the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 15 Mar 2021
Evaluation of the global glacier inventories and assessment of glacier elevation changes over north-western Himalaya
Abstract. The study evaluates the global glacier inventories available for the study area viz., RGI, GAMDAM and ICIMOD, with the newly generated Kashmir University Glacier Inventory (KUGI) for three Himalaya basins; Jhelum, Suru and Chenab in the north-western Himalaya, comprising of 2096 glaciers spread over an area of 3300 km2. The KUGI was prepared from the Landsat data supplemented by Digital Elevation Model, Google Earth images and limited field surveys. The KUGI comprises of 154 glaciers in the Jhelum, 328 in the Suru and 1614 in the Chenab basin, corresponding to the glacier area of 85.9 ± 11.4 km2, 487 ± 16.2 km2 and 2727 ± 90.2 km2 respectively. The investigation revealed that most of the glaciers in the study area are < 1 km2 in size, however, the glaciers in 1–5 km2 size class cover most (55.8 %) of the glacier area. Majority of the glaciers, both in terms of number and area, are at 4500–5500 m asl except in the Jhelum where the glaciers are mostly situated between 4000–5000 m asl altitude. The glaciers in the three basins mainly harbor slopes ranging from 10–30°. It was also observed that the southern aspects host more number of large-sized glaciers than the northern aspects. Comparative analyses of the inventories revealed that the GAMDAM (RAB = 0.75) and RGI (RAB = 0.73) inventories are consistent with the KUGI. However, discrepancies were observed in the debris-covered and shadowed glaciers particularly in the ICIMOD inventory. The glacier elevation changes were also estimated for glaciers in the three basins using the Tandem-X and SRTM-C DEMs from 2000 to 2012. The investigation revealed a strong control of glacial morphology, topography, and debris cover on glacier thinning. Glacier elevation change of −1.33 ± 0.8 m a−1 was observed in the Jhelum basin but a similar glacier elevation changes of −1.08 ± 0.7 m a−1 and −1.09 ± 0.8 m a−1 was observed in the Suru and Chenab basins respectively. Evaluation of the glacier inventories and assessment of glacier elevation change in the data-scarce Himalaya, reported in this article, would constitute a reliable database for research particularly in hydrology, glaciology, and climate change. The dataset is freely available at http://doi.org/10.5281/zenodo.4461799 (Romshoo et al., 2021).
- Preprint
(1606 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on essd-2021-28', Anonymous Referee #1, 08 Apr 2021
The authors present a new glacier inventory for north-western Himalaya, which is mainly based on manual glacier mapping using different data sources. They compared the new inventory with existing inventories and figured out limitations and differences of the individual inventories. Additionally, the authors used InSAR DEMs to compute glacier elevation changes between 2000 and 2012 of the study area.
The paper is well structured but the sections regarding the glacier elevation/mass change computations are very unclear and confusing. The authors are talking about mass balances but never provide any mass balance values. There are many flaws in the respective sections. Thus, I would suggest to remove the elevation/mass change computation sections completely, since the main focus of the paper is the evaluation of the glacier inventories.
Moreover, the results sections are too long and can be strongly condensed by focusing on tables and graphs.
At many places it is unclear, if the authors talk about mean/media values of certain variables (e.g. elevation, slope ….) or pixel wise values. A more precise wording is needed throughout the paper. (see details below).
The computation of several “uncertainty” values is unclear. Please provide formulas (see details below)
The computation of the average aspect values is unclear and might be buggy (see details below)
The comparison of the different glacier inventories is OK but can be certainly extended. It would be interesting to compute the the overlap ratio r_ov also for e.g. DC, clean and shadowed glaciers to evaluate the difference between the inventories.
It is also unclear, if the topographic parameters of the other inventories were taken form the inventory meta data or computed by the authors. The used DEMs might differ. Thus, it would be more meaningful to use a consistent source for topographic information before doing the comparison.
Once, the paper is revised it should be properly proof read. I am not a native speaker, but I got the feeling that the English can be improved. Many sentences are quite complicated and unclear or maybe got just grammatical errors.
Detailed comments (* significant issues):
L31: delete “for the study area”
l35: by “a” Digital…
l37: to glacier areas…
L40: are you talking about the mean or median glacier elevation?
L44: whats the meaning of the “R” values. Completely unclear.
L48: 2000 an 2012
L71: what about Brun et al. 2017
L97 and following: please list here more recent publications
l104: please list some of the variables
l106: why is the reproducibility not assured? Not clear.
L109: what about Brun et al 2017, Shean et al. 2020?...
L110 and following: please move the comparison to the discussion section.
L118: which basins and where? Not introduced
L123: there exist already elevation change data sets for the same period (Brun et al.2017, Shean et al. 2020). So there exists already information on the glacier behavior.
L125: please rephrase this sentence. A quite weak motivation for this study.
L130: UIB not introduced
L132: “and” 73….
l136: when is this area covered? All year long?
Fig1: Please provide country borders and names in the overview map (upper right corner) for a better orientation. Please indicate the glacier coverage also outside the 3 basins. What are the sources of glacier outlines, debris cover and glacier volume?
*L154: are you talking about mean or median altitudes? Not clear, the same for the other basins in the following.
L163: … in the northeast of the study area..
L192: ...use of…
*Table1: could you please add the Path and Row numbers of the Landsat data. ASTER GDEM not listed. URL for ICIMOD inventory is missing., please provide also the date ranges of the inventories for your study area
l208: add “C-band”
l211: please introduce the abbreviation “DEM” at the place, where it is used the first time.
*l209: please rephrase. TanDEM-X is still acquiring data. You are talking about the worldDEM phase
l234: between or only in 1999 and 2003
l298: no capital letters for Base and Target
Section 4.3: This section is a bit unprecise and many details are missing. e.g. which DEMs did you use? How did you estimate the penetration bias
l350: cite here Rolstad et al. 2009
*l352: Seehaus et al. 2020, did not use the total glacier area for A. They used the area of each glacier complex.
Section 4.4. b) This section is quite confusing and the equation to compute the uncertainty of the mass balance is missing. Please revise the whole section and use clear and individual variables!
*l365: How did you compute the glacier volume ? Not mentioned in the Methods Section
Table 2: How did you compute the glacier volume? Why does it differ so strongly e.g. between KUGI and RGI at Jhelum? How did you assume the uncertainty in glacier area for the different inventories? Not explained!
380 and following: are you talking about mean or media elevations? Or the total elevation span of the whole glacier?
Table 3,4,5,6,7: Units are missing. What means “A” and “N” and “DC”? not clear
l396. delete sentence. Already mentioned in the methods
*Section 5.1: The whole Section can be strongly condensed. All information can be found in the tables and does not need to be repeated in the text.
*Table 6: how did you compute the average glacier aspects? Please provide the formula somewhere.
l483: how did you estimate the variations? Not explained!!!
*Section 5.2: Same as for Section 5.1.! It can be strongly condensed and most of the information can be summarized in nice tables and/or graphs. The text is very long and the information is hard to find. Tables and graphs would be beneficial for the reader
*Table 10: How did you define the elevation category? All pixels with in the interval? Or all glaciers with mean/median elevation within this interval? Unclear! How did you compute the uncertainties?
Fig. 2,3,4: Please add a background. The glacier outlines would be also nice. The bar plot is too small an impossible to read. Does it show the mean elevation changes per glacier? Explain!
l629: Are you talking about average elevation changes per glacier? Please clarify.
Table 11: is the slope take pixel by pixel or is it base on the mean slope per glacier?
Table 12: same as for Table 11. is the aspect take from each pixel or the mean of each glaciers
l659: Maybe the bigger glaciers are located at lower altitudes? Please check
Table 13: Units are missing for area
l685: By inspecting Fig. 2-4, it looks like most glaciers are not south facing. Please check your aspect computation! Do not use a simple mean of all pixel wise aspect values of a glacier. See RGI6.0 technical report.#
Fig.5: Date and source of background image?
Fig.6: Date and source of background image? There are at least 2 glacier tongues. Most likely they were connected in the past, but the mapped state shows 2 individual major glacier tongues. Therefore it is not wrong to split the glacier are in 2 polygons. Please rephrase accordingly, also in the main text.
Fig. 7: the outlines are hard to see. Use different colors or wider lines. Date and source of background image?
Table 16: Can be merged with Table 2 to avoid doubling of data.
l794: Unclear sentence
Citation: https://doi.org/10.5194/essd-2021-28-RC1 -
AC1: 'Reply on RC1', Romshoo Shakil Ahmad, 08 Jun 2021
Reviewer #1
General Comment: The authors present a new glacier inventory for north-western Himalaya, which is mainly based on manual glacier mapping using different data sources. They compared the new inventory with existing inventories and figured out limitations and differences of the individual inventories. Additionally, the authors used InSAR DEMs to compute glacier elevation changes between 2000 and 2012 of the study area.
The paper is well structured but the sections regarding the glacier elevation/mass change computations are very unclear and confusing. The authors are talking about mass balances but never provide any mass balance values. There are many flaws in the respective sections. Thus, I would suggest to remove the elevation/mass change computation sections completely, since the main focus of the paper is the evaluation of the glacier inventories.
Moreover, the results sections are too long and can be strongly condensed by focusing on tables and graphs.
Response: We express our gratitude to the reviewer for the elaborative and useful section-wise and line-by-line detailed review of the manuscript. We have responded point-by-point to all the comments and suggestions of the reviewer. The concerns of the reviewer regarding elevation change have been addressed entirely to the best of our ability and knowledge. The valuable comments and suggestions provided by the reviewer have greatly helped to improve the contents and quality of the manuscript.
Furthermore, we agree with the reviewer that the manuscript is focused on the evaluation of the glacier inventories, however, we believe that the elevation change information of the three basins will add value to the database, providing, in addition to the glacier inventory, a quick overview of the glacier elevation and mass changes of the glaciers in the database. As such we find merit in retaining the elevation change section while addressing the concerns of the reviewer about elevation change estimation. The point-by-point response to the detailed comments and suggestions raised by the reviewer is provided as follows:
Comment #1: At many places it is unclear, if the authors talk about mean/media values of certain variables (e.g. elevation, slope ….) or pixel wise values. A more precise wording is needed throughout the paper. (see details below).
Response: We have used the mean value of the topographic variables, obtained by averaging single-cell elevation, slope and aspect values from the DEM. This has been specified in the revised manuscript.
Comment #2: The computation of several “uncertainty” values is unclear. Please provide formulas (see details below)
Response: Clarification/correction provided wherever suggested in the revised manuscript. A detailed response is provided against each specific comment below.
Comment #3: The computation of the average aspect values is unclear and might be buggy (see details below)
Response: As mentioned in the manuscript, the aspect was calculated by averaging the aspect value of DEM cells within the extents of each glacier. However, the aspect has been now recalculated in the revised manuscript as specified in the RGI technical document. The aspect sines and cosines of each of the glacier’s DEM grid cells were summed and the mean aspect was calculated as the arctangent of the quotient of the two sums. The calculated values of aspect were transformed to the eight cardinal directions (N, NE, E, SE, S, SW, W, and NW) with each cardinal direction having range of 45 degrees, half to each side.
Comment #4: The comparison of the different glacier inventories is OK but can be certainly extended. It would be interesting to compute the the overlap ratio r_ov also for e.g. DC, clean and shadowed glaciers to evaluate the difference between the inventories.
Response: Agreed, that the overlap ratio of different glacier classes will be more useful, however, it is pertinent to mention that the debris covered portions of the glaciers, particularly in the ICIMOD inventory, have been excluded from the glacier extents as such the same glacier classified as debris-covered in the KUGI gets classified as “clean” (as the debris covered area is excluded) in ICIMOD for example, thereby making it a bit difficult to find true overlap ratios.
Comment #5: It is also unclear, if the topographic parameters of the other inventories were taken from the inventory meta data or computed by the authors. The used DEMs might differ. Thus, it would be more meaningful to use a consistent source for topographic information before doing the comparison.
Response: Since the source DEMs for topographic information in the existing inventories are different, the topographic parameters for the KUGI and other inventories, evaluated in this study, were derived from the ASTER GDEM2, which is now specified in the revised manuscript.
Comment #6: Once, the paper is revised it should be properly proof read. I am not a native speaker, but I got the feeling that the English can be improved. Many sentences are quite complicated and unclear or maybe got just grammatical errors.
Response: Thanks for the suggestion, we have checked and revised the manuscript for any grammatical errors. We have revised and rephrased a few complex sentences in the revised manuscript for more clarity and better understanding. As suggested, we shall get the revised manuscript check by an English language editor/faculty in the University.
Detailed comments (* significant issues):
Comment #7: L31: delete “for the study area”
Response: The phrase “for the study area” has been deleted from the revised manuscript as suggested.
Comment #8: l35: by “a” Digital…
Response: As suggested, the Digital Elevation Model has been prefixed by the letter “a” in the revised manuscript.
Comment #9: l37: to glacier areas…
Response: As suggested “area” replaced by “areas” in the revised manuscript.
Comment #10: L40: are you talking about the mean or median glacier elevation?
Response: It is the mean glacier elevation and has been specified in the revised manuscript.
Comment #11: L44: whats the meaning of the “R” values. Completely unclear.
Response: The overlap ratio of individual glaciers is represented by “rov” whereas, the “R” (RBA) has been used to represent the average overlapping ratio of the base and target glacier inventories (overlap ratio averaged for all the glaciers in a particular inventory combination e.g, KUGI-RGI for individual basins). This has been described under the methodology section in the revised manuscript.
Comment #12: L48: 2000 an 2012
Response: The typo has been corrected as “…2000 and 2012” in the revised manuscript
Comment #13: L71: what about Brun et al. 2017
Response: The references has been incorporated in the revised manuscript.
Comment #14: L97 and following: please list here more recent publications
Response: More recent publications suggested in the “general comments” and several other relevant publications suggested by other reviewers have been incorporated in the revised manuscript.
Comment #15: l104: please list some of the variables
Response: The variables include glacier number and area, which has been specified in the revised manuscript.
Comment #16: l106: why is the reproducibility not assured? Not clear.
Response: Since different inventories use different approaches, datasets and at times the definition of the glacier itself varies among the inventories, therefore reproducibility is a challenge. Furthermore, in case of the manually or at times the semi-automatically delineated glacier boundaries, glacier area estimate will depend on the perception of the analyst as such the results are often not reproducible.
Comment #17: L109: what about Brun et al 2017, Shean et al. 2020?...
Response: We agree that the mentioned studies have reported elevation/mass changes over the Himalayan region and have therefore been cited at appropriate places in the revised manuscript. However, here we have specifically mentioned the studies where the dataset is publicly available.
Comment #18: L110 and following: please move the comparison to the discussion section.
Response: As suggested, the section has been moved to the Discussion section in the revised manuscript.
Comment #19: L118: which basins and where? Not introduced
Response: The basin names (i.e., Jhelum, Suru and Chenab) has been specified in the revised manuscript.
Comment #20: L123: there exist already elevation change data sets for the same period (Brun et al.2017, Shean et al. 2020). So there exists already information on the glacier behavior.
Response: Agreed that the elevation change studies already exist over the region, however, it is pertinent to mention that these studies are carried over a larger spatial domain. In the present study, we carried out the elevation changes at local scale, furthermore, we also tried to assess the impact of topographic and morphological parameters including glacier size, DC, elevation, slope and aspect on the elevation changes which is included in the database.
Comment #21: L125: please rephrase this sentence. A quite weak motivation for this study.
Response: Thanks for the comment. As suggested, we have modified the motivation for the research work in the revised manuscript keeping in view the following argument.
Primarily, the motivation for the KUGI is to develop a high-resolution glacier inventory with improved accuracy with visual interpretation and manual delineation of glaciers from Landsat satellite data supported by the limited ground truth and supplementing the glacier outlines with additional data like debris-cover, thickness changes and other glaciological parameters, that are either missing or incorrect in the existing databases so that the database is made available to the large research community for various applications.
Purportedly the global and regional glacier databases that were chosen for comparison in this study have been generated using a semi-automated method allowing less human error, quick delivery, and high accuracy. However, it was found in this study that there are significant errors in the evaluated databases due to the misinterpretation of seasonal snow cover particularly on the glacier headwalls at high altitudes, shadow-covered glaciers and debris-cover.
Keeping in view the worldwide use and applications of global and regional databases, it is important that a rigorous evaluation of these global and regional inventories is undertaken for the continued refinement of the methodology which is a fundamental requirement for any meaningful application of the global or regional database. It is hoped that the future releases of the databases will improve these and other shortcomings identified in this manuscript.
Comment #22: L130: UIB not introduced
Response: UIB stands for Upper Indus Basin and the full form of the UIB has been incorporated upfront in the revised manuscript.
Comment #23: L132: “and” 73….
Response: The word “and” has been inserted between the latitude and longitude values in the revised manuscript.
Comment #24: l136: when is this area covered? All year long?
Response: Most parts of the study area above 3600 m asl remain snow-covered for the entire or most of the year. The sentence has been modified accordingly in the revised manuscript for more clarity and instead of 3600 m, we have mentioned ~4000 m asl to include that all the areas covered with snow.
Comment #25: Fig1: Please provide country borders and names in the overview map (upper right corner) for a better orientation. Please indicate the glacier coverage also outside the 3 basins. What are the sources of glacier outlines, debris cover and glacier volume?
