of mineral dust fluxes from high-latitude sources remains difficult due to a known paucity of year-round in situ observations and known limitations of satellite remote sensing data (e.g., cloud cover and dust detection). Here we explore the chronology of dust emissions at a known and instrumented high latitude dust source: Lhù’ààn Mân (Kluane Lake) in Yukon, Canada. At this location we combine ground instrumentation, space-based remote sensing platforms, ground-based AERONET data, and oblique camera images to (i) investigate the daily to annual chronology of dust emissions recorded by these instrumental and remote sensing methods (at timescales ranging from minutes to years), and (ii) use data intercomparisons to comment on the principal factors that control the detection of dust in each case. Dust emissions were observed using oblique time-lapse (RC) cameras installed at Lhù’ààn Mân for up to 23 hours a day. These were used as a baseline for analysis of aerosol retrievals from in situ metrological data, AERONET, and co-incident MODIS MAIAC.
Use of high-cadence remote camera (RC) data collected during dust events allowed us to optimise the use of combination of date quality (DQ) 1 (aerosol optical depth - AOD) and DQ2 (single scattering albedo and Angstöm exponent) to best represent AOD dust retrievals from AERONET. Nevertheless, when compared with time series of RC data, optimised AERONET data only manage an overall 26 % detection rate for events (sub day) but 100 % detection rate for dust event days (DED) when dust was within the field of view. Here, in this instance, RC and remote sensing data were able to suggest that the low event detection rate was attributed to fundamental variations in dust advection trajectory, dust plume height, and inherent restrictions in sun angle at high latitudes. Working with a time series of optimised AOD data (covering 2018/2019), we were able to investigate the gross impacts of DQ choice on DED detection at the month/year scale. Relative to ground observations, AERONET’s DQ2.0 cloud screening algorithm may remove as much as 97 % of known dust events (3 % detection). Finally, when undertaking an AOD comparison for DED and non-DED retrievals, we find that cloud screening of MODIS/AERONET lead to a combined low sample of co-incident dust events, and weak correlations between retrievals. Our results quantify and explain the extent of under-representation of dust in both ground and space remote sensing method; a factor which impacts on the effective calibration and validation of global climate and dust models.
-
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
(2047 KB)
-
Supplement
(537 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2047 KB) -
Supplement
(537 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Rosemary Alice Huck et al.
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1156', Anonymous Referee #1, 07 Feb 2023
Review of
“ The importance of detection thresholds for the quantification of source and timing of high-latitude dust emission using remote sensing”
Rosemary Huck, Robert G. Bryant, James King
https://doi.org/10.5194/egusphere-2022-1156
Feb/2023
Summary
This work reports on observations of dust activity in a dry flood plain in NW Canada. The site is example of a high latitude dust source. Dust mobilization is characterized primary by optical sensors (surface photometer part of the Aeronet, meteorological sensors and remote cameras). The report provides a very detailed characterization of dust activity of a rather overlooked dust source and to my knowledge this study constitutes a first focusing on this type of geophysical event using such tools. The study provides information on the typical optical properties useful to be used as guidelines for future investigations. This study provides a number of new results: it demonstrates with clear examples that dust activity of this nature is often not well captured of many mainstream optical sensors (satellites and sunphotometers) and as result the magnitude of such activity is often overlooked in global dust surveys. Thus, lower latitude dust sources are overemphasized just because they can be clearly observed more often with optical sensors. It also provides guidelines on how to characterize these events and what typical detection thresholds (different than normally used for lower latitude dust) can be used in these environments. I welcome this new information about this overlooked phenomenon and this study deserves to be published. Therefore, I advise the editor to do so. I do have some general and more detailed comments listed below which mostly concern readability, clarity in figures and in some paragraphs. Also, I have a comment on the title of the paper.
----------------
General major comment.
While I think that the current title is fine, I believe that it does not highlight some of the more important results from this study. In my opinion, the more important points are two: 1) it unequivocally demonstrates with observations that high latitude dust activity can be very frequent and abundant 2) that existing mainstream instrumentation such as satellite and Aeronet can miss significantly a number of events and demonstrate they are not suitable for a climatological studies. I think these two facts are more relevant and of importance from the view of incorporating HLD in global surveys and modeling efforts. In addition, this study demonstrates something that was already reported in the Urban et al and Baddock et al (cited) papers where they excellently demonstrate how modern polar satellites very often miss dust activity to the point that it is clearly undercounting a significant amount of events. As a result , global assessments that rely in satellite data are biased towards lower latitudes. This study further contributes to this concept with the novelty that this is a largely unknown dust activity regime at latitudes not considered in the above studies.
