Exploring Imaging Methods for In Situ Measurements of the Visual Appearance of Snow

Round 1

Reviewer 1 Report

Brief summary:

The manuscript provides a description of a field campaign, where three different sites were studied during 19 days in February-March 2023. Images were taken with a camera and reflectance was measured with TLS850 device including an artificial light source. Additionally, very simple snow classification was made. Amount of sparkling of snow was estimated with sparkle contrast and sparkle density in the images. Authors tried to connect the estimated snow type with those parameters by using a classification algorithm resulting large classification error indicating that there is no evidence on that. The second part of the study estimated absorption and scattering coefficients from the reflectance measurements, which were differing from the previous studies by others. They also tried to connect the estimated snow type with the coefficients by using the same classification algorithm as in the first part of the study resulting again that connection is not evident. Even though the study was not able to show connection between estimated snow type and sparkle or scattering coefficient, it is potential for future studies when more detailed observations could be performed. 

General comments:

This study was lacking proper analysis of snow microstructure. For example, using specific surface area instead of grain type could provide more favourable results in the future. Additionally, measuring other physical properties, such as density and wetness, could help concluding the results. It would be important also to measure temperature in the measurement location to catch the snow metamorphism. Additionally, sparkle measurements could be complemented with irradiance measurements so that changes in solar radiation could be detected, especially if there are any thin clouds.

Reflectance is related to grain size, shape and orientation in the snow surface. It could be interesting to measure sparkle also with ~1000-1300 nm nir wavelengths (near infrared camera) since it is more sensitive to grain size than used visible light wavelengths. Similarly, it would be interesting to use nir wavelengths also for the absorption and scattering study. Moreover, it would be interesting to compare the observed sparkle with albedo measurements.

You should address these points to improve the manuscript:

·     I would recommend some restructuring for the manuscript since the main results and conclusions are not clearly presented. You could concentrate presenting field campaign, measurement methods, data and different aspects affecting results (e.g. focal length) with details since it is evident that new field campaign is needed to connect sparkle with snow microstructure. Demonstrate also what is new compared to the previous publication [21] of the method.

·        You could describe the field observations with more details. Location, description of the area, photos, location of temperature data compared to field campaign location could be provided. Overview pictures of snow scenes (field sites) would be good.

·        You could add more detailed analysis related to temperature changes and observed sparkle contrast (lines 176-189) by using hourly resolution temperature data.

·        Did you noted that was TLS850 device warming during the measurement so that there would be potential for melting of surface snow? If yes, please add this to the discussion. Did you noticed any change in surface grains due to the pressure of the TLS850 instrument? Like compression or breaking the grains? If yes, please add this to the discussion.

Specific comments:

Line 1: Change as “We explored imaging methods to perform in-situ measurements of snow physical properties by using visual appearance of snow.”

Line 2: What is “a single reflectance measurement”?

Line 3: Change as “We also measured sparkle indicators, contrast and density of the sparkles, from digital images of snow.”

Line 4: Change as “In the case of sparkle of snow surface,”

Line 4-5: Change as “We found that there is a potential to use the sparkle indicators for classification of grain type, but it requires larger data set coupled with grain types defined by an expert.”

Line 8: Explain diffusion theory.

Line 8: It is well known that snow albedo is high, it could be mentioned here or in the introduction chapter.

Line 8: “These features were acquired on several 8 days across winter on similar locations.“ -> “Measurements were made in several days in February-March 2023 on three locations in Norway. “

Line 22-26: Text is repeated, remove.

Line 44: which measurements?

Line 71: Change as “The following sections will present used imaging methods for each studied parameter.”

Line 76: Change as “We performed our in-situ data collection in Norway on three snow sites between February and March 2023.”

Line 77: Change as “The three sites selected were kept across all measurements to ensure a consistency and a point of comparison.“

Line 78: “A reason to explain this choice is to monitor changes due to temperature and new fallen snow on the indicators we aim to measure” -> ”We monitored changes on the measured parameters due to temperature and newly fallen snow.”

Figure 2: “Snow images taken during our acquisition campaign.”  -> “Snow images taken during our field campaign.”

