Automated Detection of Glacier Surges from Sentinel-1 Surface Velocity Time Series—An Example from Svalbard
Round 1
Reviewer 1 Report
Since the 1920s, the end of the little ice age, the surface mass balance of the archipelago is negative. The Arctic archipelago of Svalbard in the North Atlantic is at roughly 57 % glacierized and houses a variety of glacier types like cirque and valley glaciers, as well as 183 ice fields and ice caps. Surge type glaciers cover ~50 % of the glacierized area of Svalbard. It is presented in 16 pages with 42 references, 6 figures, 3 tables. The topic is high interesting.
Comments:
1 km - space between number and km (L185) and m too (Table 2)
L65 Sentinel-1 SAR - SAR is an abbreviation, which is not known. Please, explain.
Fig. 1 and 6: captures. Top Axis - 100 O? Maybe W?
In the title I see right spelling "time series". However, in the text wrong - timeseries. Please, keep it "time series" everywhere.
Fig. 3 please, put after the mention in the text.
Table 1, 2, 3 please, look into MDPI style. Also, please put "threshold" down as a whole entire word.
L. 359, 379 wrong number. So, please, check in the text too. Table 2 and 3.
Fig. 5 bottom axis is hardly visible. Please, enlarge it.
L115-116, L155 [25] propose, [4] suggest... I think, either authors [25] suggest, or paper [25] suggestS.
I suggest citing or comparing results with Surge dynamics in the Nathorstbreen glacier system, Svalbard
- April 2014
- The Cryosphere 8(2):623-638
- License
- CC BY 3.0
- Monica Sund
- Tom Rune Lauknes
- Trond Eiken
Make a short review of surge areas, citing https://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XLIII-B3-2021/503/2021/isprs-archives-XLIII-B3-2021-503-2021.pdf .
I suggest making a title more specific. As there are quite previous publications with the same idea, but using other satellites, e.g. Julian A. Dowdeswell & Toby J. Benham"A surge of Perseibreen, Svalbard, examined using aerial photography and ASTER high resolution satellite imagery". So n my opinion, in your title you could add Sentinel-1.
For the Karakorum. I suggest citing publication from MDPI Ke, L.; Zhang, J.; Fan, C.; Zhou, J.; Song, C. Large-Scale Monitoring of Glacier Surges by Integrating High-Temporal- and -Spatial-Resolution Satellite Observations: A Case Study in the Karakoram. Remote Sens. 2022, 14, 4668. https://doi.org/10.3390/rs14184668.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
1. Half of the Abstract is about defining the glacier surge. It needs to be rewritten.
2. The writer’s name is not mentioned with reference. This makes it difficult to understand the “About glacier surges” paragraph.
3. Figures 2, and 3 are not readable properly. Increase the font size.
4. In line 296 equation number is not mentioned.
5. In L344 and L359 both tables are mentioned as Table number 1. The table is not well categorized. L346 “In table 1-3 darker red equals smaller correlation coefficients, while darker green indicates the highest” please check the tables for the colour scheme and provide the legend along with the table too.
6. L376 mentioned table 3, which is written wrong.
7. The Gantt chart in Figure 5 is not clearly visible, and the legend is also missing here.
8. Figure 6 “Distribution of active Surges detected in this study from 2015-2021 (dark blue)” in this map, locations mentioned within are not clearly visible.
9. A table with the quantified value of glacier surge for surging glaciers will add value to the paper and research
Also, check the attached PDF for some minor corrections.
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
In this study, the authors use data from Sentinel-1 RETREAT data set and a time series outlier detection algorithm, to determine not only surge-type glaciers but also the time of surge onset. To demonstrate the viability of this method, the authors have conducted detailed sensitivity studies for the algorithm, and identified 18 glacier surges and the timing of their active phase. The manuscript suggests a new quantifying approach of surging activities, which can be popularized to surge-type glaciers in other regions. We think that the study is generally suitable for the journal Remote Sensing, but some issues still need to be addressed.
The major comments:
1. In Section 5. Rsults, it is necessary to compare the detected surge events and the timing of detected surges between previous work and the results of this paper, in the form table. You have mentioned the previous work in Discussion, and readers prefer to browse comparison directly in the article.
2. Sentinel-1 RETREAT data is a very welcomed data set, which is independently from weather conditions and sun illumination, but the length of the time series is shorter than those generated from Landsat-8 optical (ITS_LIVE, GoLIVE) data. In Section 6. Discussion, it is shown that the only potential misclassification in this study is Stonebreen. The probable reason is that Stonebreen started surging before September 2015, so it may be helpful to use ITS_LIVE as supplement data, in order to detect surge events before September 2015.
Detail comments:
Line184: “More than 1.000 individual glaciers”, please identify the thousand separator to prevent misunderstanding. In the United States, this character is a comma (,). In Germany, it is a period (.).
Line188: Please add reference to this sentence “Since the 1920s, the end of the little ice age, the surface mass balance of the archipelago is negative.”
Reference
1. Dowdeswell, J.A.; Hodgkins, R.; Nuttall, A.-M.; Hagen, J.O.; Hamilton, G.S. Mass balance change as a control on the frequency and occurrence of glacier surges in Svalbard, Norwegian High Arctic. Geophys. Res. Lett 1995, 22, 2909-2912, doi: 10.1029/95GL02821.
Line246-248: Figure 2. is somewhat ambiguous, and it is recommended to use a higher DPI plotting method.
Line250: Function na.approx() can be used to replace missing values by linear interpolation. Is it better to use “interpolate missing values” than “close data gaps”?
Line332-333: As shown in Figure 4., when higher ‘a’ is set, more data points outside the envelop are classified as an outlier. So higher values of ‘a’ are more prone to incorrect classifications of “normal” observations?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