Overall comments about satellite images. I read this manuscript in a printed version of the paper. All satellite images (except perhaps figure 2) had poor contrast and the darks were too dark and without definition. I can't tell if it was a problem in my printer, but this is a fact you may want to check before final submission. The PDF in the computer screen looked much better than in print.
Abstract:
It would be desirable to add information of the periods of time (months/years) of the surveys.
Overall the abstract highlights too much the technical aspect of detecting of changes thresholds and does not report a more important fact: dust activity is much more frequent than previously expected and this project has quantified it. So for example, stating here what frequency was measured with the remote cameras and by Aeronet is a very important fact in my opinion.
Figure 1. Some of the stations in easter Patagonia are high latitude and do report both proglacial and depression dust activity so they should be tagged in pink. If I recall correctly the Neuquen, Comodoro Rivadavia and Rio Gallegos sites are such cases .
Figure 2: can you add the location of Burwash landing?
Line 202 : The Aeronet ... is a FEDERATED network ....
Line 209-2010: add year of operation for those months.
Line 212 : "...and marine" , really marine aerosol here? it does not make sense to even mention this. Probably you are referring to the optically based aerosols models that can be distinguished with Aeronet. But the way this is phrased, it sounds like these aerosols are present.
Figure 4: can you place location of video cameras in this figure?
Table 2 is not referenced anywhere in the text. With respect MAIAC data, you could add the collection or version of the MAIAC algorithm .
4.1 Event scale Observations. Can you please provide rough numerical estimate of the tops of the dust plumes? are we talking about tens of meters? a few hundred meters height? this is useful for contextual information.
Figure 6. This is a nice and informative figure. But what is the purpose of the labels a,b,c and d if they are not referenced in the text? Also, please make clear in the x-axis that it is local time. Perhaps you could add in one of the mountain slopes a reference height to compare with the dust cloud?. Also, the distance from cameras to mountain visible across the valley would be useful information.
Line 359 - . I found this reasoning difficult to follow because I could not see well in the images the camera locations.
Line 400-404 I think it should be mentioned here the number of clear/cloudy days that Aeronet observed the Sun and how many of those dust was observed.
Figure 8 Caption. The description is a bit difficult to read. Are the vertical bars the DED/week?
Also the coloured lines have poor contrast. Please consider changing and add the colour information in the caption too.
Line 455-459. Please note that while relaxing the threshold criteria makes sense , it also introduces the possibility of cirrus contamination in the Aeronet data. I think and only in this case, it can be circumvented by inspecting the remote camera images for the days with Aeronet observations and check if there are cirrus in the background sky. This could be a quick and dirty way to check that Aeronet data is not contaminated.
Line 457. This is the first instance that Figure 9 is mentioned and it is referred in way as the reader is already familiar with the figure, which is not the case. So please rearrange the text to first introduce the figure and the refer to different sections of it.
Lines 476-480 and 486-490. While I think it makes sense to use thresholds used in other Aeroent dust sites for this case, it is not entirely surprising that there are detection differences. First of all , this site is extremely close to the dust source something that not necessarily is the case in the reference sites used in lower latitudes. In particular, the rapid variability of dust concentrations in puffs of dust is probably one of the main differences. So for example, given the distance to the source, it is likely that this dust has a higher coarse mode contribution to the observed AOD and AE than in lower latitude sites. While I do not think that you can do much to improve on this, I do think that this fact should be mentioned and discussed as probable impacts in observed AODs and AEs.
Line 503-504. Not clear what you mean with "aerosol phases" , what are you referring to?
Figure 11 Caption: Add a clarification that Aeronet retrievals of size distribution and SSA are carried out only for AOD>0.4
Figure 13: the way this is plotted, it suggests MODIS observed the area continuously which probably it did not happen. Can you add symbols to the days where there was a MODIS observation?
General comment triggered by Figure 13
One reason why MODIS may have trouble in this place is that the MODIS pixels are too big, or the observed pixel contains variable combination of bright and dark surfaces all in one pixel that can't be accounted for the retrieval. So perhaps you could clarify somewhere the width of the valley. For example, MODIS pixels are in the 500-1000m size. How do these compare with the typical size of the dust sources in the flood plain?.