Line 84: “The camera was held by a human user from their height, and so the camera was not fixed on a tripod. “ -> “The camera was not fixed on a tripod and it was held by a user from their height.”

Line 85: Did you recorded direction of the sunlight? was it the same in every image?

Line 87: Explain what is graininess.

Line 88: Change as “Examples of the images are shown on Figure 2. In some of them, it is possible to observe some part of graininess.”

Line 95: Explain a variant set.

Line 97: Explain grazing angle.

Line 99: Change as “With these image sets, we aim to establish an uncertainty scale on the sparkle indicators, the contrast and the density of sparkle spots, defined in our previous study [21]. “

Line 102: Change as “absorption and scattering coefficients of snow,”

Line 114: Change as “For the case of snow, it is important to be in contact but not let the weight of the sensor to compress the snow and affect the snow structure.”

Line 120-122: Sensor wavelengths could be mentioned already in line 106.

Line 136: RTE, radiative transfer equation?

Line 139: What has been used before, the instrument or some equation?

Line 142: Change as “Field campaign”. Consider to add more detailed description of the campaign and move it to line 72.

Line 143: Remove the first sentence “Studies on snow are seasonal and highly dependent on the amount of snow one can have access to, especially in the case of in-situ studies.”.

Line 148: “One reason behind the reduced amount of day for the data is due to the specific weather conditions required.” -> “Specific weather conditions were required for data collection.”

Line 150: Why use diffuse radiation is not recommended?

Line 151: “For optical properties,” -> “For absorption and scattering,”

Line 152: There are changes in snow microstructure even within the same day.

Line 154: “42 days of acquisition” -> “42 measurement days”

Line 156: Change as “We compiled three categories for the types of snow based only on our observations shown in Table 1. Three categories are considered and determined: fresh snow (F) when there has been a snow fall recently, dense snow (D) when snow has been densified in cold temperatures for some time without new snow, and old snow (O) for snow which melting process has started.”

Table 1. “Snow labels for each of acquisition.” -> “Snow type for the measurement days.”

Line 163: “each day” -> “each day of field measurements.”

Line 165: Estimated temperature gradient was not used in the study. Remove “, to estimate the gradient of temperature that the snow had”.

Line 169: Contrast and density of sparkle could be defined already in the introduction.

Figure 5. “Tracking of temperatures for the days of acquisition” -> “Temperatures for the measurement days.” How far is the temperature measurement location compared to the field site? You could plot temperature with better resolution, especially if you would like to check some part with more details.

Line 180: “acquisitions” -> “field measurements”

Line 182: Remove “This is when this trend on the contrast indicators are observed.”

Line 185: “It has been reported that temperature has an effect on the snow [4], 185 and due to higher temperatures, snow grains may have accumulated to form larger grains. 186 Due to sudden cold temperatures, the melting process did not start and thus the surface of 187 snow grain able to reflect light might have been bigger.” -> “ Temperature has an effect on the snow metamorphism and therefore grain type [4]. Due to positive temperatures, snow grains grow fast larger in melt metamorphism. Due to following cold temperatures, the melt water refreezes and the surface of snow grains might be are able to reflect light better.” Is this correct? If the daily maximum temperature has been positive, surface probably have been melted. It would be great to have better information on grain size, shape, wetness and hardness to be able to estimate reasons for the changes in the contrast.

You should check temperature variability within hours close to the increase in the sparkle contrast. Possibly surface hoar formation was possible? On the other hand, possibly some grain type classification is possible from the images, you could add an example from each measurement day.

Lines 191-211: How changing imaging distance would affect? Is it the same as having larger focal length? Would it be better to have a constant distance in the future study? If yes, it could be added to discussion.

Line 198: You could define when which focal length was used, for example add a table or add it to table 1.

Line 199: “So we obtained images for which the zoom was more important than others, which results in snow grains being bigger and sparkle easier to detect for the algorithm.” What this mean? Did you select some of the images or what? Grains are bigger only compared to image size but actual size is the same.

Figure 6. Change as “Box plots for the computed sparkle contrast indicator in the three snow sites.” Remove “For each day, we collected images and computed indicators of contrast and density.”