Response: Thanks for the suggestion. The figure has been modified accordingly in the revised manuscript. The glacier outlines and the debris cover information is based on the KUGI. Further, the glacier volume has been derived using the slope-dependent scaling approach with the glacier area and slope information derived from KUGI.
Comment #26: *L154: are you talking about mean or median altitudes? Not clear, the same for the other basins in the following.
Response: These are the mean elevations as mentioned earlier and this has been specified in the revised manuscript.
Comment #27: L163: … in the northeast of the study area..
Response: As suggested “east” has been replaced by “northeast” in the revised manuscript.
Comment #28: L192: ...use of…
Response: Sorry for the typo, the word “use” this has been corrected in the revised manuscript.
Comment #29: *Table1: could you please add the Path and Row numbers of the Landsat data. ASTER GDEM not listed. URL for ICIMOD inventory is missing., please provide also the date ranges of the inventories for your study area
Response: The suggested missing information pertaining to the data sets used in developing the glacier database has been incorporated in the revised manuscript. The images with path/rows 149/36, 148/36, 148/37, 147/37 dating between 2000 and 2002 have been used for the inventory development.
Comment #30: l208: add “C-band”
Response: As suggested the band information has been incorporated in the revised manuscript.
Comment #31: l211: please introduce the abbreviation “DEM” at the place, where it is used the first time.
Response: The full form of the DEM (Digital elevation Model) has been mentioned at the first occurrence (in abstract section) in the revised manuscript.
Comment #32: *l209: please rephrase. TanDEM-X is still acquiring data. You are talking about the worldDEM phase
Response: Agreed, TanDEM-X mission is still acquiring the data, the information provided in the manuscript is relevant for the product version of the DEM (used in the present study) released in 2018 only. The text has been therefore modified accordingly in the revised manuscript.
Comment #33: l234: between or only in 1999 and 2003
Response: It is between as the authors have used data sets dated 1999, 2000, 2001 and 2002 for the generating the glacier inventory.
Comment #34: l298: no capital letters for Base and Target
Response: As suggested, the words “Base” and “Target” have been uncapitalized in the revised manuscript.
Comment #35: Section 4.3: This section is a bit unprecise and many details are missing. e.g. which DEMs did you use? How did you estimate the penetration bias
Response: The missing details and further description pertaining to the methodology has been provided in the revised manuscript. The penetration bias was computed as a function of altitude after Vijay and Braun, (2016).
Comment #36: l350: cite here Rolstad et al. 2009
Response: The reference has been incorporated in the revised manuscript.
Comment #37: *l352: Seehaus et al. 2020, did not use the total glacier area for A. They used the area of each glacier complex.
Response: Agreed that Seehaus et al. (2020) used the area of each glacier complex. We also used the term to represent the analyzed area.
Comment #38: Section 4.4. b) This section is quite confusing and the equation to compute the uncertainty of the mass balance is missing. Please revise the whole section and use clear and individual variables!
Response: The uncertainty of glacier-wide specific elevation change (Δh) is computed as:
The , , and are uncertainty of DEM differencing, uncertainty due to void filling (since the DEMs especially SRTM has voids over the study area as such the DEMs coverage for each having voids >30% were excluded from the analysis whereas, the glaciers with <30% voids were filled with nature neighbor interpolation algorithm), temporal uncertainty of TanDEM-X and uncertainty of radar signal penetration respectively. The uncertainty of each of the individual parameter is described in detail in the revised manuscript.
Comment #39: *l365: How did you compute the glacier volume ? Not mentioned in the Methods Section
Response: The glacier volume was estimated using the slope-dependent volume estimation approach (Haeberli and Hoelzle, 1995), the methodology to estimate the glacier volume has been incorporated in the revised manuscript.
Comment #40: Table 2: How did you compute the glacier volume? Why does it differ so strongly e.g. between KUGI and RGI at Jhelum? How did you assume the uncertainty in glacier area for the different inventories? Not explained!
Response: As mentioned, we used the slope dependent approach for volume estimation. The difference in volume estimates between KUGI and RGI is most probably due to the difference in the number of glaciers in the inventories.
Comment #41: 380 and following: are you talking about mean or media elevations? Or the total elevation span of the whole glacier?
Response: These are the mean elevation values.
Comment #42: Table 3,4,5,6,7: Units are missing. What means “A” and “N” and “DC”? not clear
Response: The “A” represents glacier area in km2, “DC” is glacier debris cover again in km2, whereas, “N” indicates the glacier number (count). The units as well as the description of each letter(s) has been specified in the revised manuscript in all the tables.
Comment #43: l396. delete sentence. Already mentioned in the methods
Response: The sentence has been removed from the revised manuscript as suggested.
Comment #44: *Section 5.1: The whole Section can be strongly condensed. All information can be found in the tables and does not need to be repeated in the text.
Response: Thanks for the suggestion. Accordingly, we have trimmed this section in the revised manuscript.
Comment #45: *Table 6: how did you compute the average glacier aspects? Please provide the formula somewhere.
Response: As mentioned above, the aspect has been now recalculated in the revised manuscript. Kindly see the response to Comment #3. The aspect sines and cosines of each of the glacier’s DEM grid cells were summed and the mean aspect was calculated as the arctangent of the quotient of the two sums.
Comment #46: l483: how did you estimate the variations? Not explained!!!
Response: By the variation, we mean the difference not the statistical variance. The sentence has been rephrased accordingly for clarity in the revised manuscript.
Comment #47: *Section 5.2: Same as for Section 5.1.! It can be strongly condensed and most of the information can be summarized in nice tables and/or graphs. The text is very long and the information is hard to find. Tables and graphs would be beneficial for the reader
Response: Thanks, as suggested the information has been already put in the form of tables and were referred at appropriate places in the revised manuscript. Further, the section has been modified/condensed in the revised manuscript as suggested.
Comment #48: *Table 10: How did you define the elevation category? All pixels with in the interval? Or all glaciers with mean/median elevation within this interval? Unclear! How did you compute the uncertainties?
Response: In each elevation category, glaciers are grouped together in the elevation bins based on the mean elevation of individual glaciers. This has been specified in table caption in the revised manuscript. Furthermore, to assign the uncertainties for a sample average (for example glaciers in the elevation range between 5000-5500), we calculated the uncertainty of the sample-wide elevation change ( ) using the following (Huber et al.) , where σ_Δh is the uncertainty of each item i (glacier in our case) and n is the number of item in the sample.
Comment #49: Fig. 2,3,4: Please add a background. The glacier outlines would be also nice. The bar plot is too small an impossible to read. Does it show the mean elevation changes per glacier? Explain!
Response: Thanks for the suggestion. The figures have been modified accordingly. Yes, you are right, the bar graph is based on the mean elevation change per glacier.
Comment #50: l629: Are you talking about average elevation changes per glacier? Please clarify.
Response: Yes, we are talking about average elevation changes per glacier. This has been specified in the figure captions in the revised manuscript.
Comment #51: Table 11: is the slope take pixel by pixel or is it base on the mean slope per glacier?
Response: The slope is based on the mean slope per glacier. Specified in the table caption in the revised manuscript.
Comment #52: Table 12: same as for Table 11. is the aspect take from each pixel or the mean of each glaciers
Response: The aspect is based on the mean aspect of each glacier. Specified in the table caption in the revised manuscript.
Comment #53: l659: Maybe the bigger glaciers are located at lower altitudes? Please check
Response: Yes, the glaciers tend to be smaller at higher altitudes in the study area. For example, glaciers in the Suru and Chenab basins situated above 5500 m asl are generally smaller compared to the glaciers situated between 5000-5500 m asl (Table 4 in the revised manuscript).
Comment #54: Table 13: Units are missing for area
Response: Units (km2) of area have been provided in the revised manuscript.
Comment #55: l685: By inspecting Fig. 2-4, it looks like most glaciers are not south facing. Please check your aspect computation! Do not use a simple mean of all pixel wise aspect values of a glacier. See RGI6.0 technical report.#
Response: As suggested, the aspect was recalculated in the revised manuscript as specified in the RGI technical document and accordingly, the correct aspect is depicted in the revised manuscript.
Comment #56: Fig.5: Date and source of background image?
Response: The background image is FCC (7,4,2) of Landsat ETM+ dated 04-09-2000. The information has been added to the figure caption.
Comment #57: Fig.6: Date and source of background image? There are at least 2 glacier tongues. Most likely they were connected in the past, but the mapped state shows 2 individual major glacier tongues. Therefore, it is not wrong to split the glacier are in 2 polygons. Please rephrase accordingly, also in the main text.
Response: The date and source of the image has been provided in the revised manuscript. We, agree that there are two tongues, however, when we drape the 2000 satellite images over DEM, the ridge-topography is not prominent enough to manifest the flow direction/ridge divide on the satellite images especially in the accumulation zone, as such we did not divide the glacier in multiple polygons.
Comment #58: Fig. 7: the outlines are hard to see. Use different colors or wider lines. Date and source of background image?
Response: Thanks for the suggestion. Accordingly, the figure has been modified as suggested for better presentation of the outlines in the revised manuscript.
Comment #59: Table 16: Can be merged with Table 2 to avoid doubling of data.
Response: We have merged Table 16 with Table 8 to represent the difference in glacier number and area in the revised manuscript. The information in the Table 16 seems more relevant to Table 8, therefore the repeating information (area) has been removed from the merged tables in the revised manuscript.
Comment #60: l794: Unclear sentence
Response: The sentence attributes the higher thinning of south oriented glacier to the higher solar insolation received by southerly slopes. The sentence has been rephrased in the revised manuscript for clarity.
Citation: https://doi.org/10.5194/essd-2021-28-AC1
-
AC1: 'Reply on RC1', Romshoo Shakil Ahmad, 08 Jun 2021
-
RC2: 'Comment on essd-2021-28', Anonymous Referee #2, 16 Apr 2021
This is a useful contribution and in general seems a well-conducted assessment. Avoiding measureemnt bias is paramount for geodetic mass balance in particular, and some more evidence is needed to demonstrate that this has been achieved.
Sections 2.1, 2.2 and 2.3: please check the Jhelum minimum temperature - should be negative? Please provide consistent descriptive stats, e.g. annual precip in each case rather than mixing monthly and annual.
Line 279 - area calculation in 'ArcGIS environment' - what projection was used? Please ensure that an area-preserving projection is used when defining the glacier areas and, by extension, their volume changes.
Equation 6: good error quantification is vital in geodetic mass balance calculations. Please explain the sigma_z DEM uncertainty term. Is this a generic estimate of DEM quality, or is it specifically applicable to snow-covered surfaces and steep slopes? These are notoriously difficult to map topographically, particularly in opticial images. The voidfill error term is not defined. Please explain and justify using the penetration error as a random, uncorrelated error in this case. If penetration is wrong, it will be a systematic error and so should be added and not combined in quadrature.
Line 342: error assessment in off-glacier areas is good to do, but this is not reported or shown in the dh/dt figures. If it reveals systemtaic biases then these should be corrected to zero. This requires assessment at as many off-glacier locations as possible, at a range of altitudes, as the DEM biases are often not uniform across a scene. The iniital and corrected off-glacier stats should be reported and shown in the figures.
Table 2: KUGI glacier volumes - the calculation of volume is not trivial and is not explained. Where do these come from?
Figures 2 and 3: please show a background map, scale bar and the apparent dh/dt values off-glacier.
Citation: https://doi.org/10.5194/essd-2021-28-RC2 -
AC2: 'Reply on RC2', Romshoo Shakil Ahmad, 08 Jun 2021
Reviewer #2
General Comment: This is a useful contribution and in general seems a well-conducted assessment. Avoiding measurement bias is paramount for geodetic mass balance in particular, and some more evidence is needed to demonstrate that this has been achieved.
Response: We express our gratitude to the reviewer for the very useful review of the manuscript. The valuable comments and suggestions provided by the reviewer have improved the contents of the manuscript. We have responded point-by-point to all the comments and suggestions of the reviewer. The concerns of the reviewer about error estimates have been addressed entirely to the best of our ability and knowledge. The revised manuscript looks significantly improved. The point-by-point response to the detailed comments and suggestions raised by the reviewer are provided as follows:
Comment#1: Sections 2.1, 2.2 and 2.3: please check the Jhelum minimum temperature - should be negative? Please provide consistent descriptive stats, e.g. annual precip in each case rather than mixing monthly and annual.
Response: Thanks for the suggestion, we rechecked the mean minimum temperature and found it above zero, this has been previously reported. These sections have been however revised providing temperature and precipitation on same time scales.
Comment#2: Line 279 - area calculation in 'ArcGIS environment' - what projection was used? Please ensure that an area-preserving projection is used when defining the glacier areas and, by extension, their volume changes.
Response: In the present study we used the Mercator projection system considered suitable for use in areas between 84°N to 80°S. Also since all the inventories under consideration use the same projection system, therefore, it is assumed that any distortion in area would be same for all the inventories and would not affect the comparative analyses of the inventories evaluated in this study.
Comment#3: Equation 6: good error quantification is vital in geodetic mass balance calculations. Please explain the sigma_z DEM uncertainty term. Is this a generic estimate of DEM quality, or is it specifically applicable to snow-covered surfaces and steep slopes? These are notoriously difficult to map topographically, particularly in opticial images. The voidfill error term is not defined. Please explain and justify using the penetration error as a random, uncorrelated error in this case. If penetration is wrong, it will be a systematic error and so should be added and not combined in quadrature.
Response: Thanks for the comment and suggestions. The , , and are uncertainty of DEM differencing, uncertainty due to void filling (Since the DEMs especially SRTM has voids over the study area as such the DEMs coverage for each having voids >30% were excluded from the analysis whereas, the glaciers with <30% voids were filled with nature neighbor interpolation algorithm), temporal uncertainty of TanDEM-X and uncertainty of radar signal penetration respectively.
was calculated using the widely accepted approach,considering the spatial autocorrelation as:
where σΔh is the off-glacier NMAD, A is the glacier area analyzed and Acor = πd2, with d being the decorrelation length, we used a d=950 m observed for Jammu and Kashmir Himalaya, ecompassing the present study area (Abdullah et al. 2020). For the estimation voidfill uncertainty we again followed (Abdullah et al. 2020) where a set of glaciers (void free) distributed across the study region were selected and >30% voids were artificially created. The artificially created voids were then filled using different interpolation algorithms and the results where compared with the original values (glacier without voids). The voids filled with using natural neighbor algorithm were found in good agreement with the original values with just ±0.07 difference between the original and interpolated values. The difference of ±0.07 introduced due to the void filling was therefore considered as uncertainty due to voidfill (Abdullah et al. 2020). The penetration bias was calculated using the following exponential function determined for the neighboring Lahaul-Sipti region by Vijay and Braun, 2016:
where ‘x’ is absolute surface elevation and ‘y’ is the relative penetration bias between SRTM X and C band. The uncertainty of each of the individual parameter is described in detail in the revised manuscript.
Comment#4: Line 342: error assessment in off-glacier areas is good to do, but this is not reported or shown in the dh/dt figures. If it reveals systematic biases then these should be corrected to zero. This requires assessment at as many off-glacier locations as possible, at a range of altitudes, as the DEM biases are often not uniform across a scene. The initial and corrected off-glacier stats should be reported and shown in the figures.
Response: The elevation change estimates are based on Abdullah et al. 2020 where the error estimation is described in detail including the off glacier biases in various 5° slope bins. The study reported mean off-glacier elevation difference of 0.06 m a-1.
Comment#5: Table 2: KUGI glacier volumes - the calculation of volume is not trivial and is not explained. Where do these come from?
Response: The glacier volume was estimated using the slope-dependent volume estimation approach (Haeberli and Hoelzle, 1995). The methodology to estimate volume has been incorporated in the revised manuscript.
Comment#6: Figures 2 and 3: please show a background map, scale bar and the apparent dh/dt values off-glacier.
Response: As suggested, the figures have been modified in the revised manuscript.
Citation: https://doi.org/10.5194/essd-2021-28-AC2
-
AC2: 'Reply on RC2', Romshoo Shakil Ahmad, 08 Jun 2021
-
CC1: 'Comment on essd-2021-28', Glacier Ice, 01 May 2021
The paper by Romshoo et al. is interesting research on evaluating the existing glacier inventories over the north-western Himalayan region of Jammu and Kashmir. However, there are certain loopholes, some of them very gross, which have been pointed out below:
The title mentions "global glacier inventories", however, the very first sentence of the Abstract section mentions ICIMOD (for Himalaya only) and GAMDAM (for Asia only) inventories which are regional. As such I would suggest the authors modify the text wherever required. Or maybe mention "Evaluation of the existing glacier inventories..."
What do the authors mean by limited field surveys? How many glaciers were actually field surveyed? Specifically, what type of data was collected from the field, and would it qualify as a representative sample for Quality Control?
The uncertainty of mapping is more in Jhelum (~13%) as compared to Suru and Chenab (~3.3%). Since there are a lot of debris-covered glaciers in Chenab and Suru, uncertainty should be more in these basins as compared to Jhelum (where once predominantly finds clean-ice glaciers with few exceptions). Please explain.
When the authors say "most of the glaciers in the study area are <1 km2 in size><1km2", they should mention the number and percentage of these glaciers. Also for the 1-5 km2 category.
"Majority of the glaciers....". Please quantify. Similarly "... Jhelum where the glaciers are mostly...". Again MOSTLY is subjective.
Rather than saying "mainly harbor slopes ranging from 10-30°", it would be better to mention the average slopes of the glaciers in all three basins.
What is RBA? Mention at first occurrence. If KUGI is "consistent" with RGI and GAMDAM, I wonder then as to what is the purpose of KUGI?
I would suggest removing the portion about geodetic mass changes since the authors have already published geodetic mass changes in Scientific Reports very recently (See the reference below). This would mean plagiarism/redundancy.
Abdullah, T., Romshoo, S. A., & Rashid, I. (2020). The satellite observed glacier mass changes over the Upper Indus Basin during 2000–2012. Scientific Reports, 10(1), 1-9.
"Evaluation of the glacier inventories and assessment of glacier elevation change in the data-scarce Himalaya, reported in this article, would constitute a reliable database for research particularly in hydrology, glaciology, and climate change". This is not convincing. How will this effort help, given the fact that authors mention that KUGI is "consistent" with RGI and GAMDAM? How is the KUGI more reliable than GAMDAM and RGI?
Line 70: Azam et al., 212 should be Azam et al. 2012
Line 73: "Indicated" should be "indicates".
I do not find the introduction section very convincing. Especially Line 60-105 appears more of discontinuous sentences where authors jump between various glaciological assessments (area changes, frontal retreat, geodetic and glaciological mass changes). This becomes irrelevant since the focus of the paper is the "evaluation" of glacier inventories over 3 river basins. Besides, I find certain sentences over-referenced and others poorly referenced. In many places, the authors have not even bothered to cite the recent literature (See details after the comment. Although I do not know whether they will be relevant if the MS is revised and contextualized for comparison of glacier inventories). For example, the authors say that using the freely available glacier datasets for glacier change assessment and future projections is not recommended as the glacier inventories have inconsistencies in terms of different glacier variables. Does it mean the regional glacier-related assessments (cited by authors in first paragraphs) are not imprecise and not reliable? The authors further go on to say that "the glacier inventory database by Shukla et al., (2020), restricted to the Suru basin, is primarily based on the automatic approach (normalized-difference snow index) unlike the present study where the glaciers are mapped manually using on-screen digitization." Do they mean the inventory by Shukla et al is not credible? Since on-screen digitization is highly subjective and dependent on the cognition/skill of analysts, the approach could be contested especially when it comes to inventory mapping over large areas. How can/have the uncertainties about cognition been addressed by authors?
Merely saying Google Earth was used for validation will not have many takers among the remote sensing glaciology community. I tried to dig into Google Earth data of the 2000s for the three basins but found massively snow/cloud-covered data for the assessment period. The authors need to come up clean on this and say precisely where Google Earth data was used for correcting the glacier outlines. And also since Google Earth and Landsat data do not have an exact overlap, how was coregistration achieved.
RECENT LITERATURE:
Nie, Y., Pritchard, H. D., Liu, Q., Hennig, T., Wang, W., Wang, X., ... & Chen, X. (2021). Glacial change and hydrological implications in the Himalaya and Karakoram. Nature Reviews Earth & Environment, 1-16.
Farinotti, D., Immerzeel, W. W., de Kok, R. J., Quincey, D. J., & Dehecq, A. (2020). Manifestations and mechanisms of the Karakoram glacier Anomaly. Nature geoscience, 13(1), 8-16.
Shean, D. E., Bhushan, S., Montesano, P., Rounce, D. R., Arendt, A., & Osmanoglu, B. (2020). A systematic, regional assessment of high mountain Asia glacier mass balance. Frontiers in Earth Science, 7, 363.
Soheb, M., Ramanathan, A., Angchuk, T., Mandal, A., Kumar, N., & Lotus, S. (2020). Mass-balance observation, reconstruction and sensitivity of Stok glacier, Ladakh region, India, between 1978 and 2019. Journal of Glaciology, 66(258), 627-642.
Mehta, M., Kumar, V., Garg, S., & Shukla, A. (2021). Little Ice Age glacier extent and temporal changes in annual mass balance (2016–2019) of Pensilungpa Glacier, Zanskar Himalaya. Regional Environmental Change, 21(2), 1-18.
Farinotti, D., Immerzeel, W. W., de Kok, R. J., Quincey, D. J., & Dehecq, A. (2020). Manifestations and mechanisms of the Karakoram glacier Anomaly. Nature geoscience, 13(1), 8-16.
Nie, Y., Pritchard, H. D., Liu, Q., Hennig, T., Wang, W., Wang, X., ... & Chen, X. (2021). Glacial change and hydrological implications in the Himalaya and Karakoram. Nature Reviews Earth & Environment, 1-16.
Pritchard, H. D. (2019). Asia’s shrinking glaciers protect large populations from drought stress. Nature, 569(7758), 649-654.
Immerzeel, W. W., Lutz, A. F., Andrade, M., Bahl, A., Biemans, H., Bolch, T., ... & Baillie, J. E. M. (2020). Importance and vulnerability of the world’s water towers. Nature, 577(7790), 364-369.
Hugonnet, R., McNabb, R., Berthier, E. et al. Accelerated global glacier mass loss in the early twenty-first century. Nature 592, 726–731 (2021). https://doi.org/10.1038/s41586-021-03436-z
The authors suggest having used satellite data of 2000±3 years to delineate inventories whereas ICIMOD glacier inventory has used satellite data of 2005±5 years. Wouldn't it be comparing apples with oranges? This becomes important especially in the case of small glaciers and needs to be factored for.
"It is hoped that the KU glacier inventory and elevation change databases presented in this paper shall further help in promoting research in fields like climate change, hydrology, and other allied fields." This is common for any inventory. The rationale should be how KUGI will help to further it. This needs to be mentioned.
Line 132: Need to place "and" between latitude and longitude values.
Line 135-36: "The area above 3600 m asl in general remains covered with perennial snow and glaciers". Not true mostly. This has to be ~4000 m asl for the J&K region.
Line 136-37: Why have authors quoted numbers from RGI inventory and not GAMDAM?
Line 138: "thus making the study area the most glaciated terrain". Reframe.
Line 139-40: The authors quote Kamp et al. (2011) and suggest that glaciers are cirque-type in Ladakh which is not true with all the glaciers in the Suru basin and also neighboring Zanskar region. Glaciers in North Ladakh (Siachin area, Rimo group) arent cirque-type either. Please reframe.
Line 140-41: "All the major tributaries" instead of "Most of the major tributaries"
Fig. 1: The caption should mention the following: What does the inset map represent? What is the background image (a DEM or what)? Have a legend for elevation if it is so. Mention may be "GA" for glacier area instead of "A" since "A" also represents Jhelum Basin. The text could be made bold and a little larger for histograms in the study area map. What is the source of number of glaciers, glacier area, debris cover, and glacier volume? Maybe plot glacier volume and debris cover on the secondary axis since the associated values are small.
Line 150-180: Could be better represented as a table. For climatology of Jhelum Basin use: Zaz et al 2019 (ACP). For Suru met data are improper. Authors used Schmidt and Nüsser, 2012 (which mentions a different area of Ladakh, Kang Yatze massif, and not Suru) while Chevuturi et al. (2018) report climate of Leh and not Suru. Need to correct it. Similarly for Chenab, the authors quote Azam et al. (for Chotta Shigri area). Why not cite Bhutiyani et al. 2007 (Climatic change) and Bhutiuyani et al 2010 (Int J of Climatology)?
Line 192: "The of moderate resolution". Please correct.
Line 193-96: Some of the references have been quoted above and do not necessarily need to be mentioned here.
Table 1:
The authors mentioned wrong dates for imagery used for ICIMOD inventory. It is 2005±3 years (Weblink: https://lib.icimod.org/record/9419. See page 7>Section Satellite images> the second paragraph). I would again repeat my above question: Can 2000 data (in RGI, GAMDAM, and KUGI) be compared with 2005 data (as used in ICIMOD inventory)? Definitely not. Please justify. Also modify the respective entry in the Table. Details of all satellite scenes should be provided as supplementary data.
Google Earth: Mention the date of Google Earth images, if at all they were used for correcting/validating glacier outlines. Maybe have a supplementary file for mentioning which glaciers were validated using Google Earth imagery.
Line 213-14: Delete " hereafter named Kashmir University Glacial Inventory" as it has already been defined as KUGI in the abstract.
Line 214: Delete "global".
Line 129: "acquired during 2002 to 2008". See my earlier comments.
Line 225-28: The authors mention "The RGI glacier outlines have been extracted semi-automatically from the Landsat satellite images between 1998 and 2009. However, most of the glaciers (~98%) in the inventory over the study area have been extracted from the images acquired during 1998-2002". How many glaciers were delineated from 1998-2002 data in RGI inventory? See attribute of RGI shapefile. Again comparison seems a problem here; not only due to dates but also the technique used.
Line 244: What do authors mean by "limited field surveys"? How many glaciers (%) were ground surveyed? What kind of data was collected? Need to reflect all that in the MS.
Line 256-57, 262, 267-270: "were verified from the Google EarthTM". The GE data for 2000 is almost not usable for the region. Please explain. Is it a deliberate attempt of misinformation or what?
Line 262-63: "The thin debris layer on the glacier surface, often bearing lower surface temperature". Do the authors mean differential wrt ice or neighboring landscape?
Line 200, 280-282: The authors mention ASTER DEM here having been used to derive glacier-specific topographic parameters. But there is no mention of ASTER DEM in Table 1. Why was ASTER DEM used when CARTO DEM with a similar resolution is available over the region? Seems repetition of (line 201-202) here (280-82).
Line 285-295: When the techniques used for mapping the glacier outlines are different, it is but obvious that there won't be a high overlap. Would it be so? Add a justification.
Remove section 4.3 as explained in the beginning and also uncertainty related to geodetic mass balance.
Line 330: Rp/Ap is a constant. What does it represent?
Line 368: "Jhelum" instead of "Jehlum". Be consistent with spelling.
Line 368: "The glaciers range in size from 0.01 km2". This means ~11 pixels. I wonder if such small-sized glaciers (1 ha as mentioned by authors) could be mapped from 30 m Landsat data? This would be highly uncertain. Could it be that some of them were snowpacks and not glaciers especially when ascertaining from 2000-02 data?
Table 2: The authors mention glacier volume but have not provided any information as to how ice volumes were derived? Did the authors use VA scaling and why if it is known that VA scaling estimates are highly uncertain, even for entire mountain ranges.
Line 383-84: "mean glacier slope in the basin varies between 9° and 50°". The glaciers with a mean slope of 50 degrees are highly unlikely since by the definition such areas (> 30-degree slope) could be avalanche feeding zones. Please speak otherwise it raises concerns about the inventory itself.
Table 3: Why do the authors need to mention glacier area categories from >20->50 in the Jhelum basin. >50 should not be the category. Let it be >50-the highest glacier area in the respective basin. Similarly, if there is a category 1-2, it should be followed with >2-5, >5-10, so on and so forth.
Line 395-397: Delete as it is already mentioned in 275-76.
Table 4: Elevation categories from 5500-7000 are not relevant for Jhelun and as such could be deleted. If the first category starts from <=4000, the next category should be >4000-4500, so on and so forth.
Mention uncertainties about each A (glacier areas) in table 3, 4, 5 and 6.
The uncertainty in TGA for the Jhelum basin is ~13.3% compared to Suru and Chenab (both 3.3%). Why is this so? This should have been the other way around since there are more clean-ice glaciers in Jhelum. Explain.
How different are the estimates of Table 7 different from Scherler et al 2018 (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018GL080158)
Line 474: Delete "regional and global".
Table 8: Should be "ΔN" instead of "N". Percentage difference could be mentioned in brackets.
Line 490-500: Use ICIMOD and GAMDAM rather than ICIMODGI and GAMDAMGI.
Table 9: Above in methods authors mention having used median glacier elevation rather than mean glacier elevation.
Section 5.3 and associated tables/figures should be removed.
Line 675-680: The authors mention large glaciers at high altitudes and then low glacier cover at high altitudes which is a kind of contradiction. Large glaciers as authors suggest in results contribute to glacier cover. Please explain.
Table 14: Remove
The authors do not discuss much about the aspect (neither in results nor discussion) except very qualitatively.
Section 6.2 should be "Inconsistencies in existing glacier inventories"
Figure 5: Mention background image and band combination.
Line 720-724: The overlap ratio could be misleading since the inventories were computed using different methods. Please justify.
Figure 6: SG. On checking from Google Earth, the existing inventories have rightly followed GLIMS definition of glacier delineation and divided the ice into two polygons since the ice fluxes move in different ablation zones. However, the authors have erred here (and maybe in many such cases) by considering it as a single glacier. The authors should know that although the ice masses are connected (in the accumulation zone) the movement of ice in different directions owing to the ridge-topography divides the ice into two polygons and hence two glaciers (refer to GLIMS definition of glaciers). This appears to be a huge flaw with the interpretation by authors as the number of glaciers is massively underestimated in KUGI. This needs to be corrected in the data as well as explanations to the data.
Figure 7: Mention background image and band combination.
Remove section 6.3
Conclusions:
While the authors mention "limited field survey" at many places in the text, they have failed to showcase the data collected, the photographs depicting the glacier environments in these three catchments. They should show field data and photographs from all three catchments and demonstrate its usefulness in inventorying the glaciers in these three respective areas.
Delete sentences about geodetic mass changes.
The authors fail to convince the robustness of KUGI compared to at least RGI and GAMDAM. This needs to come up very in the results, discussion as well as the conclusion.
Citation: https://doi.org/10.5194/essd-2021-28-CC1 -
AC3: 'Reply on CC1', Romshoo Shakil Ahmad, 08 Jun 2021
CC#1
General Comment: The paper by Romshoo et al. is interesting research on evaluating the existing glacier inventories over the north-western Himalayan region of Jammu and Kashmir. However, there are certain loopholes, some of them very gross, which have been pointed out below:
Response: The authors thank the anonymous reviewer/commentator “Glacier Ice” for the useful comments and suggestion on the manuscript. The comments/suggestions have been responded point-by-point in the following sections and the useful suggestions have been incorporated in the revised manuscript which have improved the quality of the revised manuscript.
Comment #1: The title mentions "global glacier inventories", however, the very first sentence of the Abstract section mentions ICIMOD (for Himalaya only) and GAMDAM (for Asia only) inventories which are regional. As such I would suggest the authors modify the text wherever required. Or maybe mention "Evaluation of the existing glacier inventories..."
Response: Agreed, that the ICIMOD and GAMDAM are regional inventories and the RGI is a global glacier inventory. Accordingly, the title and text have been modified in the revised manuscript. The “Global glacier inventories” has been replaced with “Global and Regional glacier inventories” in the title of the revised manuscript.
Comment #2: What do the authors mean by limited field surveys? How many glaciers were actually field surveyed? Specifically, what type of data was collected from the field, and would it qualify as a representative sample for Quality Control?
Response: In the present study, we have done field surveys/validation on 20 glaciers located across the study area. The locations and the field photographs of these glaciers has been provided as Supplementary figure in the revised manuscript. We have collected the snout positional data of the debris-covered tongues of these glaciers to validate the glacier delineation. The field measurements of these glaciers acted as an interpretation tool for delineation of the debris-covered glaciers in the study area. The terminus of the heavily debris covered glaciers like the Hoksar glacier in Jhelum basin was not easily delineated even using the thermal and google earth imagery. We therefore, mapped the glacier terminus on field and further observed that the debris on the glacier is relatively smooth and aligned in the direction of glacier flow when compared to the debris-cover in the surroundings which was found a useful field-based information for mapping of debris-covered termini of other glaciers when viewed on Google earth. Further, eight of these twelve glaciers have been designated as benchmark glaciers and are continuously studied for mass balance, GPR, debris thickness, Surface mapping temperature profiling and other glaciological studies since the last 5-8 years. In addition, glacier outlines of several other glaciers in the vicinity of these 8 benchmark glaciers have been verified during annual glacier field expeditions during the last 5-8 years.
Additionally, all the heavily debris covered glaciers and a majority of the clean glaciers, numbering more than 850 were qualitatively verified on Google Earth image for the rectification of any delineation error. A similar approach of quality check using Google Earth has been previously adopted by Nagai et al. (2016) and several others and is an accepted method for validation of such a large number of glaciers located in inaccessible complex terrain.
Field photographs of the about 20 selected glaciers in the three basins, which have been visited over the last few years for field measurements/validation along with their GLIMS ID are presented in the revised manuscript (The field photographs of some of these glaciers are provided above). A locational map of these glaciers in the form of a KML file would be provided as a supplemental material in the revised manuscript (The field photographs have also been attached as a supplement file in pdf format).
Comment #3: The uncertainty of mapping is more in Jhelum (~13%) as compared to Suru and Chenab (~3.3%). Since there are a lot of debris-covered glaciers in Chenab and Suru, uncertainty should be more in these basins as compared to Jhelum (where once predominantly finds clean-ice glaciers with few exceptions). Please explain.
Response: For the uncertainty assessment of the glacier area, we have referred to the previous detailed uncertainty analysis by Paul et al. (2013) that reported an error of ~3% for the alpine glaciers. Paul et al. (2013), however, reported a perimeter-area ratio of 5.03 km−1, unlike the perimeter-area ratio of 0.96 km-1 observed for Chenab in this study, 3.9 km-1 for Suru and 6.22 km-1 for the Jhelum basin. We therefore applied a scaling after Braun et al, (2019) to determine the uncertainty in glacier outline delineation in the three basins in this study. As indicated by the value of Perimeter-Area ratio of 6.22 km-1 for the Jhelum basin, it is quite obvious that the basin is expected to have higher uncertainty as the perimeter-area ratio of the basin is higher amongst the three studied basins. This has been specified in the revised manuscript.
Comment #4: When the authors say "most of the glaciers in the study area are <1 km2 in size><1km2", they should mention the number and percentage of these glaciers. Also for the 1-5 km2 category.
Response: The detailed statistics of the glacier number and area in each size category is described in detail in the Results section and already provided in Table 3 of the manuscript. On an average around 91% of the glaciers (glacier number) are in the size category < 1 km2 in the Jhelum basin. Similarly, around 77% of the glaciers (number) are falling in the < 1 km2 area class in the Suru and Chenab basins (Table 3). In terms of the glacier coverage (area), the area class 1-5 km2 harbor ~44% of the glacier area in the Chenab basin, followed by ~55% in the Suru basin and ~90% in the Jhelum basin (Table 3). The statistics are averaged for all the glacier inventories and the numbers vary between the individual inventories (Table 3). This has been specified in the revised manuscript.
Comment #5: "Majority of the glaciers....". Please quantify. Similarly "... Jhelum where the glaciers are mostly...". Again MOSTLY is subjective.
Response: The elevation range 4500-5500 for the Chenab and Suru basins harbor around 85.5% and 91.3% of the glacier number and area respectively. Whereas, the elevation band 4000-5000 harbors 91.8% (number) and 95.8% (area) of glacier number and area respectively in the Jhelum basin (Information already provided in Table 4). However, the information has been specified in the revised manuscript as suggested.
Comment #6: Rather than saying "mainly harbor slopes ranging from 10-30°", it would be better to mention the average slopes of the glaciers in all three basins.
Response: The slope range (10-30°) harbors 78% of the glaciers in terms of number and 84% of the glacier area (Table 4). The numbers have been specified in the revised manuscript.
Comment #7: What is RBA? Mention at first occurrence. If KUGI is "consistent" with RGI and GAMDAM, I wonder then as to what is the purpose of KUGI?
Response: The overlap ratio of individual glaciers is represented by “rov” whereas, the “R” (RBA) has been used to represent the average overlapping ratio of the base and target glacier inventories (overlap ratio averaged for all the glaciers in a particular inventory combination e.g, KUGI-RGI for each basin). This has been described under the methodology section in the revised manuscript. The inventories are consistent in general pattern and distribution which means that a majority of the glaciers in all the inventories are found in a particular area class, slope category etc. however, large differences occur in individual glaciers and overall statistics in terms of the area/size (Table 3-6). However, the specific purpose of the KUGI is provided is provided in the response to the Comment #9 below.
Comment #8: I would suggest removing the portion about geodetic mass changes since the authors have already published geodetic mass changes in Scientific Reports very recently (See the reference below). This would mean plagiarism/redundancy.
Abdullah, T., Romshoo, S. A., & Rashid, I. (2020). The satellite observed glacier mass changes over the Upper Indus Basin during 2000–2012. Scientific Reports, 10(1), 1-9.
Response: The geodetic mass changes of the entire region of Upper Indus Basin comprising of 12000 glaciers, of which Jhelum, Chenab and Suru are a part, has been published in the Scientific Reports which has been referred to in the manuscript and therefore would not in any way amount to plagiarism. However, we wanted to retain the elevation change information in the inventory data base for the ready availability of this valuable database to scientific community as the same provides a valuable additional information about the behavior/dynamics of the glaciers in the study area.
Comment #9: "Evaluation of the glacier inventories and assessment of glacier elevation change in the data-scarce Himalaya, reported in this article, would constitute a reliable database for research particularly in hydrology, glaciology, and climate change". This is not convincing. How will this effort help, given the fact that authors mention that KUGI is "consistent" with RGI and GAMDAM? How is the KUGI more reliable than GAMDAM and RGI?
Response: Please see the response to the comment above where it has been clarified that there is “consistency” in terms of the general pattern/distribution, however there are significant differences in various glacier parameters reported in the three inventories compared to the KUGI. The value addition and novelty of the KUGI over the three inventories evaluated in this study has been prominently mentioned at relevant places in the revised manuscript and the same is consolidated and reproduced here for your perusal is also reproduced as follows:
“Primarily, the motivation for the KUGI is to develop a high-resolution glacier inventory with improved accuracy with visual interpretation and manual delineation of glaciers from Landsat satellite data supported by the limited ground truth and supplementing the glacier outlines with additional data like debris-cover, thickness changes and other glaciological parameters, that are either missing or incorrect in the existing databases so that the database is made available to the large research community for various applications.
Purportedly the global and regional glacier databases that were chosen for comparison in this study have been generated using a semi-automated method (manual for GAMDAMGI) allowing less human error, quick delivery, and high accuracy. However, it was found in this study that there are significant errors in the evaluated databases due to the misinterpretation of seasonal snow cover particularly on the glacier headwalls at high altitudes, shadow-covered glaciers and debris-cover. Against the reports/claims of the overall accuracy of the global/regional glacier databases, ~3% for ICIMOD (Bajracharya er al, 2014), ~5% for RGI (Pfeffer et al. 2014; RGI Consortium, 2017) and 15% for GAMDAM (Nuimura et al. 2015), it was found that, compared to the KUGI, the ICIMOD is overestimating glacier area by 12.2%, RGI underestimates the glacier area by 2.4% and the GAMDAM inventory by 1.5%. However, the three databases overestimate the glacier numbers in the three basins considerably; RGI by ~45%, ICMOD by ~68% and GAMDAM by ~56%. Gardelle et al. (2013) found that in the southeastern Tibet, RGI 2.0 database has glacier extent 88% greater than their estimate. Similarly, Nuimura et al. (2015) while comparing GAMDAM with ICIMOD found a significant discrepancy between the two inventories. Frey et al. (2014) and Mölg et al. (2018), have highlighted the presence of debris-cover, seasonal snow and cloud cover as the main source of uncertainty in the Himalayan region. Mohammad et al. (2019) has also highlighted the differences between the existing glacier inventories in Indus basin.
Keeping in view the worldwide use and applications of global and regional databases, it is important that a rigorous evaluation of these global and regional inventories is undertaken for the continued refinement of the methodology which is a fundamental requirement for any meaningful application of the global or regional database. It is hoped that the future releases of the databases will improve these and other shortcomings identified in this manuscript. Although this cross-checking improved the quality of the data, the mapped glacier outlines are also affected by various other types of obscurities, which are mostly dependent on image resolution which is the also the case with KUGI.
KUGI improved the mis-mapped glacier outlines/boundaries from existing global and regional inventories and any mismatches of the glacier geometry due to the seasonal/temporal snow cover and shadows were manually corrected using additional Landsat images and Google earth images. Further the mapped glaciers with better georeferencing were overlaid with high resolution images in Google Earth environment for validation wherever available. Though, the mis-mapped/mis-located outlines, observed on the global/regional inventories, may have only limited effect on measurements of glacier area, but can introduce serious errors into applications that rely on absolute positioning (e.g. co-registration to other datasets such as DEMs). The only realistic way to correct them is to provide more accurate outlines as done in the KUGI which would serve as source of improved outlines for the scientific community interested in conducting various application studies using the glacier outlines.
The analysis of the debris-cover (>19%)-the criterion we used to classify the debris-covered glaciers (Brun et al., 2019) showed that the RGI, ICIMOD and GAMDAM glacier inventories have underestimated the debris-covered glaciers by ~15%, ~25%, 8% respectively. Debris cover, present on 44% of Earth’s glaciers, significantly influences glacier melt. Despite its significant importance, the debris cover has not been mapped with accuracy in the three global/regional glacier databases evaluated in this study. Due to lack of debris-cover map at the global level (Sam Herreid & Francesca Pellicciotti, 2020), debris cover has been omitted from global glacier models and forecasts of their response to a changing climate. Therefore, the KUGI, has fundamentally resolved this omission and provided improved debris-cover outlines of the three basins in the Northwest Himalaya. KUGI has added a separate debris-cover database in the three basins which is missing in the global/regional databases. This is a major improvement and correction to the existing global/regional databases evaluated in this study. Use of the KUGI outlines of the debris-covered parts of glaciers in glacier-melt models will enable improved estimates of melt over the three basins.
Other than the debris-covered glaciers, the discrepancy related to the shadowed glaciers is another major error with the glacier inventories. Though, KUGI has not generated a database of these glaciers, but the same is under preparation for inclusion in the revised version of the KUGI database being submitted after this revision”
Comment #10: Line 70: Azam et al., 212 should be Azam et al. 2012
Response: The typo has been corrected in the revised manuscript.
Comment #11: Line 73: "Indicated" should be "indicates".
Response: The word “indicated” has been replaced by “indicates” in the revised manuscript.
Comment #12: I do not find the introduction section very convincing. Especially Line 60-105 appears more of discontinuous sentences where authors jump between various glaciological assessments (area changes, frontal retreat, geodetic and glaciological mass changes). This becomes irrelevant since the focus of the paper is the "evaluation" of glacier inventories over 3 river basins. Besides, I find certain sentences over-referenced and others poorly referenced. In many places, the authors have not even bothered to cite the recent literature (See details after the comment. Although I do not know whether they will be relevant if the MS is revised and contextualized for comparison of glacier inventories). For example, the authors say that using the freely available glacier datasets for glacier change assessment and future projections is not recommended as the glacier inventories have inconsistencies in terms of different glacier variables. Does it mean the regional glacier-related assessments (cited by authors in first paragraphs) are not imprecise and not reliable? The authors further go on to say that "the glacier inventory database by Shukla et al., (2020), restricted to the Suru basin, is primarily based on the automatic approach (normalized-difference snow index) unlike the present study where the glaciers are mapped manually using on-screen digitization." Do they mean the inventory by Shukla et al is not credible? Since on-screen digitization is highly subjective and dependent on the cognition/skill of analysts, the approach could be contested especially when it comes to inventory mapping over large areas. How can/have the uncertainties about cognition been addressed by authors?
Response: The introduction section provides a broader overview of the research under consideration and therefore, the authors felt that the reference to geodetic mass balance and other glacial studies is relevant. However, as suggested, a few relevant references suggested by the anonymous commentator have been incorporated in the revised manuscript.
We never said that the existing inventories are unreliable. Instead, we reiterate that a thorough evaluation of the glacier boundaries, as reported in this paper, is required before using them for impact assessment or any other climatic and hydrological application especially when the spatial domain of investigation is small (basins or sub-basins). The discrepancy in the glacier area might result in significant uncertainty in the use of the data for various applications.
The commentator is again insinuating when he emphasizes in italics something that is not meant by the authors-Not our words. We did not discredit the glacier inventory by Shukla et al, in fact, a part of the study area in Suru overlaps with the glaciers studied by Shukla et al. and therefore it is necessary to refer the paper. However, we observed a difference in glacier coverage reported between for a few glaciers in the two inventories and it is therefore important to explain this difference which we have attributed to the different techniques used in the present study.
Regarding minimizing the error due to the onscreen digitization by the analyst, it was made sure that analysts (authors) use the same criteria in terms of pre-processing, mapping scale etc. for the delineation of glacier boundaries. Furthermore, all the glacier boundaries were checked for quality control and corrected by the Lead author before finalization. This minimized the uncertainty due to the skill/interpretation of the analyst. The uncertainty approach used in the present study is also based on detailed analysis of glacier uncertainties mapped by multiple analysts (Paul et 2013) and therefore we believe that this approach is quite valid and addresses the uncertainties well.
Comment #13: Merely saying Google Earth was used for validation will not have many takers among the remote sensing glaciology community. I tried to dig into Google Earth data of the 2000s for the three basins but found massively snow/cloud-covered data for the assessment period. The authors need to come up clean on this and say precisely where Google Earth data was used for correcting the glacier outlines. And also since Google Earth and Landsat data do not have an exact overlap, how was coregistration achieved.
Response: This is very unfortunate that the anonymous commentator is again using very strong words like “need to come clean” in his review comments which sound like personal and unprofessional. We do agree that the Google Earth data for 2000s for the study area is cloud and snow covered. However, it is pertinent mention here that the Google Earth image of post-dating 2000s were used for quality check and verification of ambiguous glacier outlines. For example, glacier headwalls of many glaciers covered with thin layer of snow appeared smooth on the Landsat imagery, however, when checked on the high-resolution Google EarthTM imagery, it usually turned out an undulated non glaciated surface or a rock surface covered with thin snow. Besides all the heavily debris covered glaciers were qualitatively verified on Google Earth for rectification of any delineation error. A similar approach of quality check using Google Earth has been previously adopted by Nagai et al. (2016) and several others and is an accepted method for such a large number of glaciers located in inaccessible complex terrain. Further, we did not use the Google Earth data for the quantification of glacier area, and therefore an exact-overlap was not required. All the statistic quantifications and analyses in the present study is based on the Landsat images. This has been specified in the revised manuscript. The Google Earth images were used for the verification and validation of the glacier outlines only. The location of the 850 glaciers verified from the snow- and cloud-free Google Earth images would be provided as supplemental KML file in the revised manuscript.
RECENT LITERATURE:
Nie, Y., Pritchard, H. D., Liu, Q., Hennig, T., Wang, W., Wang, X., ... & Chen, X. (2021). Glacial change and hydrological implications in the Himalaya and Karakoram. Nature Reviews Earth & Environment, 1-16.
Farinotti, D., Immerzeel, W. W., de Kok, R. J., Quincey, D. J., & Dehecq, A. (2020). Manifestations and mechanisms of the Karakoram glacier Anomaly. Nature geoscience, 13(1), 8-16.
Shean, D. E., Bhushan, S., Montesano, P., Rounce, D. R., Arendt, A., & Osmanoglu, B. (2020). A systematic, regional assessment of high mountain Asia glacier mass balance. Frontiers in Earth Science, 7, 363.
Soheb, M., Ramanathan, A., Angchuk, T., Mandal, A., Kumar, N., & Lotus, S. (2020). Mass-balance observation, reconstruction and sensitivity of Stok glacier, Ladakh region, India, between 1978 and 2019. Journal of Glaciology, 66(258), 627-642.
Mehta, M., Kumar, V., Garg, S., & Shukla, A. (2021). Little Ice Age glacier extent and temporal changes in annual mass balance (2016–2019) of Pensilungpa Glacier, Zanskar Himalaya. Regional Environmental Change, 21(2), 1-18.
Farinotti, D., Immerzeel, W. W., de Kok, R. J., Quincey, D. J., & Dehecq, A. (2020). Manifestations and mechanisms of the Karakoram glacier Anomaly. Nature geoscience, 13(1), 8-16.
Nie, Y., Pritchard, H. D., Liu, Q., Hennig, T., Wang, W., Wang, X., ... & Chen, X. (2021). Glacial change and hydrological implications in the Himalaya and Karakoram. Nature Reviews Earth & Environment, 1-16.
Pritchard, H. D. (2019). Asia’s shrinking glaciers protect large populations from drought stress. Nature, 569(7758), 649-654.
Immerzeel, W. W., Lutz, A. F., Andrade, M., Bahl, A., Biemans, H., Bolch, T., ... & Baillie, J. E. M. (2020). Importance and vulnerability of the world’s water towers. Nature, 577(7790), 364-369.
Hugonnet, R., McNabb, R., Berthier, E. et al. Accelerated global glacier mass loss in the early twenty-first century. Nature 592, 726–731 (2021). https://doi.org/10.1038/s41586-021-03436-z
Response: The relevant literature has been cited at appropriate places in the revised manuscript.
Comment #14: The authors suggest having used satellite data of 2000±3 years to delineate inventories whereas ICIMOD glacier inventory has used satellite data of 2005±5 years. Wouldn't it be comparing apples with oranges? This becomes important especially in the case of small glaciers and needs to be factored for.
Response: The difference in the dates of source images does not explain the significant glacier area over-estimation observed in the Jhelum and Chenab basins. Using 2005±3 data (compared to 2000±2 in case of KUGI), normally an under estimation in glacier area is expected in ICIMOD but the comparison of the ICIMOD database with KUGI shows over-estimation of the glacier area which has been attributed to inclusion snow covered glacier headwalls and at places some season snowpacks also. Further, the area underestimation of 10.97% in case of the Suru basin is not fully explained by the expected area change between 2000±2 and 2005±3. Such comparisons using data gap of 2-3 years have also been carried out in previous studies (Nuimura et al. 2015) and therefore, the comparison is valid. However, the data difference has been explained in the revised manuscript.
Comment #15: "It is hoped that the KU glacier inventory and elevation change databases presented in this paper shall further help in promoting research in fields like climate change, hydrology, and other allied fields." This is common for any inventory. The rationale should be how KUGI will help to further it. This needs to be mentioned.
Response: Please see the response to the similar comment above. The novelty, value addition and usefulness of the KUGI over other inventories has been discussed in the revised manuscript and is reproduced in the authors response against the Comment#9.
Comment #16: Line 132: Need to place "and" between latitude and longitude values.
Response: “and” placed between the between latitude and longitude in the revised manuscript.
Comment #17: Line 135-36: "The area above 3600 m asl in general remains covered with perennial snow and glaciers". Not true mostly. This has to be ~4000 m asl for the J&K region.
Response: In general, areas above 3600 masl remain snow covered with snow for the entire year in the study area. However, there might be some exceptions. Therefore, as suggested the sentence has been modified in the revised manuscript to show that areas ~4000 m asl remain covered with snow.
Comment #18: Line 136-37: Why have authors quoted numbers from RGI inventory and not GAMDAM?
Response: Here, we are providing an overall estimate of the glacier cover over the entire Jammu, Kashmir and Ladakh region extracted from RGI and used in the previous study (Abdullah et al. 2020).
Comment #19: Line 138: "thus making the study area the most glaciated terrain". Reframe.
Response: The sentence has been rephrased as suggested in the revised manuscript.
Comment #20: Line 139-40: The authors quote Kamp et al. (2011) and suggest that glaciers are cirque-type in Ladakh which is not true with all the glaciers in the Suru basin and also neighboring Zanskar region. Glaciers in North Ladakh (Siachin area, Rimo group) aren’t cirque-type either. Please reframe.
Response: Agreed, there are a few exceptions. Therefore, the sentence has been modified in the revised manuscript.
Comment #21: Line 140-41: "All the major tributaries" instead of "Most of the major tributaries"
Response: The sentence has been modified as suggested in the revised manuscript.
Comment #22: Fig. 1: The caption should mention the following: What does the inset map represent? What is the background image (a DEM or what)? Have a legend for elevation if it is so. Mention may be "GA" for glacier area instead of "A" since "A" also represents Jhelum Basin. The text could be made bold and a little larger for histograms in the study area map. What is the source of number of glaciers, glacier area, debris cover, and glacier volume? Maybe plot glacier volume and debris cover on the secondary axis since the associated values are small.
Response: The suggested modifications in the figure have been incorporated in the revised manuscript.
Comment #23: Line 150-180: Could be better represented as a table. For climatology of Jhelum Basin use: Zaz et al 2019 (ACP). For Suru met data are improper. Authors used Schmidt and Nüsser, 2012 (which mentions a different area of Ladakh, Kang Yatze massif, and not Suru) while Chevuturi et al. (2018) report climate of Leh and not Suru. Need to correct it. Similarly for Chenab, the authors quote Azam et al. (for Chotta Shigri area). Why not cite Bhutiyani et al. 2007 (Climatic change) and Bhutiuyani et al 2010 (Int J of Climatology)?
Response: We agree that the meteorological data in Schmidt and Nüsser, (2012) is for the neighboring area, however, due to lack of meteorological data observations in the study region, we used the data from the referred paper assuming that the climatology of the Ladakh region does not vary much spatially from the Zanaskar region, as both being the part of the cold desert climatic zone. Again, due to non-availability of the meteorological data specific to Chenab basin, we cited Azam et al. which is located in the upper Chenab basin. Like, Chevuturi et al. (2018), Bhutiyani et al. (2007) also reported the climate data for the Ladakh Range based on the data available at the Leh station as there is no climate data available for the Zanaskar region. However, the time series in case of Chevuturi et al. (2018) is longer and as such, we cited the same in the manuscript. However, we have cited the relevant literature cited by the commentator in the revised manuscript.
Comment #24: Line 192: "The of moderate resolution". Please correct.
Response: The sentence have been corrected to “The use of ….” in the revised manuscript.
Comment #25: Line 193-96: Some of the references have been quoted above and do not necessarily need to be mentioned here.
Response: As suggested some of the references has been removed from here in the revised manuscript.
Comment #26: Table 1:
The authors mentioned wrong dates for imagery used for ICIMOD inventory. It is 2005±3 years (Weblink: https://lib.icimod.org/record/9419. See page 7>Section Satellite images> the second paragraph). I would again repeat my above question: Can 2000 data (in RGI, GAMDAM, and KUGI) be compared with 2005 data (as used in ICIMOD inventory)? Definitely not. Please justify. Also modify the respective entry in the Table. Details of all satellite scenes should be provided as supplementary data.
Response: Sorry for inadvertent typo and the same has been corrected in the table. The rest of the comment has already been responded to (Please see the response to Comment#14 above.)
Comment #27: Google Earth: Mention the date of Google Earth images, if at all they were used for correcting/validating glacier outlines. Maybe have a supplementary file for mentioning which glaciers were validated using Google Earth imagery.
Response: This comment has been already responded to above. All the debris-covered glacier and several clean numbering more than 850 have been validated/corrected using the Google Earth data. Google earth images are dated, 2009-2011 for the Jhelum basin: 2006-2013 for the Suru basin and 2000-2006 for Chenab basin have been used to verify the glacier outlines. The locational location of the glaciers verified on Google Earth is provided as KML file in the supplemental material of the revised manuscript.
Comment #28: Line 213-14: Delete " hereafter named Kashmir University Glacial Inventory" as it has already been defined as KUGI in the abstract.
Response: The sentence has been modified, as suggested, in the revised manuscript.
Comment #29: Line 214: Delete "global".
Response: The modifications, as suggested, have been incorporated throughout the revised manuscript.
Comment #30: Line 129: "acquired during 2002 to 2008". See my earlier comments.
Response: This has been explained in the response to Comment#14 above and specified in the revised manuscript.
Comment #31: Line 225-28: The authors mention "The RGI glacier outlines have been extracted semi-automatically from the Landsat satellite images between 1998 and 2009. However, most of the glaciers (~98%) in the inventory over the study area have been extracted from the images acquired during 1998-2002". How many glaciers were delineated from 1998-2002 data in RGI inventory? See attribute of RGI shapefile. Again comparison seems a problem here; not only due to dates but also the technique used.
Response: The information in the text is relevant for the entire Jammu, Kashmir and Ladakh region which has been specifically mentioned in the revised manuscript. Out of 15064 glaciers in RGI, 14894 glaciers were delineated from source images acquired between 1998-2002 (1998: 2228; 1999:2065; 2000:4034; 2001: 2117; 20002:3789; 2006:17; 2009:153). For the three basins under consideration, all the glaciers except one in the Suru basin, have been delineated from source images acquired between 2000-2002. This has been specified in the revised manuscript.
Comment #32: Line 244: What do authors mean by "limited field surveys"? How many glaciers (%) were ground surveyed? What kind of data was collected? Need to reflect all that in the MS.
Response: This is again a repeated comment and has been responded to above (Please see response to the Comment#2 above)
Comment #33: Line 256-57, 262, 267-270: "were verified from the Google EarthTM". The GE data for 2000 is almost not usable for the region. Please explain. Is it a deliberate attempt of misinformation or what?
Response: This is very unfortunate that the commentator is repeating the use of strong language like “Misinformation or what”. This is comment is repeated 4th time in the review. However, the comment stands already responded above. We in general appreciate the suggestions of the anonymous reviewer but the repeated use of the personal and unscientific language is unnecessary and therefore very unfortunate.
Comment #34: Line 262-63: "The thin debris layer on the glacier surface, often bearing lower surface temperature". Do the authors mean differential wrt ice or neighboring landscape?
Response: Yes, it is with reference to the neighboring landscape. The same has been mentioned in the revised manuscript.
Comment #35: Line 200, 280-282: The authors mention ASTER DEM here having been used to derive glacier-specific topographic parameters. But there is no mention of ASTER DEM in Table 1. Why was ASTER DEM used when CARTO DEM with a similar resolution is available over the region? Seems repetition of (line 201-202) here (280-82).
Response: The information pertaining GDEM2 has been incorporated in the revised manuscript. We preferred ASTER GDEM2 as its use for glacier studies is well established as mentioned in section 3.2. Furthermore, since both the DEMs have same spatial resolution and we did not find any study reporting any specific advantage of using CartoDEM for glacier inventory studies.
The repeating sentence at line 201-202 has been deleted in the revised manuscript.
Comment #36: Line 285-295: When the techniques used for mapping the glacier outlines are different, it is but obvious that there won't be a high overlap. Would it be so? Add a justification.
Response: Of course, there won’t be a high overlap and in fact we have already mentioned this (different techniques used for glacier mapping) as one of the reasons for the discrepancy observed in glacier outlines in terms of the overlap ratio
Comment #37: Remove section 4.3 as explained in the beginning and also uncertainty related to geodetic mass balance.
Response: As mentioned in response to Comment #8 above, we believe that the elevation change information will be a value addition to the database as such there is merit in retaining this section.
Comment #38: Line 330: Rp/Ap is a constant. What does it represent?
Response: Please refer to Paul et al, (2013). is the perimeter-area ration reported by Paul et al, (2013) in a detailed analysis aimed at the uncertainty assessment of glacier mapping which is equal to 5.03 km -1. In the present study the perimeter-area ratio however, varied from 0.96 – 6.22 km-1, we therefore applied a scaling after Braun et al, (2019) to determine the uncertainty in glacier delineation over the study region. We have incorporated the description in the revised manuscript.
Comment #39: Line 368: "Jhelum" instead of "Jehlum". Be consistent with spelling.
Response: The typo has been corrected in the revised manuscript.
Comment #40: Line 368: "The glaciers range in size from 0.01 km2". This means ~11 pixels. I wonder if such small-sized glaciers (1 ha as mentioned by authors) could be mapped from 30 m Landsat data? This would be highly uncertain. Could it be that some of them were snowpacks and not glaciers especially when ascertaining from 2000-02 data?
Response: Glaciers of the size (0.01 km2) have been previously mapped by Paul et al. (2002); Paul et al. (2009); Pfeffer et al. (2014); Abdullah et al (2020). Shukla et al. (2020) have also mapped glaciers with minimum size of 0.01 km2 in a recent study comprising a part of the study region using 30 m Landsat data. Also, the glaciers of the same size have been mapped in the GAMDAM inventory using Landsat 2000 data (Nuimura et al. 2015). So, mapping such glaciers from 30 m data is not a problem, however, to ensure that snow-packs are not misinterpreted for glaciers, we specifically checked satellite images dating before and after the satellite image under consideration. We also used Google earth imagery (post-dating) to verify that snow-packs are not misinterpreted as glaciers as already explained in response to your previous comment.
Comment #41: Table 2: The authors mention glacier volume but have not provided any information as to how ice volumes were derived? Did the authors use VA scaling and why if it is known that VA scaling estimates are highly uncertain, even for entire mountain ranges.
Response: The glacier volume was estimated using the slope-dependent volume estimation approach (Haeberli and Hoelzle, 1995), this has been mentioned in the manuscript.
Comment #42: Line 383-84: "mean glacier slope in the basin varies between 9° and 50°". The glaciers with a mean slope of 50 degrees are highly unlikely since by the definition such areas (> 30-degree slope) could be avalanche feeding zones. Please speak otherwise it raises concerns about the inventory itself.
Response: In the Jhelum basin, there is only one glacier with the mean slope of 49.7° and similarly there are a few more in the other basins. We rechecked these glaciers and found them situated at higher altitudes with very little exposed headwall area, and therefore, there is not enough terrain to form avalanche zone.
Comment #43: Table 3: Why do the authors need to mention glacier area categories from >20->50 in the Jhelum basin. >50 should not be the category. Let it be >50-the highest glacier area in the respective basin. Similarly, if there is a category 1-2, it should be followed with >2-5, >5-10, so on and so forth.
Response: As suggested, the non-relevant categories have been removed from the tables and the tables have been modified in the revised manuscript.
Comment #44: Line 395-397: Delete as it is already mentioned in 275-76.
Response: As suggested, the line is deleted in the revised manuscript.
Comment #45: Table 4: Elevation categories from 5500-7000 are not relevant for Jhelun and as such could be deleted. If the first category starts from <=4000, the next category should be >4000-4500, so on and so forth.
Response: The non-relevant categories have been removed from the tables in the revised manuscript.
Comment #46: Mention uncertainties about each A (glacier areas) in table 3, 4, 5 and 6.
Response: Thanks for the suggestion. The uncertainties for each area, elevation, slope and aspect class has been provided in the revised manuscript.
Comment #48: The uncertainty in TGA for the Jhelum basin is ~13.3% compared to Suru and Chenab (both 3.3%). Why is this so? This should have been the other way around since there are more clean-ice glaciers in Jhelum. Explain.
Response: This is again a repeating comment and has been already responded to (Please see the response to Comment#3 above)
Comment #49: How different are the estimates of Table 7 different from Scherler et al 2018 (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018GL080158)
Response: On comparison of KUGI with Scherler et al. (2018), We found that the DC estimates are in general higher compared to the Scherler et al. (2018) for all the three basins. For example, for the Jhelum basin, the DC estimate of 7.2 km2 is around 8% higher than Scherler et al. (6.6 km2).
Comment #50: Line 474: Delete "regional and global".
Response: Deleted in the revised manuscript.
Comment #51: Table 8: Should be "ΔN" instead of "N". Percentage difference could be mentioned in brackets.
Response: The table is modified as suggested in the revised manuscript with percentage (%) difference provided in brackets and the difference in number is represented by “ΔN” instead of “N”
Comment #51: Line 490-500: Use ICIMOD and GAMDAM rather than ICIMODGI and GAMDAMGI.
Response: GI was added to reduce the word count and also it becomes monotonous to append the phrase “glacier inventory” each time an inventory name is mentioned. Therefore, we are retaining GIs after global, and regional inventories.
Comment #53: Table 9: Above in methods authors mention having used median glacier elevation rather than mean glacier elevation.
Response: Sorry, we have used the mean rather than median and the same been corrected in the methods section of the revised manuscript.
Comment #54: Section 5.3 and associated tables/figures should be removed.
Response: As mentioned in response to Comment#8 and Comment#37, we find merit in retaining the elevation change information in the revised manuscript.
Comment #55: Line 675-680: The authors mention large glaciers at high altitudes and then low glacier cover at high altitudes which is a kind of contradiction. Large glaciers as authors suggest in results contribute to glacier cover. Please explain.
Response: The glacier cover observed in the study region is in general concentrated at higher altitudes for example above 4500 m asl in the Suru and Chenab basins which is justified by the elevation and temperature relationship for glacier growth. However, glacier coverage starts to decrease considerably with the further increase in the altitude e.g., above 5500 m asl in the Suru and Chenab which has been attributed to the steeper headwalls above this altitude facilitating snow/ice avalanches, thereby precluding the glacier formation. The sentences have been rephrased in the revised manuscript for more clarity and better understanding.
Comment #56: Table 14: Remove
Response: Please see our response to the suggestion above (Comment #37 and Comment#54)
Comment #57: The authors do not discuss much about the aspect (neither in results nor discussion) except very qualitatively.
Response: More details regarding aspect have been added in the revised manuscript.
Comment #58: Section 6.2 should be "Inconsistencies in existing glacier inventories"
Response: The heading has been modified, as suggested, in the revised manuscript.
Comment #59: Figure 5: Mention background image and band combination.
Response: The background images is FCC (7,4,2) of Landsat ETM+ dated 04-09-2000. The information has been added to the figure caption.
Comment #60: Line 720-724: The overlap ratio could be misleading since the inventories were computed using different methods. Please justify.
Response: Nagai et al. (2016) have demonstrated usefulness of the overlap ratio to assess the consistency of glacier outlines including location shifts which would be difficult to assess using the absolute value of delineated areas. Furthermore, the results of the overlap ratio observed in the present study are corroborated by the comparison of the glacier inventories in absolute terms (number and area). For example, the overlap ratio between KUGI-ICIMOD combination is relatively poor and the same is reflected by relatively larger differences in glacier number and area. Therefore, overlap ratio is quite useful indicator to assess the consistency of glacier inventory irrespective of the methodology used to delineate glacier boundaries.
Comment #61: Figure 6: SG. On checking from Google Earth, the existing inventories have rightly followed GLIMS definition of glacier delineation and divided the ice into two polygons since the ice fluxes move in different ablation zones. However, the authors have erred here (and maybe in many such cases) by considering it as a single glacier. The authors should know that although the ice masses are connected (in the accumulation zone) the movement of ice in different directions owing to the ridge-topography divides the ice into two polygons and hence two glaciers (refer to GLIMS definition of glaciers). This appears to be a huge flaw with the interpretation by authors as the number of glaciers is massively underestimated in KUGI. This needs to be corrected in the data as well as explanations to the data.
Response: We have followed the GLIMS criteria, by dividing a glacier into two or more polygons as determined by the underlying ridge topography during the inventory. It is fact that the latest Google Earth imagery shows that the glacier has fragmented, however, when we closely look at the Google Earth data, the glacier has fragmented or the under lying ridge divides the glacier far away from the position it was previously divided in case of GAMDAM and ICIMOD glacier inventories. Neither, GAMDAM nor the ICIMOD has divided the glacier where it appears to have fragmented in the recent years. Also, when we drape the 2000 satellite data over DEM, the ridge-topography is not prominent enough to manifest the flow direction/ridge divide on the satellite images especially in the accumulation zone. Therefore, there is no question of dividing a glacier into multiple polygons.
Comment #62: Figure 7: Mention background image and band combination.
Response: The image source and the band combination has been specified for the figure in the revised manuscript.
Comment #63: Remove section 6.3
Response: Please see our response to the similar comments above.
Comment #64: Conclusions:
While the authors mention "limited field survey" at many places in the text, they have failed to showcase the data collected, the photographs depicting the glacier environments in these three catchments. They should show field data and photographs from all three catchments and demonstrate its usefulness in inventorying the glaciers in these three respective areas.
Response: This is again repeating comment. Please see our response to Comment #2; Comment #9.
Comment #65: Delete sentences about geodetic mass changes.
Response: Again, as explained above, we want to retain this section in the revised manuscript.
Comment #66: The authors fail to convince the robustness of KUGI compared to at least RGI and GAMDAM. This needs to come up very in the results, discussion as well as the conclusion.
Response: This is a repeating comment. Please see the response to the similar comments above, particularly the author response to Comment #9.
-
AC3: 'Reply on CC1', Romshoo Shakil Ahmad, 08 Jun 2021
-
CC2: 'Comment on essd-2021-28', Sher Muhammad, 04 May 2021
The authors derive a new glacier inventory for selected Himalayan river basins using manual delineation and various data sources. The authors also highlight the strength of their inventory through the field data. The derived inventory is compared with RGI, ICIMOD, and GAMDAM inventories and highlighted the limitations in the mentioned inventories. In addition to the comparison of inventories, the authors estimated the surface elevation changes of glaciers in the basin between 2000 and 2012. It is important and interesting to see the comparison of various inventories (e.g., Muhammad et al., 2019a) to support the glaciological community to use the most appropriate inventory for their research. I only review part of the manuscript and suggest few comments to incorporate in the revision to strengthen their manuscript.
- Interesting to see that ICIMOD inventory is not only underestimating (as in the Karakoram (Muhammad et al., 2019a) but also overestimating. The main reason for underestimation in the Karakoram by ICIMOD inventory is the slope criteria. Most of the glaciers are avalanche-fed in the Karakoram and the accumulation falls on the steep slopes which is mostly not considered. However, I found that the inventory here shows that there is overestimation as well in the ICIMOD inventory. The authors are suggested to discuss the overestimation in ICIMOD inventory and its potential reasons and also discuss the results in comparison with Muhammad et al., 2019a.
- The authors manually digitize the glaciers which is extremely inconvenient in the presence of state of the art automatic techniques considering the >2000 glaciers. Mapping only a single (medium to large) glacier with manual digitization takes several hours. Usually, automatically derived extents are improved using manual digitization but the approach is different here. The authors may explain why they selected manual digitization.
- Also, it is unclear why the authors use topographic parameters if they use manual digitization? These parameters are useful when the glaciers are automatically mapped.
- The authors indicate field surveys data for glacier inventory validation but did not show the results of the survey anywhere (in any figure or text). The authors are suggested to add detailed information of the field survey including 1) the number of glaciers surveyed in the field, 2) what kind of information/data collected in the field, 3) how the survey information/data improved/validated the remote sensing results?
References
Farhan, SB, Zhang, Y, Ma, Y, Guo, Y and Ma, N (2015) Hydrological regimes under the conjunction of westerly and monsoon climates: a case investigation in the Astore Basin, Northwestern Himalaya. Clim. Dynam., 44(11–12), 3015–3032
Muhammad, Sher, Lide Tian, and Asif Khan. "Early twenty-first century glacier mass losses in the Indus Basin constrained by density assumptions." Journal of Hydrology 574 (2019a): 467-475.
Muhammad, S., Tian, L., & Nüsser, M. (2019b). No significant mass loss in the glaciers of Astore Basin (North-Western Himalaya), between 1999 and 2016. Journal of Glaciology, 65(250), 270-278. doi:10.1017/jog.2019.5
Citation: https://doi.org/10.5194/essd-2021-28-CC2 -
AC4: 'Reply on CC2', Romshoo Shakil Ahmad, 08 Jun 2021
CC#2
General Comment: The authors derive a new glacier inventory for selected Himalayan river basins using manual delineation and various data sources. The authors also highlight the strength of their inventory through the field data. The derived inventory is compared with RGI, ICIMOD, and GAMDAM inventories and highlighted the limitations in the mentioned inventories. In addition to the comparison of inventories, the authors estimated the surface elevation changes of glaciers in the basin between 2000 and 2012. It is important and interesting to see the comparison of various inventories (e.g., Muhammad et al., 2019a) to support the glaciological community to use the most appropriate inventory for their research. I only review part of the manuscript and suggest few comments to incorporate in the revision to strengthen their manuscript.
Response: We express our gratitude to the reviewer for suggestions and comments on the manuscript. The comparison of the present study with Muhammad et al., (2019) has been incorporated in the revised manuscript. However, the point by point response to the detailed comments and suggestions raised by the reviewer are provided as follows:
Comment#1: Interesting to see that ICIMOD inventory is not only underestimating (as in the Karakoram (Muhammad et al., 2019a) but also overestimating. The main reason for underestimation in the Karakoram by ICIMOD inventory is the slope criteria. Most of the glaciers are avalanche-fed in the Karakoram and the accumulation falls on the steep slopes which is mostly not considered. However, I found that the inventory here shows that there is overestimation as well in the ICIMOD inventory. The authors are suggested to discuss the overestimation in ICIMOD inventory and its potential reasons and also discuss the results in comparison with Muhammad et al., 2019a.
Response: The overestimation observed in case of the ICIMOD inventory is largely attributed to the misinterpretation of snowpacks as glaciers as demonstrated in the Fig. 5. Furthermore, the description of the results with respect to Muhammad et al. (2019) is provided in the revised manuscript. Both these points have been discussed in details in the revised manuscript.
Comment#2: The authors manually digitize the glaciers which is extremely inconvenient in the presence of state of the art automatic techniques considering the >2000 glaciers. Mapping only a single (medium to large) glacier with manual digitization takes several hours. Usually, automatically derived extents are improved using manual digitization but the approach is different here. The authors may explain why they selected manual digitization.
Response: Agreed that the automatic glacier boundary delineation followed by manual correction is one of the preferred approach for glacier mapping from satellite images especially over large regions because of the less time required when compared to the manual digitization techniques. However, automatic glacier delineation technique poses a significant challenge for mapping debris covered glaciers particularly the glacier terminus in the Himalaya. In fact, the reflectance of the supra-glacial debris cover is similar to the surroundings which results in the exclusion of such areas from the glacier extents. Furthermore, seasonal snow, cloud cover and shadow also pose a significant challenge in mapping Himalayan glaciers using automatic image delineation techniques. Therefore, to overcome these challenges in mapping glaciers in the Himalaya, we used multiple data sets including thermal data, high resolution imagery, time series of satellite data etc. which is not possible to use in the automatic approach. Furthermore, the local knowledge/field experience of an analyst also proves very useful in precise glacier delineation which is again not possible in the automatic approach. Since a considerable number of glaciers in the present study have debris-covered termini and we found it appropriate and necessary to map the glaciers manually to minimize the errors and uncertainties in the glacier inventory. Furthermore, advantage of the manual approach over the automatic approach for mapping glaciers when debris covered, shadowed and seasonal snow-covered area has been previously demonstrated by Paul et al. (2013); Nuimura et al. (2015).
KUGI improved the mis-mapped glacier outlines/boundaries from the automatic approach and any mismatches of the glacier geometry due to the seasonal/temporal snow cover and shadows were manually corrected using additional Landsat images and Google earth images. Further the mapped glaciers with better geo-referencing were overlaid with high resolution images in Google Earth environment for validation wherever available. Though, the mis-mapped/mis-located outlines, observed on the global/regional inventories, may have only limited effect on measurements of glacier area, but can introduce serious errors into applications that rely on absolute positioning (e.g. co-registration to other datasets such as DEMs). The only realistic way to correct them is to provide more accurate outlines using manual approach as was done in the KUGI which would serve as source of improved outlines for the scientific community interested in conducting various application studies using the glacier outlines.
Comment#3: Also, it is unclear why the authors use topographic parameters if they use manual digitization? These parameters are useful when the glaciers are automatically mapped.
Response: As specified in the methodology section (section 4.1), under the surface conditions on headwalls with slopes (topographic parameter) exceeding 40˚, we specifically verified such glaciers from the Google Earth image for accurate delineation of glacier extents. Further, the satellite images draped on DEM (hillshade) was found useful in demarcating glacier outlines when the ridges particularly in the accumulation zone were covered with seasonal snow (Paul et al. 2004; Paul et al. 2009). The overall visualization of a glacier in 3D helped in the precise mapping of glacier outlines in KUGI.
Comment#4: The authors indicate field surveys data for glacier inventory validation but did not show the results of the survey anywhere (in any figure or text). The authors are suggested to add detailed information of the field survey including 1) the number of glaciers surveyed in the field, 2) what kind of information/data collected in the field, 3) how the survey information/data improved/validated the remote sensing results?
Response: In the present study, we have done field surveys on 20 glaciers located across the study area, which we visit almost annually for other glaciological studies. The locations and the field photographs of these glaciers has been provided as Supplementary figure in the revised manuscript. We have collected the snout positional data of the debris-covered tongues of these glaciers to validate the glacier delineation. The field measurements of these glaciers acted as an interpretation tool for delineation of the debris-covered glaciers in the study area. The terminus of the heavily debris covered glaciers like the Hoksar glacier in Jhelum basin was not easily mappable even using the thermal and google earth imagery. We therefore, mapped the glacier terminus in the field and further observed that the debris on the glacier is relatively smooth and aligned in the direction of glacier flow when compared to the debris-cover in the surroundings which was found a useful field-based information for mapping debris-covered termini of other glaciers when viewed on Google earth images. Further, eight of these twenty glaciers have been designated as benchmark glaciers and are continuously studied for mass balance, GPR, debris thickness, Surface mapping temperature profiling and other glaciological studies since the last 5-8 years. In addition, the glacier outlines of several other glaciers in the vicinity of these 8 glaciers in the three basins have been verified during annual glacier field expeditions during the last 5-8 years.
Additionally, all the heavily debris covered glaciers and a majority of the clean glaciers, numbering more than 850, were qualitatively verified on the Google Earth image for the rectification of any delineation error. A similar approach of quality check using Google Earth has been previously adopted by Nagai et al. (2016) and several others and is an accepted method for validation of such a large number of glaciers located in inaccessible complex terrain.
Field photographs of the about 20 selected glaciers in the three basins, which have been visited over the last few years for field measurements/validation along with their GLIMS ID are presented in the revised manuscript (The field photographs of some of these glaciers are provided below). A locational map of these glaciers in the form of a KML file would be provided as a supplemental material in the revised manuscript ((The field photographs have also been attached as a supplement file in pdf format)).
Comment#5: References
Farhan, SB, Zhang, Y, Ma, Y, Guo, Y and Ma, N (2015) Hydrological regimes under the conjunction of westerly and monsoon climates: a case investigation in the Astore Basin, Northwestern Himalaya. Clim. Dynam., 44(11–12), 3015–3032
Muhammad, Sher, Lide Tian, and Asif Khan. "Early twenty-first century glacier mass losses in the Indus Basin constrained by density assumptions." Journal of Hydrology 574 (2019a): 467-475.
Muhammad, S., Tian, L., & Nüsser, M. (2019b). No significant mass loss in the glaciers of Astore Basin (North-Western Himalaya), between 1999 and 2016. Journal of Glaciology, 65(250), 270-278. doi:10.1017/jog.2019.5
Response: The suggested literatures references have been incorporated in the revised manuscript.
Status: closed
-
RC1: 'Comment on essd-2021-28', Anonymous Referee #1, 08 Apr 2021
The authors present a new glacier inventory for north-western Himalaya, which is mainly based on manual glacier mapping using different data sources. They compared the new inventory with existing inventories and figured out limitations and differences of the individual inventories. Additionally, the authors used InSAR DEMs to compute glacier elevation changes between 2000 and 2012 of the study area.
The paper is well structured but the sections regarding the glacier elevation/mass change computations are very unclear and confusing. The authors are talking about mass balances but never provide any mass balance values. There are many flaws in the respective sections. Thus, I would suggest to remove the elevation/mass change computation sections completely, since the main focus of the paper is the evaluation of the glacier inventories.
Moreover, the results sections are too long and can be strongly condensed by focusing on tables and graphs.
At many places it is unclear, if the authors talk about mean/media values of certain variables (e.g. elevation, slope ….) or pixel wise values. A more precise wording is needed throughout the paper. (see details below).
The computation of several “uncertainty” values is unclear. Please provide formulas (see details below)
The computation of the average aspect values is unclear and might be buggy (see details below)
The comparison of the different glacier inventories is OK but can be certainly extended. It would be interesting to compute the the overlap ratio r_ov also for e.g. DC, clean and shadowed glaciers to evaluate the difference between the inventories.
It is also unclear, if the topographic parameters of the other inventories were taken form the inventory meta data or computed by the authors. The used DEMs might differ. Thus, it would be more meaningful to use a consistent source for topographic information before doing the comparison.
Once, the paper is revised it should be properly proof read. I am not a native speaker, but I got the feeling that the English can be improved. Many sentences are quite complicated and unclear or maybe got just grammatical errors.
Detailed comments (* significant issues):
L31: delete “for the study area”
l35: by “a” Digital…
l37: to glacier areas…
L40: are you talking about the mean or median glacier elevation?
L44: whats the meaning of the “R” values. Completely unclear.
L48: 2000 an 2012
L71: what about Brun et al. 2017
L97 and following: please list here more recent publications
l104: please list some of the variables
l106: why is the reproducibility not assured? Not clear.
L109: what about Brun et al 2017, Shean et al. 2020?...
L110 and following: please move the comparison to the discussion section.
L118: which basins and where? Not introduced
L123: there exist already elevation change data sets for the same period (Brun et al.2017, Shean et al. 2020). So there exists already information on the glacier behavior.
L125: please rephrase this sentence. A quite weak motivation for this study.
L130: UIB not introduced
L132: “and” 73….
l136: when is this area covered? All year long?
Fig1: Please provide country borders and names in the overview map (upper right corner) for a better orientation. Please indicate the glacier coverage also outside the 3 basins. What are the sources of glacier outlines, debris cover and glacier volume?
*L154: are you talking about mean or median altitudes? Not clear, the same for the other basins in the following.
L163: … in the northeast of the study area..
L192: ...use of…
*Table1: could you please add the Path and Row numbers of the Landsat data. ASTER GDEM not listed. URL for ICIMOD inventory is missing., please provide also the date ranges of the inventories for your study area
l208: add “C-band”
l211: please introduce the abbreviation “DEM” at the place, where it is used the first time.
*l209: please rephrase. TanDEM-X is still acquiring data. You are talking about the worldDEM phase
l234: between or only in 1999 and 2003
l298: no capital letters for Base and Target
Section 4.3: This section is a bit unprecise and many details are missing. e.g. which DEMs did you use? How did you estimate the penetration bias
l350: cite here Rolstad et al. 2009
*l352: Seehaus et al. 2020, did not use the total glacier area for A. They used the area of each glacier complex.
Section 4.4. b) This section is quite confusing and the equation to compute the uncertainty of the mass balance is missing. Please revise the whole section and use clear and individual variables!
*l365: How did you compute the glacier volume ? Not mentioned in the Methods Section
Table 2: How did you compute the glacier volume? Why does it differ so strongly e.g. between KUGI and RGI at Jhelum? How did you assume the uncertainty in glacier area for the different inventories? Not explained!
380 and following: are you talking about mean or media elevations? Or the total elevation span of the whole glacier?
Table 3,4,5,6,7: Units are missing. What means “A” and “N” and “DC”? not clear
l396. delete sentence. Already mentioned in the methods
*Section 5.1: The whole Section can be strongly condensed. All information can be found in the tables and does not need to be repeated in the text.
*Table 6: how did you compute the average glacier aspects? Please provide the formula somewhere.
l483: how did you estimate the variations? Not explained!!!
*Section 5.2: Same as for Section 5.1.! It can be strongly condensed and most of the information can be summarized in nice tables and/or graphs. The text is very long and the information is hard to find. Tables and graphs would be beneficial for the reader
*Table 10: How did you define the elevation category? All pixels with in the interval? Or all glaciers with mean/median elevation within this interval? Unclear! How did you compute the uncertainties?
Fig. 2,3,4: Please add a background. The glacier outlines would be also nice. The bar plot is too small an impossible to read. Does it show the mean elevation changes per glacier? Explain!
l629: Are you talking about average elevation changes per glacier? Please clarify.
Table 11: is the slope take pixel by pixel or is it base on the mean slope per glacier?
Table 12: same as for Table 11. is the aspect take from each pixel or the mean of each glaciers
l659: Maybe the bigger glaciers are located at lower altitudes? Please check
Table 13: Units are missing for area
l685: By inspecting Fig. 2-4, it looks like most glaciers are not south facing. Please check your aspect computation! Do not use a simple mean of all pixel wise aspect values of a glacier. See RGI6.0 technical report.#
Fig.5: Date and source of background image?
Fig.6: Date and source of background image? There are at least 2 glacier tongues. Most likely they were connected in the past, but the mapped state shows 2 individual major glacier tongues. Therefore it is not wrong to split the glacier are in 2 polygons. Please rephrase accordingly, also in the main text.
Fig. 7: the outlines are hard to see. Use different colors or wider lines. Date and source of background image?
Table 16: Can be merged with Table 2 to avoid doubling of data.
l794: Unclear sentence
Citation: https://doi.org/10.5194/essd-2021-28-RC1 -
AC1: 'Reply on RC1', Romshoo Shakil Ahmad, 08 Jun 2021
Reviewer #1
General Comment: The authors present a new glacier inventory for north-western Himalaya, which is mainly based on manual glacier mapping using different data sources. They compared the new inventory with existing inventories and figured out limitations and differences of the individual inventories. Additionally, the authors used InSAR DEMs to compute glacier elevation changes between 2000 and 2012 of the study area.
The paper is well structured but the sections regarding the glacier elevation/mass change computations are very unclear and confusing. The authors are talking about mass balances but never provide any mass balance values. There are many flaws in the respective sections. Thus, I would suggest to remove the elevation/mass change computation sections completely, since the main focus of the paper is the evaluation of the glacier inventories.
Moreover, the results sections are too long and can be strongly condensed by focusing on tables and graphs.
Response: We express our gratitude to the reviewer for the elaborative and useful section-wise and line-by-line detailed review of the manuscript. We have responded point-by-point to all the comments and suggestions of the reviewer. The concerns of the reviewer regarding elevation change have been addressed entirely to the best of our ability and knowledge. The valuable comments and suggestions provided by the reviewer have greatly helped to improve the contents and quality of the manuscript.
Furthermore, we agree with the reviewer that the manuscript is focused on the evaluation of the glacier inventories, however, we believe that the elevation change information of the three basins will add value to the database, providing, in addition to the glacier inventory, a quick overview of the glacier elevation and mass changes of the glaciers in the database. As such we find merit in retaining the elevation change section while addressing the concerns of the reviewer about elevation change estimation. The point-by-point response to the detailed comments and suggestions raised by the reviewer is provided as follows:
Comment #1: At many places it is unclear, if the authors talk about mean/media values of certain variables (e.g. elevation, slope ….) or pixel wise values. A more precise wording is needed throughout the paper. (see details below).
Response: We have used the mean value of the topographic variables, obtained by averaging single-cell elevation, slope and aspect values from the DEM. This has been specified in the revised manuscript.
Comment #2: The computation of several “uncertainty” values is unclear. Please provide formulas (see details below)
Response: Clarification/correction provided wherever suggested in the revised manuscript. A detailed response is provided against each specific comment below.
Comment #3: The computation of the average aspect values is unclear and might be buggy (see details below)
Response: As mentioned in the manuscript, the aspect was calculated by averaging the aspect value of DEM cells within the extents of each glacier. However, the aspect has been now recalculated in the revised manuscript as specified in the RGI technical document. The aspect sines and cosines of each of the glacier’s DEM grid cells were summed and the mean aspect was calculated as the arctangent of the quotient of the two sums. The calculated values of aspect were transformed to the eight cardinal directions (N, NE, E, SE, S, SW, W, and NW) with each cardinal direction having range of 45 degrees, half to each side.
Comment #4: The comparison of the different glacier inventories is OK but can be certainly extended. It would be interesting to compute the the overlap ratio r_ov also for e.g. DC, clean and shadowed glaciers to evaluate the difference between the inventories.
Response: Agreed, that the overlap ratio of different glacier classes will be more useful, however, it is pertinent to mention that the debris covered portions of the glaciers, particularly in the ICIMOD inventory, have been excluded from the glacier extents as such the same glacier classified as debris-covered in the KUGI gets classified as “clean” (as the debris covered area is excluded) in ICIMOD for example, thereby making it a bit difficult to find true overlap ratios.
Comment #5: It is also unclear, if the topographic parameters of the other inventories were taken from the inventory meta data or computed by the authors. The used DEMs might differ. Thus, it would be more meaningful to use a consistent source for topographic information before doing the comparison.
Response: Since the source DEMs for topographic information in the existing inventories are different, the topographic parameters for the KUGI and other inventories, evaluated in this study, were derived from the ASTER GDEM2, which is now specified in the revised manuscript.
Comment #6: Once, the paper is revised it should be properly proof read. I am not a native speaker, but I got the feeling that the English can be improved. Many sentences are quite complicated and unclear or maybe got just grammatical errors.
Response: Thanks for the suggestion, we have checked and revised the manuscript for any grammatical errors. We have revised and rephrased a few complex sentences in the revised manuscript for more clarity and better understanding. As suggested, we shall get the revised manuscript check by an English language editor/faculty in the University.
Detailed comments (* significant issues):
Comment #7: L31: delete “for the study area”
Response: The phrase “for the study area” has been deleted from the revised manuscript as suggested.
Comment #8: l35: by “a” Digital…
Response: As suggested, the Digital Elevation Model has been prefixed by the letter “a” in the revised manuscript.
Comment #9: l37: to glacier areas…
Response: As suggested “area” replaced by “areas” in the revised manuscript.
Comment #10: L40: are you talking about the mean or median glacier elevation?
Response: It is the mean glacier elevation and has been specified in the revised manuscript.
Comment #11: L44: whats the meaning of the “R” values. Completely unclear.
Response: The overlap ratio of individual glaciers is represented by “rov” whereas, the “R” (RBA) has been used to represent the average overlapping ratio of the base and target glacier inventories (overlap ratio averaged for all the glaciers in a particular inventory combination e.g, KUGI-RGI for individual basins). This has been described under the methodology section in the revised manuscript.
Comment #12: L48: 2000 an 2012
Response: The typo has been corrected as “…2000 and 2012” in the revised manuscript
Comment #13: L71: what about Brun et al. 2017
Response: The references has been incorporated in the revised manuscript.
Comment #14: L97 and following: please list here more recent publications
Response: More recent publications suggested in the “general comments” and several other relevant publications suggested by other reviewers have been incorporated in the revised manuscript.
Comment #15: l104: please list some of the variables
Response: The variables include glacier number and area, which has been specified in the revised manuscript.
Comment #16: l106: why is the reproducibility not assured? Not clear.
Response: Since different inventories use different approaches, datasets and at times the definition of the glacier itself varies among the inventories, therefore reproducibility is a challenge. Furthermore, in case of the manually or at times the semi-automatically delineated glacier boundaries, glacier area estimate will depend on the perception of the analyst as such the results are often not reproducible.
Comment #17: L109: what about Brun et al 2017, Shean et al. 2020?...
Response: We agree that the mentioned studies have reported elevation/mass changes over the Himalayan region and have therefore been cited at appropriate places in the revised manuscript. However, here we have specifically mentioned the studies where the dataset is publicly available.
Comment #18: L110 and following: please move the comparison to the discussion section.
Response: As suggested, the section has been moved to the Discussion section in the revised manuscript.
Comment #19: L118: which basins and where? Not introduced
Response: The basin names (i.e., Jhelum, Suru and Chenab) has been specified in the revised manuscript.
Comment #20: L123: there exist already elevation change data sets for the same period (Brun et al.2017, Shean et al. 2020). So there exists already information on the glacier behavior.
Response: Agreed that the elevation change studies already exist over the region, however, it is pertinent to mention that these studies are carried over a larger spatial domain. In the present study, we carried out the elevation changes at local scale, furthermore, we also tried to assess the impact of topographic and morphological parameters including glacier size, DC, elevation, slope and aspect on the elevation changes which is included in the database.
Comment #21: L125: please rephrase this sentence. A quite weak motivation for this study.
Response: Thanks for the comment. As suggested, we have modified the motivation for the research work in the revised manuscript keeping in view the following argument.
Primarily, the motivation for the KUGI is to develop a high-resolution glacier inventory with improved accuracy with visual interpretation and manual delineation of glaciers from Landsat satellite data supported by the limited ground truth and supplementing the glacier outlines with additional data like debris-cover, thickness changes and other glaciological parameters, that are either missing or incorrect in the existing databases so that the database is made available to the large research community for various applications.
Purportedly the global and regional glacier databases that were chosen for comparison in this study have been generated using a semi-automated method allowing less human error, quick delivery, and high accuracy. However, it was found in this study that there are significant errors in the evaluated databases due to the misinterpretation of seasonal snow cover particularly on the glacier headwalls at high altitudes, shadow-covered glaciers and debris-cover.
Keeping in view the worldwide use and applications of global and regional databases, it is important that a rigorous evaluation of these global and regional inventories is undertaken for the continued refinement of the methodology which is a fundamental requirement for any meaningful application of the global or regional database. It is hoped that the future releases of the databases will improve these and other shortcomings identified in this manuscript.
Comment #22: L130: UIB not introduced
Response: UIB stands for Upper Indus Basin and the full form of the UIB has been incorporated upfront in the revised manuscript.
Comment #23: L132: “and” 73….
Response: The word “and” has been inserted between the latitude and longitude values in the revised manuscript.
Comment #24: l136: when is this area covered? All year long?
Response: Most parts of the study area above 3600 m asl remain snow-covered for the entire or most of the year. The sentence has been modified accordingly in the revised manuscript for more clarity and instead of 3600 m, we have mentioned ~4000 m asl to include that all the areas covered with snow.
Comment #25: Fig1: Please provide country borders and names in the overview map (upper right corner) for a better orientation. Please indicate the glacier coverage also outside the 3 basins. What are the sources of glacier outlines, debris cover and glacier volume?
Response: Thanks for the suggestion. The figure has been modified accordingly in the revised manuscript. The glacier outlines and the debris cover information is based on the KUGI. Further, the glacier volume has been derived using the slope-dependent scaling approach with the glacier area and slope information derived from KUGI.
Comment #26: *L154: are you talking about mean or median altitudes? Not clear, the same for the other basins in the following.
Response: These are the mean elevations as mentioned earlier and this has been specified in the revised manuscript.
Comment #27: L163: … in the northeast of the study area..
Response: As suggested “east” has been replaced by “northeast” in the revised manuscript.
Comment #28: L192: ...use of…
Response: Sorry for the typo, the word “use” this has been corrected in the revised manuscript.
Comment #29: *Table1: could you please add the Path and Row numbers of the Landsat data. ASTER GDEM not listed. URL for ICIMOD inventory is missing., please provide also the date ranges of the inventories for your study area
Response: The suggested missing information pertaining to the data sets used in developing the glacier database has been incorporated in the revised manuscript. The images with path/rows 149/36, 148/36, 148/37, 147/37 dating between 2000 and 2002 have been used for the inventory development.
Comment #30: l208: add “C-band”
Response: As suggested the band information has been incorporated in the revised manuscript.
Comment #31: l211: please introduce the abbreviation “DEM” at the place, where it is used the first time.
Response: The full form of the DEM (Digital elevation Model) has been mentioned at the first occurrence (in abstract section) in the revised manuscript.
Comment #32: *l209: please rephrase. TanDEM-X is still acquiring data. You are talking about the worldDEM phase
Response: Agreed, TanDEM-X mission is still acquiring the data, the information provided in the manuscript is relevant for the product version of the DEM (used in the present study) released in 2018 only. The text has been therefore modified accordingly in the revised manuscript.
Comment #33: l234: between or only in 1999 and 2003
Response: It is between as the authors have used data sets dated 1999, 2000, 2001 and 2002 for the generating the glacier inventory.
Comment #34: l298: no capital letters for Base and Target
Response: As suggested, the words “Base” and “Target” have been uncapitalized in the revised manuscript.
Comment #35: Section 4.3: This section is a bit unprecise and many details are missing. e.g. which DEMs did you use? How did you estimate the penetration bias
Response: The missing details and further description pertaining to the methodology has been provided in the revised manuscript. The penetration bias was computed as a function of altitude after Vijay and Braun, (2016).
Comment #36: l350: cite here Rolstad et al. 2009
Response: The reference has been incorporated in the revised manuscript.
Comment #37: *l352: Seehaus et al. 2020, did not use the total glacier area for A. They used the area of each glacier complex.
Response: Agreed that Seehaus et al. (2020) used the area of each glacier complex. We also used the term to represent the analyzed area.
Comment #38: Section 4.4. b) This section is quite confusing and the equation to compute the uncertainty of the mass balance is missing. Please revise the whole section and use clear and individual variables!
Response: The uncertainty of glacier-wide specific elevation change (Δh) is computed as:
The , , and are uncertainty of DEM differencing, uncertainty due to void filling (since the DEMs especially SRTM has voids over the study area as such the DEMs coverage for each having voids >30% were excluded from the analysis whereas, the glaciers with <30% voids were filled with nature neighbor interpolation algorithm), temporal uncertainty of TanDEM-X and uncertainty of radar signal penetration respectively. The uncertainty of each of the individual parameter is described in detail in the revised manuscript.
Comment #39: *l365: How did you compute the glacier volume ? Not mentioned in the Methods Section
Response: The glacier volume was estimated using the slope-dependent volume estimation approach (Haeberli and Hoelzle, 1995), the methodology to estimate the glacier volume has been incorporated in the revised manuscript.
Comment #40: Table 2: How did you compute the glacier volume? Why does it differ so strongly e.g. between KUGI and RGI at Jhelum? How did you assume the uncertainty in glacier area for the different inventories? Not explained!
Response: As mentioned, we used the slope dependent approach for volume estimation. The difference in volume estimates between KUGI and RGI is most probably due to the difference in the number of glaciers in the inventories.
Comment #41: 380 and following: are you talking about mean or media elevations? Or the total elevation span of the whole glacier?
Response: These are the mean elevation values.
Comment #42: Table 3,4,5,6,7: Units are missing. What means “A” and “N” and “DC”? not clear
Response: The “A” represents glacier area in km2, “DC” is glacier debris cover again in km2, whereas, “N” indicates the glacier number (count). The units as well as the description of each letter(s) has been specified in the revised manuscript in all the tables.
Comment #43: l396. delete sentence. Already mentioned in the methods
Response: The sentence has been removed from the revised manuscript as suggested.
Comment #44: *Section 5.1: The whole Section can be strongly condensed. All information can be found in the tables and does not need to be repeated in the text.
Response: Thanks for the suggestion. Accordingly, we have trimmed this section in the revised manuscript.
Comment #45: *Table 6: how did you compute the average glacier aspects? Please provide the formula somewhere.
Response: As mentioned above, the aspect has been now recalculated in the revised manuscript. Kindly see the response to Comment #3. The aspect sines and cosines of each of the glacier’s DEM grid cells were summed and the mean aspect was calculated as the arctangent of the quotient of the two sums.
Comment #46: l483: how did you estimate the variations? Not explained!!!
Response: By the variation, we mean the difference not the statistical variance. The sentence has been rephrased accordingly for clarity in the revised manuscript.
Comment #47: *Section 5.2: Same as for Section 5.1.! It can be strongly condensed and most of the information can be summarized in nice tables and/or graphs. The text is very long and the information is hard to find. Tables and graphs would be beneficial for the reader
Response: Thanks, as suggested the information has been already put in the form of tables and were referred at appropriate places in the revised manuscript. Further, the section has been modified/condensed in the revised manuscript as suggested.
Comment #48: *Table 10: How did you define the elevation category? All pixels with in the interval? Or all glaciers with mean/median elevation within this interval? Unclear! How did you compute the uncertainties?
Response: In each elevation category, glaciers are grouped together in the elevation bins based on the mean elevation of individual glaciers. This has been specified in table caption in the revised manuscript. Furthermore, to assign the uncertainties for a sample average (for example glaciers in the elevation range between 5000-5500), we calculated the uncertainty of the sample-wide elevation change ( ) using the following (Huber et al.) , where σ_Δh is the uncertainty of each item i (glacier in our case) and n is the number of item in the sample.
Comment #49: Fig. 2,3,4: Please add a background. The glacier outlines would be also nice. The bar plot is too small an impossible to read. Does it show the mean elevation changes per glacier? Explain!
Response: Thanks for the suggestion. The figures have been modified accordingly. Yes, you are right, the bar graph is based on the mean elevation change per glacier.
Comment #50: l629: Are you talking about average elevation changes per glacier? Please clarify.
Response: Yes, we are talking about average elevation changes per glacier. This has been specified in the figure captions in the revised manuscript.
Comment #51: Table 11: is the slope take pixel by pixel or is it base on the mean slope per glacier?
Response: The slope is based on the mean slope per glacier. Specified in the table caption in the revised manuscript.
Comment #52: Table 12: same as for Table 11. is the aspect take from each pixel or the mean of each glaciers
Response: The aspect is based on the mean aspect of each glacier. Specified in the table caption in the revised manuscript.
Comment #53: l659: Maybe the bigger glaciers are located at lower altitudes? Please check
Response: Yes, the glaciers tend to be smaller at higher altitudes in the study area. For example, glaciers in the Suru and Chenab basins situated above 5500 m asl are generally smaller compared to the glaciers situated between 5000-5500 m asl (Table 4 in the revised manuscript).
Comment #54: Table 13: Units are missing for area
Response: Units (km2) of area have been provided in the revised manuscript.
Comment #55: l685: By inspecting Fig. 2-4, it looks like most glaciers are not south facing. Please check your aspect computation! Do not use a simple mean of all pixel wise aspect values of a glacier. See RGI6.0 technical report.#
Response: As suggested, the aspect was recalculated in the revised manuscript as specified in the RGI technical document and accordingly, the correct aspect is depicted in the revised manuscript.
Comment #56: Fig.5: Date and source of background image?
Response: The background image is FCC (7,4,2) of Landsat ETM+ dated 04-09-2000. The information has been added to the figure caption.
Comment #57: Fig.6: Date and source of background image? There are at least 2 glacier tongues. Most likely they were connected in the past, but the mapped state shows 2 individual major glacier tongues. Therefore, it is not wrong to split the glacier are in 2 polygons. Please rephrase accordingly, also in the main text.