Perhaps, it would be illustrative to add a MODIS/VIIRS RGB of one event to illustrate how poorly the plumes are resolved (it will lucky very fuzzy). Just suggestion, it maybe informative for a presentation but probably take too much space in the manuscript.
Line 579 ... detected by?
Citation: https://doi.org/10.5194/egusphere-2022-1156-RC1 -
AC1: 'Reply on RC1', Rosemary Huck, 20 Apr 2023
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2022-1156/egusphere-2022-1156-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Rosemary Huck, 20 Apr 2023
-
RC2: 'Comment on egusphere-2022-1156', Anonymous Referee #2, 02 Mar 2023
Review of
“The importance of detection thresholds for the quantification of source and timing of high-latitude dust emission using remote sensing”
Rosemary Huck, Robert G. Bryant, James King
https://doi.org/10.5194/egusphere-2022-1156
March 2023
Summary:
The occurrence and sources of local dust aerosol in the Arctic region is an important scientific issue, due to the pristine, fragile and climate change sensitivity there. In this work, the authors presented an interesting study on the high latitude dust aerosol over Kluane Lake, using combined observations from AERONET, Remote Camera, meteorological data and satellite data and they try to define a threshold for the quantification of source and timing of high latitude dust at this site. Given the high sensitivity of the Arctic region to climate change, it is important to quantitatively understand and untangle impacting factors such as aerosols on the surface and atmospheric radiative heating balance. Generally speaking, I believe the topic the authors choose is very important, and the findings are meaningful. The manuscript is well-written and easy to follow with some minor technical issues. I suggest the authors revise the manuscript carefully, to make the key parts inside easier to follow.
General major comment:
- Besides the AOD data and the Angstrom Exponent data, did you analyze the SDA from the sun photometer? I believe this will give further information on fine and coarse mode AOD and it would help to better differentiate the fine and coarse mode aerosols.
- I think this classification based on the threshold from Verma et al. (2015) and Dubovik et al. (2002) are very primitive and as presented in figure 9b it is not very realistic. You try to classify aerosol types based on Verma et al. (2015) and Dubovik et al. (2002) thresholds. First of all, Dubovik et al. (2002) tried to investigate the absorption and other aerosol optical properties in several key locations. As they mentioned in this paper, they used these thresholds because the values of real and imaginary parts of the refractive index, as well as single scattering albedo, are given only for the condition of τ(440) > 0.4 for Urban-industrial, mixed, and biomass burning aerosols, and for the conditions of τ ext(1020) > 0.3 and α < 0.6 for desert dust. On the other hand, Verma et al. (2015) tried to define thresholds for the Jaipur AERONET station in India and the aerosol in different regions has its own properties (as you claimed in the introduction). So, I strongly suggest authors define their thresholds for their station based on the properties of the aerosol at Kluane lake station.
- If you are using Version 3 Level 1.0 data (or DQ 1) you need to further investigate the clouds effects on your results. I quickly checked MODIS Aqua imageries for Kluane lake for the month of May 2018 and based on these images at least 30% of the days in this month were cloudy or partly cloudy.
- I suggest adding a few sentences about how you compare daily mean MAIAC AOD with daily AERONET mean AOD. It is not clear if you compare one pixel (1*1km) with AERONET AOD or if you chose a bigger area and then averaged AOD values and used the averaged value for your analysis.
specific comments:
I suggest using either “high latitude” or “high-latitude”
I suggest using either “ground based” or “ground-based”
L94: by direct sun and “sky scan” measurements
L98: The Angström exponent (α) allows estimation of aerosol particle size (effective radius) not aerosol size distribution.
L103: this map needs to be fixed. I don’t see any highlighted area in “yellow” and also circles are outside of the map. In the legend N is ~1750 but, in the caption, it says N ~ 1075. Please fix this as well.
L134-135 Since AERONET also is a remote sensing measurement maybe you need to specify the remote observations to satellite observations.
L199: Please add “Figure S1”
L 210: You need to specify these max and min observed values are for the Kluane Lake AERONET station.
L212: One of the aerosol types that you have in your classification is Marine. I don’t know if it is the case for Kluane lake with about 200 km far from the ocean and all of those mountains around the lake.
L 213-214: based on Giles et. al., (2019), Level 1.5 represents near-real-time automatic cloud screening and automatic instrument anomaly quality controls and Level 2.0 additionally applies pre-field and post-field calibrations.
L222-225: So how about cases where AOD is less than 0.3? That is the reason I think you should use the SDA product which is a standard product of AERONET and it breakdown the total AOD into fine and coarse mode AOD. So, you don’t need to define this threshold.
L310-312: Please add an explanation of why you use Spearman’s rank correlation coefficient.
L397: Not only below the mountain line but also if dust remains out of the sun photometer’s FOV still it will not be captured. So, for example, if the sun is in the south of the KLRS and the dust plume is to the north of the site, still it will not be captured.
L413-419: the caption is not very clear please use the same definition for the different variables as your y-axis labels. For example, you talk about Average weekly depth and the y labels are Lake Depth (m) and Weekly total Snowfall (cm) and it is a bit confusing.
L428: In the Figure 8 caption you said that “The AERONET station was recording from 14/05/18 until 21/10/18” and here “The AERONET station began recording on 13th May 2018”. Please fix the one that is not correct.
L480: I don’t agree with this sentence. The reason that shorter wavelengths lead to an underrepresentation of dust events and DED frequency is the threshold that you used for the shorter wavelength.
L493-494: Please rewrite this sentence. It is hard to understand.
L496-497: based on Figures 11a and 11b the distribution is not bimodal and it is trimodal with a fine mode peak of 0.2 to 0.4 μm and 2 coarse mode peaks at 2.6 and 10.1μm.
L499_500: AERONET inversion PSD bins (x-axis) are radius, not diameter but Bachelder et al. (2020) results as you mentioned here are in diameter so you need to convert one to another and then compare them together.
L501: Do you think this second peak (around 10μin radius) is real or it might be affected by the cloud? I know these results are from Level 2 inversion but if the cloud screening doesn’t work well for direct sun measurements, how much do you believe in these results from AERONET Inversion?
L503: “aerosol phases” is not clear to me. I think by “aerosol phases” you mean “aerosol types”
L510: the colours in the scatterplot are not the same as the colour in the caption, please fix this.
L519: Please fix the colour or the caption of Figure 10. I don’t see any yellow or black in this plot.
L521: Figure 11 a and b x-axis labels are radius, not diameter, please fix the labels.
L521: Why some of these plots in figure 11 are just for one year (for example Figure 11c) and some of these plots are for two years (Figures 11d and e)? Maybe it would be more reasonable if you keep the time range the same for all plots.
L527: Again, the distribution is trimodal with a fine mode peak and two coarse mode peaks.
L529-530: Again, the colours in the caption have not matched the colours in the plot.
L554: I think you mean Figures 12 and 13
Citation: https://doi.org/10.5194/egusphere-2022-1156-RC2 -
AC2: 'Reply on RC2', Rosemary Huck, 20 Apr 2023
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2022-1156/egusphere-2022-1156-AC2-supplement.pdf
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1156', Anonymous Referee #1, 07 Feb 2023
Review of
“ The importance of detection thresholds for the quantification of source and timing of high-latitude dust emission using remote sensing”
Rosemary Huck, Robert G. Bryant, James King
https://doi.org/10.5194/egusphere-2022-1156
Feb/2023
Summary
This work reports on observations of dust activity in a dry flood plain in NW Canada. The site is example of a high latitude dust source. Dust mobilization is characterized primary by optical sensors (surface photometer part of the Aeronet, meteorological sensors and remote cameras). The report provides a very detailed characterization of dust activity of a rather overlooked dust source and to my knowledge this study constitutes a first focusing on this type of geophysical event using such tools. The study provides information on the typical optical properties useful to be used as guidelines for future investigations. This study provides a number of new results: it demonstrates with clear examples that dust activity of this nature is often not well captured of many mainstream optical sensors (satellites and sunphotometers) and as result the magnitude of such activity is often overlooked in global dust surveys. Thus, lower latitude dust sources are overemphasized just because they can be clearly observed more often with optical sensors. It also provides guidelines on how to characterize these events and what typical detection thresholds (different than normally used for lower latitude dust) can be used in these environments. I welcome this new information about this overlooked phenomenon and this study deserves to be published. Therefore, I advise the editor to do so. I do have some general and more detailed comments listed below which mostly concern readability, clarity in figures and in some paragraphs. Also, I have a comment on the title of the paper.
----------------
General major comment.
While I think that the current title is fine, I believe that it does not highlight some of the more important results from this study. In my opinion, the more important points are two: 1) it unequivocally demonstrates with observations that high latitude dust activity can be very frequent and abundant 2) that existing mainstream instrumentation such as satellite and Aeronet can miss significantly a number of events and demonstrate they are not suitable for a climatological studies. I think these two facts are more relevant and of importance from the view of incorporating HLD in global surveys and modeling efforts. In addition, this study demonstrates something that was already reported in the Urban et al and Baddock et al (cited) papers where they excellently demonstrate how modern polar satellites very often miss dust activity to the point that it is clearly undercounting a significant amount of events. As a result , global assessments that rely in satellite data are biased towards lower latitudes. This study further contributes to this concept with the novelty that this is a largely unknown dust activity regime at latitudes not considered in the above studies.
Overall comments about satellite images. I read this manuscript in a printed version of the paper. All satellite images (except perhaps figure 2) had poor contrast and the darks were too dark and without definition. I can't tell if it was a problem in my printer, but this is a fact you may want to check before final submission. The PDF in the computer screen looked much better than in print.
Abstract:
It would be desirable to add information of the periods of time (months/years) of the surveys.
Overall the abstract highlights too much the technical aspect of detecting of changes thresholds and does not report a more important fact: dust activity is much more frequent than previously expected and this project has quantified it. So for example, stating here what frequency was measured with the remote cameras and by Aeronet is a very important fact in my opinion.
Figure 1. Some of the stations in easter Patagonia are high latitude and do report both proglacial and depression dust activity so they should be tagged in pink. If I recall correctly the Neuquen, Comodoro Rivadavia and Rio Gallegos sites are such cases .
Figure 2: can you add the location of Burwash landing?
Line 202 : The Aeronet ... is a FEDERATED network ....
Line 209-2010: add year of operation for those months.
Line 212 : "...and marine" , really marine aerosol here? it does not make sense to even mention this. Probably you are referring to the optically based aerosols models that can be distinguished with Aeronet. But the way this is phrased, it sounds like these aerosols are present.
Figure 4: can you place location of video cameras in this figure?
Table 2 is not referenced anywhere in the text. With respect MAIAC data, you could add the collection or version of the MAIAC algorithm .
4.1 Event scale Observations. Can you please provide rough numerical estimate of the tops of the dust plumes? are we talking about tens of meters? a few hundred meters height? this is useful for contextual information.
Figure 6. This is a nice and informative figure. But what is the purpose of the labels a,b,c and d if they are not referenced in the text? Also, please make clear in the x-axis that it is local time. Perhaps you could add in one of the mountain slopes a reference height to compare with the dust cloud?. Also, the distance from cameras to mountain visible across the valley would be useful information.
Line 359 - . I found this reasoning difficult to follow because I could not see well in the images the camera locations.
Line 400-404 I think it should be mentioned here the number of clear/cloudy days that Aeronet observed the Sun and how many of those dust was observed.
Figure 8 Caption. The description is a bit difficult to read. Are the vertical bars the DED/week?
Also the coloured lines have poor contrast. Please consider changing and add the colour information in the caption too.
Line 455-459. Please note that while relaxing the threshold criteria makes sense , it also introduces the possibility of cirrus contamination in the Aeronet data. I think and only in this case, it can be circumvented by inspecting the remote camera images for the days with Aeronet observations and check if there are cirrus in the background sky. This could be a quick and dirty way to check that Aeronet data is not contaminated.
Line 457. This is the first instance that Figure 9 is mentioned and it is referred in way as the reader is already familiar with the figure, which is not the case. So please rearrange the text to first introduce the figure and the refer to different sections of it.
Lines 476-480 and 486-490. While I think it makes sense to use thresholds used in other Aeroent dust sites for this case, it is not entirely surprising that there are detection differences. First of all , this site is extremely close to the dust source something that not necessarily is the case in the reference sites used in lower latitudes. In particular, the rapid variability of dust concentrations in puffs of dust is probably one of the main differences. So for example, given the distance to the source, it is likely that this dust has a higher coarse mode contribution to the observed AOD and AE than in lower latitude sites. While I do not think that you can do much to improve on this, I do think that this fact should be mentioned and discussed as probable impacts in observed AODs and AEs.
Line 503-504. Not clear what you mean with "aerosol phases" , what are you referring to?
Figure 11 Caption: Add a clarification that Aeronet retrievals of size distribution and SSA are carried out only for AOD>0.4
Figure 13: the way this is plotted, it suggests MODIS observed the area continuously which probably it did not happen. Can you add symbols to the days where there was a MODIS observation?
General comment triggered by Figure 13
One reason why MODIS may have trouble in this place is that the MODIS pixels are too big, or the observed pixel contains variable combination of bright and dark surfaces all in one pixel that can't be accounted for the retrieval. So perhaps you could clarify somewhere the width of the valley. For example, MODIS pixels are in the 500-1000m size. How do these compare with the typical size of the dust sources in the flood plain?.
Perhaps, it would be illustrative to add a MODIS/VIIRS RGB of one event to illustrate how poorly the plumes are resolved (it will lucky very fuzzy). Just suggestion, it maybe informative for a presentation but probably take too much space in the manuscript.
Line 579 ... detected by?
Citation: https://doi.org/10.5194/egusphere-2022-1156-RC1 -
AC1: 'Reply on RC1', Rosemary Huck, 20 Apr 2023
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2022-1156/egusphere-2022-1156-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Rosemary Huck, 20 Apr 2023
-
RC2: 'Comment on egusphere-2022-1156', Anonymous Referee #2, 02 Mar 2023
Review of
“The importance of detection thresholds for the quantification of source and timing of high-latitude dust emission using remote sensing”
Rosemary Huck, Robert G. Bryant, James King
https://doi.org/10.5194/egusphere-2022-1156
March 2023
Summary:
The occurrence and sources of local dust aerosol in the Arctic region is an important scientific issue, due to the pristine, fragile and climate change sensitivity there. In this work, the authors presented an interesting study on the high latitude dust aerosol over Kluane Lake, using combined observations from AERONET, Remote Camera, meteorological data and satellite data and they try to define a threshold for the quantification of source and timing of high latitude dust at this site. Given the high sensitivity of the Arctic region to climate change, it is important to quantitatively understand and untangle impacting factors such as aerosols on the surface and atmospheric radiative heating balance. Generally speaking, I believe the topic the authors choose is very important, and the findings are meaningful. The manuscript is well-written and easy to follow with some minor technical issues. I suggest the authors revise the manuscript carefully, to make the key parts inside easier to follow.
General major comment:
- Besides the AOD data and the Angstrom Exponent data, did you analyze the SDA from the sun photometer? I believe this will give further information on fine and coarse mode AOD and it would help to better differentiate the fine and coarse mode aerosols.
- I think this classification based on the threshold from Verma et al. (2015) and Dubovik et al. (2002) are very primitive and as presented in figure 9b it is not very realistic. You try to classify aerosol types based on Verma et al. (2015) and Dubovik et al. (2002) thresholds. First of all, Dubovik et al. (2002) tried to investigate the absorption and other aerosol optical properties in several key locations. As they mentioned in this paper, they used these thresholds because the values of real and imaginary parts of the refractive index, as well as single scattering albedo, are given only for the condition of τ(440) > 0.4 for Urban-industrial, mixed, and biomass burning aerosols, and for the conditions of τ ext(1020) > 0.3 and α < 0.6 for desert dust. On the other hand, Verma et al. (2015) tried to define thresholds for the Jaipur AERONET station in India and the aerosol in different regions has its own properties (as you claimed in the introduction). So, I strongly suggest authors define their thresholds for their station based on the properties of the aerosol at Kluane lake station.
- If you are using Version 3 Level 1.0 data (or DQ 1) you need to further investigate the clouds effects on your results. I quickly checked MODIS Aqua imageries for Kluane lake for the month of May 2018 and based on these images at least 30% of the days in this month were cloudy or partly cloudy.
- I suggest adding a few sentences about how you compare daily mean MAIAC AOD with daily AERONET mean AOD. It is not clear if you compare one pixel (1*1km) with AERONET AOD or if you chose a bigger area and then averaged AOD values and used the averaged value for your analysis.
specific comments:
I suggest using either “high latitude” or “high-latitude”
I suggest using either “ground based” or “ground-based”
L94: by direct sun and “sky scan” measurements
L98: The Angström exponent (α) allows estimation of aerosol particle size (effective radius) not aerosol size distribution.
L103: this map needs to be fixed. I don’t see any highlighted area in “yellow” and also circles are outside of the map. In the legend N is ~1750 but, in the caption, it says N ~ 1075. Please fix this as well.
L134-135 Since AERONET also is a remote sensing measurement maybe you need to specify the remote observations to satellite observations.
L199: Please add “Figure S1”
L 210: You need to specify these max and min observed values are for the Kluane Lake AERONET station.
L212: One of the aerosol types that you have in your classification is Marine. I don’t know if it is the case for Kluane lake with about 200 km far from the ocean and all of those mountains around the lake.
L 213-214: based on Giles et. al., (2019), Level 1.5 represents near-real-time automatic cloud screening and automatic instrument anomaly quality controls and Level 2.0 additionally applies pre-field and post-field calibrations.
L222-225: So how about cases where AOD is less than 0.3? That is the reason I think you should use the SDA product which is a standard product of AERONET and it breakdown the total AOD into fine and coarse mode AOD. So, you don’t need to define this threshold.
L310-312: Please add an explanation of why you use Spearman’s rank correlation coefficient.
L397: Not only below the mountain line but also if dust remains out of the sun photometer’s FOV still it will not be captured. So, for example, if the sun is in the south of the KLRS and the dust plume is to the north of the site, still it will not be captured.
L413-419: the caption is not very clear please use the same definition for the different variables as your y-axis labels. For example, you talk about Average weekly depth and the y labels are Lake Depth (m) and Weekly total Snowfall (cm) and it is a bit confusing.
L428: In the Figure 8 caption you said that “The AERONET station was recording from 14/05/18 until 21/10/18” and here “The AERONET station began recording on 13th May 2018”. Please fix the one that is not correct.
L480: I don’t agree with this sentence. The reason that shorter wavelengths lead to an underrepresentation of dust events and DED frequency is the threshold that you used for the shorter wavelength.
L493-494: Please rewrite this sentence. It is hard to understand.
L496-497: based on Figures 11a and 11b the distribution is not bimodal and it is trimodal with a fine mode peak of 0.2 to 0.4 μm and 2 coarse mode peaks at 2.6 and 10.1μm.
L499_500: AERONET inversion PSD bins (x-axis) are radius, not diameter but Bachelder et al. (2020) results as you mentioned here are in diameter so you need to convert one to another and then compare them together.
L501: Do you think this second peak (around 10μin radius) is real or it might be affected by the cloud? I know these results are from Level 2 inversion but if the cloud screening doesn’t work well for direct sun measurements, how much do you believe in these results from AERONET Inversion?
L503: “aerosol phases” is not clear to me. I think by “aerosol phases” you mean “aerosol types”
L510: the colours in the scatterplot are not the same as the colour in the caption, please fix this.
L519: Please fix the colour or the caption of Figure 10. I don’t see any yellow or black in this plot.
L521: Figure 11 a and b x-axis labels are radius, not diameter, please fix the labels.
L521: Why some of these plots in figure 11 are just for one year (for example Figure 11c) and some of these plots are for two years (Figures 11d and e)? Maybe it would be more reasonable if you keep the time range the same for all plots.
L527: Again, the distribution is trimodal with a fine mode peak and two coarse mode peaks.
L529-530: Again, the colours in the caption have not matched the colours in the plot.
L554: I think you mean Figures 12 and 13
Citation: https://doi.org/10.5194/egusphere-2022-1156-RC2 -
AC2: 'Reply on RC2', Rosemary Huck, 20 Apr 2023
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2023/egusphere-2022-1156/egusphere-2022-1156-AC2-supplement.pdf
Peer review completion
Journal article(s) based on this preprint
Rosemary Alice Huck et al.
Data sets
Meterological data Rosemary Huck, James King https://doi.org/10.5281/zenodo.7249227
Remote camera images James King, Rosemary Huck https://doi.org/10.5281/zenodo.7249227
Rosemary Alice Huck et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 285 | 101 | 20 | 406 | 46 | 8 | 10 |
- HTML: 285
- PDF: 101
- XML: 20
- Total: 406
- Supplement: 46
- BibTeX: 8
- EndNote: 10
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(2047 KB) - Metadata XML
-
Supplement
(537 KB) - BibTeX
- EndNote
- Final revised paper