Figure 7. Change as “Box plots for the computed sparkle density indicator in the three snow sites.” Remove “For each day, we collected images and computed indicators of contrast and density.”

Lines 208, 216: Remove “human”

Figure 8: Legend is not good, please remove and add information to the caption. Indicate also meaning of orange and blue colors for different focal lengths.

Line 244: “label of snow” -> “snow classification”

Line 245: Explain raw observation.

Line 245: “This labelling” -> “This classification”

Line 268: What are milk samples? explain.

Line 273: Remove “material”

Line 298: You could conclude also here something like “This error value does not allow concluding on a classification of snow with scattering coefficient”.

Line 299: “our labelling” -> “our snow classification”

Line 302-314: It is unclear for me, what is aim of the rendering study?

Line 306: Explain what are rendered images.

Line 309: It is not clear what is “a 3D object modelled”. Line 314 explains a bit but it could be explained earlier.

Line 324: Conclusion includes quite much discussion, consider to restructure so that there is a separate discussion chapter.

Line 324: Discussion could mention that the covered area in the photos were not constant due to use of different focal lengths. Have you estimated how large impact that has?

Line 325: “physical correlates of the snow visual appearance” -> “physical properties correlating with the snow appearance”.

Line 326: “For the case of the sparkle correlate,” -> “For the case of the sparkle of snow,”

Line 327: “the method of acquisition and detection” -> “the measurement method”

Line 329: “indicator for snow” -> “indicator for snow type.”

Line 331: What means “content dependent”?

Line 334: “labelling for classification” -> “grain type classification”

·        Language needs improvements since the text is not easy to read and sometimes meaning is difficult to understand. Especially, I recommend to rewrite the abstract.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This is a really fascinating study, and offers insights into some new ways to look at snow.  I had never heard of "snow sparkle" being assessed in this way, or for being used to classify snow cover.  

Questions:  
1) are there similar studies in other parts of the world, or looking at snow with different water contents, age, density, etc. that could expand your background a bit?  I did find the introduction a bit light on breadth of studies, but depth was fine and you reviewed methods just fine. 
2) How does this type of classification link to remote sensing methods?  Could sparkle eventually be characterized with algorithms for drone or aerial-based surveillance?   
3) Do you have any spatial survey data at your site(s)?  I'd be interested to see plots of how your observations vary over a fixed plot (if it exists), particularly where snow morphology may have been impacted by wind forcing or other.  I recognize this may be beyond the scope of your field dataset. 

Otherwise, very good work!  I didn't detect any major issues with the writing, but as always, I encourage a careful proofread on final submission.  I'm happy to recommend this for publication, whether you take my questions/comments into consideration or not. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Modifications made improved the text and I suggest to consider only these minor changes:

Abstract:

Lines 1-3: We explored imaging methods to perform in-situ field measurements of physical correlates of visual appearance of snow. Measurements were performed on three locations in Norway between February and March 2023. We used a method to estimate absorption and scattering coefficients of snow by using only one measurement of reflectance captured by TLS850 translucency meter.

Line 4: We also measured sparkle indicators (contrast and density of sparkle spots) from digital images of snow. The contrast of sparkle spots can be defined as the median value of all the pixels identified as sparkle spots by an algorithm. The density of sparkle spots would thus be the number of sparkle spots on a given area in the image.

Lines 6-8: For the absorption and scattering properties, the measurements confirm the fact that snow is a weakly absorptive and highly scattering material when considering modelling light interactions in the snow.

Line 9: Consider to move text to the beginning as suggested above.

Tables and figures:

Table 1. replace "x" with "-"

Figure 5. 10 km is quite long distance for temperature, since depending on local topography, there might be huge temperature differences. For the future, my recommendation is to make temperature measurements on the site at least simultaneously with your measurements if continuous measurements are not feasible. Then you could compare how well the station is representing temperatures at your site.   

Figures 6 and 7. Change label "scene" to "site". Additionally, is there units for contrast and density? If yes, those should be added to all relevant figures and tables.

Readability could benefit of language check.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf