Estimating Regional Snow Line Elevation Using Public Webcam Images

Round 1

Reviewer 1 Report

The work is devoted to the analysis of images collected by a network of 16 public webcams to the main purpose of complementing the outcomes provided by satellite which are affected by spatial or temporal resolution limitations (e.g. Moderate Resolution Imaging Spectroradiometer (MODIS) can offer a daily coverage but a relatively low spatial resolution of 500 m, while Sentinel-2 provides spatial resolution of about 20 m but a 5-day revisit. The use of "public webcams" for monitoring snow cover is not new. Several scholars have worked in this direction and also the authors of the manuscript have already published several works on this same topic. Thus manuscript is weak from the point of view of novelty. Furtheromore it is a bit too oversimplified the idea that a network of 16 webcams can replace or yield more accurate description of the snow line elevation. It is true that the spatial resolution of satellite is limited, however they provide a continuosly defined grid of values over a extended region. The use of the "NDSI snow cover index" with its well-know limitations due to the arbitrariness of the constant thresholding of the images is also a bit disappointing for the reader. In particular as the constant threshold adopted for the NDSI might be far from the value at the very specific position of the camera. More accurate snow map estimates can be obtained by using alternative snow indexes which involve multiple spectral components of the satellite imagery (see below for example Ref. [1] and cited works therein). Nevertheless the work has some merits and in my opinion deserves attention by the scientific community from the methodological point of view. In particular the manuscript includes a robust statistical analysis of the correlation between the webcam data and the MODIS and Sentinel 2 and the effect (presence/absence) of clouds (which are another limitation of the satellite records). The method of cross-correlation between the local (webcam) and global (satellite) grid data might be indeed very useful for the continuous calibration of the webcam public networks to ensure they operate in the best condition to yield accurate data. Interesting developments might be devised if data from local sensors on specific spectral bands could be correlated (this would require for example IR camera). The advantage of the local photography is the possibility to use the collected data (most of them in visible region of the EM spectrum) with high temporal frequencies. Daily cycles (for example through the detection of hourly spaced data) of snow metamorphism (see for example Ref. [2] and cited works therein) at locations critical for safety (due to high risk o ice/snow avalanche) might be detected and cross-correlated with other data (satellite images or other features). I would therefore reccomend the authors to revise introduction and conclusion to redirect the focus of the reader towards the cross-correlation methodology, to outline novelty and outcome of the work indicating possible future developments, as for example using a network of different webcams validated through satellite images via snow cover indexes more accurate than NSDI or using the local photography to monitor in quasi real time snow metamorphism. All these aspects should be at least discussed by the authors even if by now they have not yet included results in this work that might be anyway considered for future developments of the proposed method. [1] Non-binary Snow Index for Multi-Component Surfaces, MM Arreola-Esquivel et al. Remote Sensing 13 (14), 2777 (2021) [2] Pinzer, B. R., & Schneebeli, M. (2009). Snow metamorphism under alternating temperature gradients: Morphology and recrystallization in surface snow. Geophysical research letters, 36(23).

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear Editor and Authors, a nice paper, well worked in any aspect. I do not have specific comments, to me the paper can be published.

Author Response

Dear Reviewer 2

Thank you for your feedback.

Reviewer 3 Report

This is an innovative paper that is interesting and also important: public webcams are ubiquitous and the use of the technology for mapping snow is novel.  The results are compared with remote sensing images and the technology is found to be effective.  I strongly recommend publication of this paper, but I have a few minor comments that should be considered before publication.

line 91: what is 50’855 in the context of images?  Does this refer to 50855 images?

lines 93-103: The authors should provide an example of the image processing pipeline in reference to this paragraph.  What is the threshold?

line 116: Can the idea of "computationally intensive" be quantified here?

line 186:  What is meant by "representative"?

line 450: Can "great agreement" be quantified in the context of this sentence?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

The paper title “Estimating Regional Snow Line Elevation using Public Webcam Images” evaluates the usefulness of images obtained with 16 webcams located in the study area, three hydrological basins of Switzerland, to calculate the altitude of the snow line in a period between October2017_June2018. The authors relate the altitude of the snow line obtained with the images of the webcams with the average value obtained from MODIS and Sentinel_2 images. The differences are between 53 and 56 m depending on whether Sentinel-2 or MODIS is used. In addition to the snow line, the authors explore the interference of cloud cover in the identification of snow, finding that the use of images from webcams minimizes this problem. The paper address relevant scientific questions within the scope of Remote Sensing Journal. Monitoring the snow cover using optical images has a serious problem with cloud cover. This means that despite the high temporal resolution of some sensors, time series are not available.  For this reason, the use of time-laps cameras for the observation of the snow cover has become widespread. Despite the low resolution of the images obtained, these are of great value in remote areas of difficult access and adverse weather. Public webcams are used in this work what makes the work novel and interesting. I think that the work is well structured and the syntax allows its understanding and reading. The figures and tables have high quality and are suitable for illustrating the work. Citations are sufficient, current and interesting for this work. However, I think the work has some problems that make it unsuitable for publishing in the current form, needs revision.

Main concerns

1.     The main problem is related to the traceability of the method. In section 2.1 (line 105-114) the procedure for georeferencing webcam images is not sufficiently explained. Authors use the name of commands or methods (pinhole camera model) that need further explanation. Why haven't the authors used a Gis tool to georeference the images? I think that in this case, this phase of the work is very important since it supports all the results obtained. The authors must make it very clear how they have related the DEM with the webcam images and what limitations the method used has. The description of experiments and calculations must be sufficiently complete and precise to allow their reproduction by fellow scientists (traceability of results).

2.     In the same way, the authors name commands (line 92) that belong to a specific software but are not said the software name. I think the authors should list the programs they have used to do this work.

Minor concerns

1.     The authors have chosen the images of the webcams (one per day and per webcam) based on their quality and sharpness. However, the cloud cover shows a very high temporal variability. In order to compare the influence of cloud cover on the utility of the different images (Webcams, MODIS, Sentinel-2) the authors should have chosen the images of the webcams by the time in which they were taken. It is very important that the hour of webcam images coincide with the exact time at which the satellite takes the image. This does not invalidate the study, but does not allow comparisons to be made about the influence of cloud cover on the utility of the different images.

Finally, I think that the conclusions of the work would have been more robust if, in addition to calculating the snow line, the pattern of the snow cover had been studied. Only by using two types of snow cover (continuous and discontinuous), could the usefulness of webcam images and the accuracy they could add to the study be more objectively evaluated.

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have satisfactorily addressed the issues raised in my previous report and in particular clarified the limits and significance of the proposed method.

Author Response

We would like to thank reviewer 1 for this feedback.