the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Dynamics of the snow grain size in a windy coastal area of Antarctica from continuous in-situ spectral albedo measurements
Sara Arioli
Ghislain Picard
Laurent Arnaud
Vincent Favier
Abstract. The grain size of the superficial snow layer is a key determinant of the surface albedo in Antarctica. Its evolution is the result of multiple interacting processes, such as dry and wet metamorphism, melt, snow drift and precipitation. Among them, snow drift has the least known and least predictable impact. The goal of this study is to relate the variations of surface snow grain size to these processes in a windy location of the Antarctic coast. For this, we retrieved the daily grain size from 5 year-long in-situ observations of the spectral albedo recorded by a new multi-band albedometer, unique in terms of autonomy and described here for the first time. An uncertainty assessment and a comparison with satellite-retrieved grain size were carried out to verify the reliability of the instrument and an RMSE up to 0.16 mm on the observed grain size was found. By relating these in-situ measurements to timeseries of snow drift, surface temperature, snow surface height and snowfall, we established that the evolution of the grain size in the presence of snow drift is complex and follows two possible pathways: 1) A decrease in the grain size (about half of our measurements) resulting from the deposition of small grains advected by the wind. Surprisingly, this decreases is often (2/3 of the cases) associated with a decrease of the surface height, i.e. a net erosion over the drift episode, 2) an increase of the grain size (the other half) either due to the removal of the surface layer, or metamorphism. However, we note that this increase is often limited with respect to the increase predicted by a theoretical metamorphism model, suggesting that a concomitant deposition of small grains is likely. At last, we found that wind also completely impedes the deposition of snowfall during half of the observed precipitation events. When this happens, the grain size evolves as if precipitation was not occurring. As a result of all these processes, we conclude that the grain size in a windy area remains more stable than it would be in the absence of snow drift, hence limiting the variations of the albedo and of the radiative energy budget.
Sara Arioli et al.
Status: closed
-
RC1: 'Comment on tc-2022-236', Anonymous Referee #1, 09 Feb 2023
Dynamics of the snow grain size in a windy coastal area of Antarctica from continuous in-situ spectral albedo measurements: Arioli et al., 2022, The Cryosphere Discussions
The manuscript presents an extensive data set of multi-year optical snow grain size retrievals from surface albedo measurements at a windy Antarctic coastal site. After a thorough characterization of the calibration steps of the albedo measurements, the authors show statistics of the evolution of the optical snow grain size in relation to various meteorological parameters such as surface temperature, wind speed, snow drift estimates, snowfall, and snow height measurements. The main conclusion of the paper is that sites affected by strong snow drift can show a more stable evolution of the snow grain size than previously expected.
I want to start with congratulating the authors for (1) producing an impressive, much needed and long-term data set of spectral surface albedo and optical snow grain size estimates at a remote Antarctic location, (2) a very careful characterization of their instrument and retrieval algorithm, (3) their extensive statistical efforts in trying to disentangle the different processes such as snow drift and snowfall, and (4) presenting their results in such a comprehensive and clear way. The structure of the manuscript is very logical and the figures are generally of very good quality, which made reading the paper easy and enjoyable.
I am happy to recommend the publication of this manuscript subject to minor revisions, as I do have some comments that I believe need further clarification.
After some general comments, more specific comments and suggestions for technical corrections are given below.
General comments
There is a huge need for precise surface albedo measurements, which require a thorough calibration and estimation of the measurement uncertainties. The different processing and calibration steps are described very clearly, however I still have some open questions:
(1) I am wondering about the cross calibration of the two sensor heads. Maybe I missed it in the manuscript, but what were the weather conditions during the side-by-side observations? In my opinion, this cross calibration should be performed using a purely diffuse light source only, as otherwise the differences in the cosine response of the two sensor heads are somehow corrected as well, even though the actual cosine response correction is performed in the next step and only for the upward-looking sensor which receives direct radiation. In my eyes, a cross-calibrating during conditions with direct incident radiation would lead to a compensation for effects between the two sensor heads which are not occurring during the actual measurements (as the downward-looking sensor head only receives diffuse radiation). Please correct me if I am wrong, but I would like you to comment on my line of thought.
(2) The bandpass filters used for the spectral measurements have a certain spectral width (usually characterized by a FWHM - full width at half maximum). Could you please specify the FWHM of the spectral filters used in the instrument, and also discuss the influence of the non-discrete wavelengths on the retrieval of the optical snow grain size using the ART formulas?
(3) Clear sky index (Lines 157-164): Why is the clear sky index (CSI) not applied to each individual measurement, but the entire day is assumed to be clear sky when 75% of the observations had a CSI below 1.25? It seems applying the clear sky/overcast distinction before calculating the daily average would reduce the uncertainty in the retrieval?
(4) This applies also to the remark on Line 216: Would it not be possible to use these filtered measurements if one would apply a CSI to each measurement?
(5) How broken are these clouds that are still considered to be clear sky? Maybe you can give some more details here from Marty and Philipona (2000), if available?
(6) The computation of the CSI uses temperature and humidity data from the AWS at D17 (which is above 400 m a.s.l.). What is the influence to use the same data and apply it to D5 (below 200 m a.s.l.)? Is this why you chose to apply the same CSI to one full day as there was no separate AWS available at D5? Or are you using reanalysis data for D5? This needs to be clarified within the manuscript.
(7) Why are the SSA/d_opt only retrieved from the diffuse albedo spectra (e.g. Line 186 and 202)? If you are correcting for the cosine response of the sensor head and use Eq. 2, is it not possible to retrieve SSA/d_opt from the direct albedo measurements?
(8) At first glance in Fig. 7, it looks like the variations in SW albedo measured by the CNR4 (shaded blue) are largest during clear-sky days (noon line shows the clear-sky symbol). This underlines your statement in the text that the influence of e.g. sastrugi is largest on clear-sky days. I can follow your argument why you are focusing on the noon observations to reduce this error, but wouldn’t this error be even lower if you would focus only on the overcast observations for this test? Especially with the different footprints (CNR4 installed at lower height), I think it is reasonable to try and reduce the uncertainties as much as possible in order to make the comparison as fair as possible between the two approaches. Thus, it would be interesting to see the statistics/deviations if you discard the clear-sky observations in the time series for this test.
(9) When comparing the d_opt from the surface albedo measurements with the satellite observations: could you please briefly discuss the influence of the different wavelengths used in the ground-based and satellite retrievals? The different wavelengths would lead to slightly different penetration depths of the radiation into the snowpack, thus the two instruments are 'seeing' slightly different layers of the snowpack.
The evolution of the snowpack over the 5 seasons presented in this study is discussed in very good detail. However, the manuscript would definitely benefit from putting this impressive data set in perspective to very similar studies at other Antarctic locations. For example, Libois et al. (2015) presented a multi-year study of SSA evolution retrieved from albedo measurements and discuss the influence of drifting snow. Also, Carlsen et al. (2017) showed the temporal evolution of the SSA from surface albedo measurements on the Antarctic plateau and compared it to in situ and – similarly to this study – optical satellite observations. A more thorough discussion on how the different studies compare would be an important addition to the discussion.
Specific comments
Introduction: wavelength dependence of surface albedo - could you give some typical values from literature?
L19: I recommend a quick, short definition of surface albedo at the first mentioning.
Figure 1: Grid lines would help, maybe include a photo of sastrugi/snow drift event as panel c?
Figure 2: Please highlight the components by annotating the photo
Section 2.1: One more sentence to the measurement principle of the snow drift volume would help at this point.
L124: It would be helpful to give a percentage of how many measurements were filtered out by the SZA criterion and the 1% total irradiance criterion.
Figure 5: the axis annotations for SSA and d_opt need a unit for both (m2/kg and mm I believe)
L251: I think ‘reproduce’ is a bit misleading here, implicating that the satellite measurements would be the ground truth, even though they come with a lot of uncertainties in themselves (as you mention before).
L305: T_air,max is mentioned here for the first time, but never explained. What is the difference to T_s,max?
L312-: It is nice to see all these statistics and to put inter-seasonal and interannual variability in relation to each other. However, it would be good to compare this to literature values of other snow drift measurements in Antarctica and maybe put these values into context in terms of weak and strong snow drift events.
Figure 11d: the d_opt values are within the shaded area to classify melting conditions (caption of Figure 9), however the surface temperature is clearly below 0°C so I do not know how helpful the grey shading in Fig. 11 is, especially as these are cases with no melt. You could consider removing this shading to foster clarity.
L357: Some further justification is needed why the 0.2mm measurement of the grain size of freshly precipitated snow by Domine et al. (2007) is used as an upper threshold here. With the given information, this seems quite arbitrary to me.
Technical corrections
L11: decrease
L27: liquid water content
L85: shortwave infrared
L94/96: irradiance instead of radiance
Table 1 caption: irradiance
L110: Fig. 2
L243: omit additional ‘that’
Figure 7 caption: 10th and 90th percentile
L280: This is the reason why both are shown.
Figure 9: the x axis should just be the months, in order to correspond to the other panels for the other years (so no specific 2022-10, 2022-11, but only 10, 11, …)
L312: ever-present
L314: It would be easier for the reader to stick to the unit kg m-2 d-1, and not switch to Mg
L322: Days satisfying this criterion correspond to 3% of the total measurements in November, 48% in December …
L327: increases by
L361: I believe you mean delta SF?
L389: and the snow height decreases
Figure 13 caption: ‘an episode of erosion’
L425: shortwave
L460: on days
References:
Libois, Q., Picard, G., Arnaud, L., Dumont, M., Lafaysse, M., Morin, S., and Lefebvre, E.: Summertime evolution of snow specific surface area close to the surface on the Antarctic Plateau, The Cryosphere, 9, 2383–2398, https://doi.org/10.5194/tc-9-2383-2015, 2015
Carlsen, T., Birnbaum, G., Ehrlich, A., Freitag, J., Heygster, G., Istomina, L., Kipfstuhl, S., Orsi, A., Schäfer, M., and Wendisch, M.: Comparison of different methods to retrieve optical-equivalent snow grain size in central Antarctica, The Cryosphere, 11, 2727–2741, https://doi.org/10.5194/tc-11-2727-2017, 2017
Citation: https://doi.org/10.5194/tc-2022-236-RC1 -
AC1: 'Reply on RC1', Sara Arioli, 31 Mar 2023
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2022-236/tc-2022-236-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Sara Arioli, 31 Mar 2023
-
RC2: 'Comment on tc-2022-236', Anonymous Referee #2, 17 Feb 2023
After reading this nice analysis on snow grain size behavior adjacent to Dumont D'Urville station, it is clear that I do not have sufficient expertise in this topic area to provide useful scientific input. On page 22, the authors conclude that snow drift impacts at D17 prevent significant changes to snow grain size due to competing effects. That is, there is relative stability to the surface albedo.
Here are some small items:
Lines 37-39: This sentence seems to be saying contradictory things. Higher precipitation should probably be associated with higher albedo.
Line 41: Drop "operated".
Line 77: Automatic rather than Automated.
Line 113: Rephrase to "buried in the snow to provide temperature stability for the electronics".
Line 211: "obstructing" rather than "obturating" would be better as much more frequent usage and less jarring.
Figure 9: Caption says "The shaded gray area marks dopt > 0.64 mm". The gray shading along the top of each panel is continuously gray
and does not coincide with dopt > 0.64 mm in most locations.
Line 543: Complete the reference for Colbeck 1973.
Citation: https://doi.org/10.5194/tc-2022-236-RC2 -
AC2: 'Reply on RC2', Sara Arioli, 31 Mar 2023
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2022-236/tc-2022-236-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Sara Arioli, 31 Mar 2023
Status: closed
-
RC1: 'Comment on tc-2022-236', Anonymous Referee #1, 09 Feb 2023
Dynamics of the snow grain size in a windy coastal area of Antarctica from continuous in-situ spectral albedo measurements: Arioli et al., 2022, The Cryosphere Discussions
The manuscript presents an extensive data set of multi-year optical snow grain size retrievals from surface albedo measurements at a windy Antarctic coastal site. After a thorough characterization of the calibration steps of the albedo measurements, the authors show statistics of the evolution of the optical snow grain size in relation to various meteorological parameters such as surface temperature, wind speed, snow drift estimates, snowfall, and snow height measurements. The main conclusion of the paper is that sites affected by strong snow drift can show a more stable evolution of the snow grain size than previously expected.
I want to start with congratulating the authors for (1) producing an impressive, much needed and long-term data set of spectral surface albedo and optical snow grain size estimates at a remote Antarctic location, (2) a very careful characterization of their instrument and retrieval algorithm, (3) their extensive statistical efforts in trying to disentangle the different processes such as snow drift and snowfall, and (4) presenting their results in such a comprehensive and clear way. The structure of the manuscript is very logical and the figures are generally of very good quality, which made reading the paper easy and enjoyable.
I am happy to recommend the publication of this manuscript subject to minor revisions, as I do have some comments that I believe need further clarification.
After some general comments, more specific comments and suggestions for technical corrections are given below.
General comments
There is a huge need for precise surface albedo measurements, which require a thorough calibration and estimation of the measurement uncertainties. The different processing and calibration steps are described very clearly, however I still have some open questions:
(1) I am wondering about the cross calibration of the two sensor heads. Maybe I missed it in the manuscript, but what were the weather conditions during the side-by-side observations? In my opinion, this cross calibration should be performed using a purely diffuse light source only, as otherwise the differences in the cosine response of the two sensor heads are somehow corrected as well, even though the actual cosine response correction is performed in the next step and only for the upward-looking sensor which receives direct radiation. In my eyes, a cross-calibrating during conditions with direct incident radiation would lead to a compensation for effects between the two sensor heads which are not occurring during the actual measurements (as the downward-looking sensor head only receives diffuse radiation). Please correct me if I am wrong, but I would like you to comment on my line of thought.
(2) The bandpass filters used for the spectral measurements have a certain spectral width (usually characterized by a FWHM - full width at half maximum). Could you please specify the FWHM of the spectral filters used in the instrument, and also discuss the influence of the non-discrete wavelengths on the retrieval of the optical snow grain size using the ART formulas?
(3) Clear sky index (Lines 157-164): Why is the clear sky index (CSI) not applied to each individual measurement, but the entire day is assumed to be clear sky when 75% of the observations had a CSI below 1.25? It seems applying the clear sky/overcast distinction before calculating the daily average would reduce the uncertainty in the retrieval?
(4) This applies also to the remark on Line 216: Would it not be possible to use these filtered measurements if one would apply a CSI to each measurement?
(5) How broken are these clouds that are still considered to be clear sky? Maybe you can give some more details here from Marty and Philipona (2000), if available?
(6) The computation of the CSI uses temperature and humidity data from the AWS at D17 (which is above 400 m a.s.l.). What is the influence to use the same data and apply it to D5 (below 200 m a.s.l.)? Is this why you chose to apply the same CSI to one full day as there was no separate AWS available at D5? Or are you using reanalysis data for D5? This needs to be clarified within the manuscript.
(7) Why are the SSA/d_opt only retrieved from the diffuse albedo spectra (e.g. Line 186 and 202)? If you are correcting for the cosine response of the sensor head and use Eq. 2, is it not possible to retrieve SSA/d_opt from the direct albedo measurements?
(8) At first glance in Fig. 7, it looks like the variations in SW albedo measured by the CNR4 (shaded blue) are largest during clear-sky days (noon line shows the clear-sky symbol). This underlines your statement in the text that the influence of e.g. sastrugi is largest on clear-sky days. I can follow your argument why you are focusing on the noon observations to reduce this error, but wouldn’t this error be even lower if you would focus only on the overcast observations for this test? Especially with the different footprints (CNR4 installed at lower height), I think it is reasonable to try and reduce the uncertainties as much as possible in order to make the comparison as fair as possible between the two approaches. Thus, it would be interesting to see the statistics/deviations if you discard the clear-sky observations in the time series for this test.
(9) When comparing the d_opt from the surface albedo measurements with the satellite observations: could you please briefly discuss the influence of the different wavelengths used in the ground-based and satellite retrievals? The different wavelengths would lead to slightly different penetration depths of the radiation into the snowpack, thus the two instruments are 'seeing' slightly different layers of the snowpack.
The evolution of the snowpack over the 5 seasons presented in this study is discussed in very good detail. However, the manuscript would definitely benefit from putting this impressive data set in perspective to very similar studies at other Antarctic locations. For example, Libois et al. (2015) presented a multi-year study of SSA evolution retrieved from albedo measurements and discuss the influence of drifting snow. Also, Carlsen et al. (2017) showed the temporal evolution of the SSA from surface albedo measurements on the Antarctic plateau and compared it to in situ and – similarly to this study – optical satellite observations. A more thorough discussion on how the different studies compare would be an important addition to the discussion.
Specific comments
Introduction: wavelength dependence of surface albedo - could you give some typical values from literature?
L19: I recommend a quick, short definition of surface albedo at the first mentioning.
Figure 1: Grid lines would help, maybe include a photo of sastrugi/snow drift event as panel c?
Figure 2: Please highlight the components by annotating the photo
Section 2.1: One more sentence to the measurement principle of the snow drift volume would help at this point.
L124: It would be helpful to give a percentage of how many measurements were filtered out by the SZA criterion and the 1% total irradiance criterion.
Figure 5: the axis annotations for SSA and d_opt need a unit for both (m2/kg and mm I believe)
L251: I think ‘reproduce’ is a bit misleading here, implicating that the satellite measurements would be the ground truth, even though they come with a lot of uncertainties in themselves (as you mention before).
L305: T_air,max is mentioned here for the first time, but never explained. What is the difference to T_s,max?
L312-: It is nice to see all these statistics and to put inter-seasonal and interannual variability in relation to each other. However, it would be good to compare this to literature values of other snow drift measurements in Antarctica and maybe put these values into context in terms of weak and strong snow drift events.
Figure 11d: the d_opt values are within the shaded area to classify melting conditions (caption of Figure 9), however the surface temperature is clearly below 0°C so I do not know how helpful the grey shading in Fig. 11 is, especially as these are cases with no melt. You could consider removing this shading to foster clarity.
L357: Some further justification is needed why the 0.2mm measurement of the grain size of freshly precipitated snow by Domine et al. (2007) is used as an upper threshold here. With the given information, this seems quite arbitrary to me.
Technical corrections
L11: decrease
L27: liquid water content
L85: shortwave infrared
L94/96: irradiance instead of radiance
Table 1 caption: irradiance
L110: Fig. 2
L243: omit additional ‘that’
Figure 7 caption: 10th and 90th percentile
L280: This is the reason why both are shown.
Figure 9: the x axis should just be the months, in order to correspond to the other panels for the other years (so no specific 2022-10, 2022-11, but only 10, 11, …)
L312: ever-present
L314: It would be easier for the reader to stick to the unit kg m-2 d-1, and not switch to Mg
L322: Days satisfying this criterion correspond to 3% of the total measurements in November, 48% in December …
L327: increases by
L361: I believe you mean delta SF?
L389: and the snow height decreases
Figure 13 caption: ‘an episode of erosion’
L425: shortwave
L460: on days
References:
Libois, Q., Picard, G., Arnaud, L., Dumont, M., Lafaysse, M., Morin, S., and Lefebvre, E.: Summertime evolution of snow specific surface area close to the surface on the Antarctic Plateau, The Cryosphere, 9, 2383–2398, https://doi.org/10.5194/tc-9-2383-2015, 2015
Carlsen, T., Birnbaum, G., Ehrlich, A., Freitag, J., Heygster, G., Istomina, L., Kipfstuhl, S., Orsi, A., Schäfer, M., and Wendisch, M.: Comparison of different methods to retrieve optical-equivalent snow grain size in central Antarctica, The Cryosphere, 11, 2727–2741, https://doi.org/10.5194/tc-11-2727-2017, 2017
Citation: https://doi.org/10.5194/tc-2022-236-RC1 -
AC1: 'Reply on RC1', Sara Arioli, 31 Mar 2023
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2022-236/tc-2022-236-AC1-supplement.pdf
-
AC1: 'Reply on RC1', Sara Arioli, 31 Mar 2023
-
RC2: 'Comment on tc-2022-236', Anonymous Referee #2, 17 Feb 2023
After reading this nice analysis on snow grain size behavior adjacent to Dumont D'Urville station, it is clear that I do not have sufficient expertise in this topic area to provide useful scientific input. On page 22, the authors conclude that snow drift impacts at D17 prevent significant changes to snow grain size due to competing effects. That is, there is relative stability to the surface albedo.
Here are some small items:
Lines 37-39: This sentence seems to be saying contradictory things. Higher precipitation should probably be associated with higher albedo.
Line 41: Drop "operated".
Line 77: Automatic rather than Automated.
Line 113: Rephrase to "buried in the snow to provide temperature stability for the electronics".
Line 211: "obstructing" rather than "obturating" would be better as much more frequent usage and less jarring.
Figure 9: Caption says "The shaded gray area marks dopt > 0.64 mm". The gray shading along the top of each panel is continuously gray
and does not coincide with dopt > 0.64 mm in most locations.
Line 543: Complete the reference for Colbeck 1973.
Citation: https://doi.org/10.5194/tc-2022-236-RC2 -
AC2: 'Reply on RC2', Sara Arioli, 31 Mar 2023
The comment was uploaded in the form of a supplement: https://tc.copernicus.org/preprints/tc-2022-236/tc-2022-236-AC2-supplement.pdf
-
AC2: 'Reply on RC2', Sara Arioli, 31 Mar 2023
Sara Arioli et al.
Sara Arioli et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
388 | 127 | 20 | 535 | 10 | 8 |
- HTML: 388
- PDF: 127
- XML: 20
- Total: 535
- BibTeX: 10
- EndNote: 8
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1