Response: The date and source of the image has been provided in the revised manuscript. We, agree that there are two tongues, however, when we drape the 2000 satellite images over DEM, the ridge-topography is not prominent enough to manifest the flow direction/ridge divide on the satellite images especially in the accumulation zone, as such we did not divide the glacier in multiple polygons.
Comment #58: Fig. 7: the outlines are hard to see. Use different colors or wider lines. Date and source of background image?
Response: Thanks for the suggestion. Accordingly, the figure has been modified as suggested for better presentation of the outlines in the revised manuscript.
Comment #59: Table 16: Can be merged with Table 2 to avoid doubling of data.
Response: We have merged Table 16 with Table 8 to represent the difference in glacier number and area in the revised manuscript. The information in the Table 16 seems more relevant to Table 8, therefore the repeating information (area) has been removed from the merged tables in the revised manuscript.
Comment #60: l794: Unclear sentence
Response: The sentence attributes the higher thinning of south oriented glacier to the higher solar insolation received by southerly slopes. The sentence has been rephrased in the revised manuscript for clarity.
Citation: https://doi.org/10.5194/essd-2021-28-AC1
-
AC1: 'Reply on RC1', Romshoo Shakil Ahmad, 08 Jun 2021
-
RC2: 'Comment on essd-2021-28', Anonymous Referee #2, 16 Apr 2021
This is a useful contribution and in general seems a well-conducted assessment. Avoiding measureemnt bias is paramount for geodetic mass balance in particular, and some more evidence is needed to demonstrate that this has been achieved.
Sections 2.1, 2.2 and 2.3: please check the Jhelum minimum temperature - should be negative? Please provide consistent descriptive stats, e.g. annual precip in each case rather than mixing monthly and annual.
Line 279 - area calculation in 'ArcGIS environment' - what projection was used? Please ensure that an area-preserving projection is used when defining the glacier areas and, by extension, their volume changes.
Equation 6: good error quantification is vital in geodetic mass balance calculations. Please explain the sigma_z DEM uncertainty term. Is this a generic estimate of DEM quality, or is it specifically applicable to snow-covered surfaces and steep slopes? These are notoriously difficult to map topographically, particularly in opticial images. The voidfill error term is not defined. Please explain and justify using the penetration error as a random, uncorrelated error in this case. If penetration is wrong, it will be a systematic error and so should be added and not combined in quadrature.
Line 342: error assessment in off-glacier areas is good to do, but this is not reported or shown in the dh/dt figures. If it reveals systemtaic biases then these should be corrected to zero. This requires assessment at as many off-glacier locations as possible, at a range of altitudes, as the DEM biases are often not uniform across a scene. The iniital and corrected off-glacier stats should be reported and shown in the figures.
Table 2: KUGI glacier volumes - the calculation of volume is not trivial and is not explained. Where do these come from?
Figures 2 and 3: please show a background map, scale bar and the apparent dh/dt values off-glacier.
Citation: https://doi.org/10.5194/essd-2021-28-RC2 -
AC2: 'Reply on RC2', Romshoo Shakil Ahmad, 08 Jun 2021
Reviewer #2
General Comment: This is a useful contribution and in general seems a well-conducted assessment. Avoiding measurement bias is paramount for geodetic mass balance in particular, and some more evidence is needed to demonstrate that this has been achieved.
Response: We express our gratitude to the reviewer for the very useful review of the manuscript. The valuable comments and suggestions provided by the reviewer have improved the contents of the manuscript. We have responded point-by-point to all the comments and suggestions of the reviewer. The concerns of the reviewer about error estimates have been addressed entirely to the best of our ability and knowledge. The revised manuscript looks significantly improved. The point-by-point response to the detailed comments and suggestions raised by the reviewer are provided as follows:
Comment#1: Sections 2.1, 2.2 and 2.3: please check the Jhelum minimum temperature - should be negative? Please provide consistent descriptive stats, e.g. annual precip in each case rather than mixing monthly and annual.
Response: Thanks for the suggestion, we rechecked the mean minimum temperature and found it above zero, this has been previously reported. These sections have been however revised providing temperature and precipitation on same time scales.
Comment#2: Line 279 - area calculation in 'ArcGIS environment' - what projection was used? Please ensure that an area-preserving projection is used when defining the glacier areas and, by extension, their volume changes.
Response: In the present study we used the Mercator projection system considered suitable for use in areas between 84°N to 80°S. Also since all the inventories under consideration use the same projection system, therefore, it is assumed that any distortion in area would be same for all the inventories and would not affect the comparative analyses of the inventories evaluated in this study.
Comment#3: Equation 6: good error quantification is vital in geodetic mass balance calculations. Please explain the sigma_z DEM uncertainty term. Is this a generic estimate of DEM quality, or is it specifically applicable to snow-covered surfaces and steep slopes? These are notoriously difficult to map topographically, particularly in opticial images. The voidfill error term is not defined. Please explain and justify using the penetration error as a random, uncorrelated error in this case. If penetration is wrong, it will be a systematic error and so should be added and not combined in quadrature.
Response: Thanks for the comment and suggestions. The , , and are uncertainty of DEM differencing, uncertainty due to void filling (Since the DEMs especially SRTM has voids over the study area as such the DEMs coverage for each having voids >30% were excluded from the analysis whereas, the glaciers with <30% voids were filled with nature neighbor interpolation algorithm), temporal uncertainty of TanDEM-X and uncertainty of radar signal penetration respectively.
was calculated using the widely accepted approach,considering the spatial autocorrelation as:
where σΔh is the off-glacier NMAD, A is the glacier area analyzed and Acor = πd2, with d being the decorrelation length, we used a d=950 m observed for Jammu and Kashmir Himalaya, ecompassing the present study area (Abdullah et al. 2020). For the estimation voidfill uncertainty we again followed (Abdullah et al. 2020) where a set of glaciers (void free) distributed across the study region were selected and >30% voids were artificially created. The artificially created voids were then filled using different interpolation algorithms and the results where compared with the original values (glacier without voids). The voids filled with using natural neighbor algorithm were found in good agreement with the original values with just ±0.07 difference between the original and interpolated values. The difference of ±0.07 introduced due to the void filling was therefore considered as uncertainty due to voidfill (Abdullah et al. 2020). The penetration bias was calculated using the following exponential function determined for the neighboring Lahaul-Sipti region by Vijay and Braun, 2016:
where ‘x’ is absolute surface elevation and ‘y’ is the relative penetration bias between SRTM X and C band. The uncertainty of each of the individual parameter is described in detail in the revised manuscript.
Comment#4: Line 342: error assessment in off-glacier areas is good to do, but this is not reported or shown in the dh/dt figures. If it reveals systematic biases then these should be corrected to zero. This requires assessment at as many off-glacier locations as possible, at a range of altitudes, as the DEM biases are often not uniform across a scene. The initial and corrected off-glacier stats should be reported and shown in the figures.
Response: The elevation change estimates are based on Abdullah et al. 2020 where the error estimation is described in detail including the off glacier biases in various 5° slope bins. The study reported mean off-glacier elevation difference of 0.06 m a-1.
Comment#5: Table 2: KUGI glacier volumes - the calculation of volume is not trivial and is not explained. Where do these come from?
Response: The glacier volume was estimated using the slope-dependent volume estimation approach (Haeberli and Hoelzle, 1995). The methodology to estimate volume has been incorporated in the revised manuscript.
Comment#6: Figures 2 and 3: please show a background map, scale bar and the apparent dh/dt values off-glacier.
Response: As suggested, the figures have been modified in the revised manuscript.
Citation: https://doi.org/10.5194/essd-2021-28-AC2
-
AC2: 'Reply on RC2', Romshoo Shakil Ahmad, 08 Jun 2021
-
CC1: 'Comment on essd-2021-28', Glacier Ice, 01 May 2021
The paper by Romshoo et al. is interesting research on evaluating the existing glacier inventories over the north-western Himalayan region of Jammu and Kashmir. However, there are certain loopholes, some of them very gross, which have been pointed out below:
The title mentions "global glacier inventories", however, the very first sentence of the Abstract section mentions ICIMOD (for Himalaya only) and GAMDAM (for Asia only) inventories which are regional. As such I would suggest the authors modify the text wherever required. Or maybe mention "Evaluation of the existing glacier inventories..."
What do the authors mean by limited field surveys? How many glaciers were actually field surveyed? Specifically, what type of data was collected from the field, and would it qualify as a representative sample for Quality Control?
The uncertainty of mapping is more in Jhelum (~13%) as compared to Suru and Chenab (~3.3%). Since there are a lot of debris-covered glaciers in Chenab and Suru, uncertainty should be more in these basins as compared to Jhelum (where once predominantly finds clean-ice glaciers with few exceptions). Please explain.
When the authors say "most of the glaciers in the study area are <1 km2 in size><1km2", they should mention the number and percentage of these glaciers. Also for the 1-5 km2 category.
"Majority of the glaciers....". Please quantify. Similarly "... Jhelum where the glaciers are mostly...". Again MOSTLY is subjective.
Rather than saying "mainly harbor slopes ranging from 10-30°", it would be better to mention the average slopes of the glaciers in all three basins.
What is RBA? Mention at first occurrence. If KUGI is "consistent" with RGI and GAMDAM, I wonder then as to what is the purpose of KUGI?
I would suggest removing the portion about geodetic mass changes since the authors have already published geodetic mass changes in Scientific Reports very recently (See the reference below). This would mean plagiarism/redundancy.
Abdullah, T., Romshoo, S. A., & Rashid, I. (2020). The satellite observed glacier mass changes over the Upper Indus Basin during 2000–2012. Scientific Reports, 10(1), 1-9.
"Evaluation of the glacier inventories and assessment of glacier elevation change in the data-scarce Himalaya, reported in this article, would constitute a reliable database for research particularly in hydrology, glaciology, and climate change". This is not convincing. How will this effort help, given the fact that authors mention that KUGI is "consistent" with RGI and GAMDAM? How is the KUGI more reliable than GAMDAM and RGI?
Line 70: Azam et al., 212 should be Azam et al. 2012
Line 73: "Indicated" should be "indicates".
I do not find the introduction section very convincing. Especially Line 60-105 appears more of discontinuous sentences where authors jump between various glaciological assessments (area changes, frontal retreat, geodetic and glaciological mass changes). This becomes irrelevant since the focus of the paper is the "evaluation" of glacier inventories over 3 river basins. Besides, I find certain sentences over-referenced and others poorly referenced. In many places, the authors have not even bothered to cite the recent literature (See details after the comment. Although I do not know whether they will be relevant if the MS is revised and contextualized for comparison of glacier inventories). For example, the authors say that using the freely available glacier datasets for glacier change assessment and future projections is not recommended as the glacier inventories have inconsistencies in terms of different glacier variables. Does it mean the regional glacier-related assessments (cited by authors in first paragraphs) are not imprecise and not reliable? The authors further go on to say that "the glacier inventory database by Shukla et al., (2020), restricted to the Suru basin, is primarily based on the automatic approach (normalized-difference snow index) unlike the present study where the glaciers are mapped manually using on-screen digitization." Do they mean the inventory by Shukla et al is not credible? Since on-screen digitization is highly subjective and dependent on the cognition/skill of analysts, the approach could be contested especially when it comes to inventory mapping over large areas. How can/have the uncertainties about cognition been addressed by authors?
Merely saying Google Earth was used for validation will not have many takers among the remote sensing glaciology community. I tried to dig into Google Earth data of the 2000s for the three basins but found massively snow/cloud-covered data for the assessment period. The authors need to come up clean on this and say precisely where Google Earth data was used for correcting the glacier outlines. And also since Google Earth and Landsat data do not have an exact overlap, how was coregistration achieved.
RECENT LITERATURE:
Nie, Y., Pritchard, H. D., Liu, Q., Hennig, T., Wang, W., Wang, X., ... & Chen, X. (2021). Glacial change and hydrological implications in the Himalaya and Karakoram. Nature Reviews Earth & Environment, 1-16.
Farinotti, D., Immerzeel, W. W., de Kok, R. J., Quincey, D. J., & Dehecq, A. (2020). Manifestations and mechanisms of the Karakoram glacier Anomaly. Nature geoscience, 13(1), 8-16.
Shean, D. E., Bhushan, S., Montesano, P., Rounce, D. R., Arendt, A., & Osmanoglu, B. (2020). A systematic, regional assessment of high mountain Asia glacier mass balance. Frontiers in Earth Science, 7, 363.
Soheb, M., Ramanathan, A., Angchuk, T., Mandal, A., Kumar, N., & Lotus, S. (2020). Mass-balance observation, reconstruction and sensitivity of Stok glacier, Ladakh region, India, between 1978 and 2019. Journal of Glaciology, 66(258), 627-642.
Mehta, M., Kumar, V., Garg, S., & Shukla, A. (2021). Little Ice Age glacier extent and temporal changes in annual mass balance (2016–2019) of Pensilungpa Glacier, Zanskar Himalaya. Regional Environmental Change, 21(2), 1-18.
Farinotti, D., Immerzeel, W. W., de Kok, R. J., Quincey, D. J., & Dehecq, A. (2020). Manifestations and mechanisms of the Karakoram glacier Anomaly. Nature geoscience, 13(1), 8-16.
Nie, Y., Pritchard, H. D., Liu, Q., Hennig, T., Wang, W., Wang, X., ... & Chen, X. (2021). Glacial change and hydrological implications in the Himalaya and Karakoram. Nature Reviews Earth & Environment, 1-16.
Pritchard, H. D. (2019). Asia’s shrinking glaciers protect large populations from drought stress. Nature, 569(7758), 649-654.
Immerzeel, W. W., Lutz, A. F., Andrade, M., Bahl, A., Biemans, H., Bolch, T., ... & Baillie, J. E. M. (2020). Importance and vulnerability of the world’s water towers. Nature, 577(7790), 364-369.
Hugonnet, R., McNabb, R., Berthier, E. et al. Accelerated global glacier mass loss in the early twenty-first century. Nature 592, 726–731 (2021). https://doi.org/10.1038/s41586-021-03436-z
The authors suggest having used satellite data of 2000±3 years to delineate inventories whereas ICIMOD glacier inventory has used satellite data of 2005±5 years. Wouldn't it be comparing apples with oranges? This becomes important especially in the case of small glaciers and needs to be factored for.
"It is hoped that the KU glacier inventory and elevation change databases presented in this paper shall further help in promoting research in fields like climate change, hydrology, and other allied fields." This is common for any inventory. The rationale should be how KUGI will help to further it. This needs to be mentioned.
Line 132: Need to place "and" between latitude and longitude values.
Line 135-36: "The area above 3600 m asl in general remains covered with perennial snow and glaciers". Not true mostly. This has to be ~4000 m asl for the J&K region.
Line 136-37: Why have authors quoted numbers from RGI inventory and not GAMDAM?
Line 138: "thus making the study area the most glaciated terrain". Reframe.
Line 139-40: The authors quote Kamp et al. (2011) and suggest that glaciers are cirque-type in Ladakh which is not true with all the glaciers in the Suru basin and also neighboring Zanskar region. Glaciers in North Ladakh (Siachin area, Rimo group) arent cirque-type either. Please reframe.
Line 140-41: "All the major tributaries" instead of "Most of the major tributaries"
Fig. 1: The caption should mention the following: What does the inset map represent? What is the background image (a DEM or what)? Have a legend for elevation if it is so. Mention may be "GA" for glacier area instead of "A" since "A" also represents Jhelum Basin. The text could be made bold and a little larger for histograms in the study area map. What is the source of number of glaciers, glacier area, debris cover, and glacier volume? Maybe plot glacier volume and debris cover on the secondary axis since the associated values are small.
Line 150-180: Could be better represented as a table. For climatology of Jhelum Basin use: Zaz et al 2019 (ACP). For Suru met data are improper. Authors used Schmidt and Nüsser, 2012 (which mentions a different area of Ladakh, Kang Yatze massif, and not Suru) while Chevuturi et al. (2018) report climate of Leh and not Suru. Need to correct it. Similarly for Chenab, the authors quote Azam et al. (for Chotta Shigri area). Why not cite Bhutiyani et al. 2007 (Climatic change) and Bhutiuyani et al 2010 (Int J of Climatology)?
Line 192: "The of moderate resolution". Please correct.
Line 193-96: Some of the references have been quoted above and do not necessarily need to be mentioned here.
Table 1:
The authors mentioned wrong dates for imagery used for ICIMOD inventory. It is 2005±3 years (Weblink: https://lib.icimod.org/record/9419. See page 7>Section Satellite images> the second paragraph). I would again repeat my above question: Can 2000 data (in RGI, GAMDAM, and KUGI) be compared with 2005 data (as used in ICIMOD inventory)? Definitely not. Please justify. Also modify the respective entry in the Table. Details of all satellite scenes should be provided as supplementary data.
Google Earth: Mention the date of Google Earth images, if at all they were used for correcting/validating glacier outlines. Maybe have a supplementary file for mentioning which glaciers were validated using Google Earth imagery.
Line 213-14: Delete " hereafter named Kashmir University Glacial Inventory" as it has already been defined as KUGI in the abstract.
Line 214: Delete "global".
Line 129: "acquired during 2002 to 2008". See my earlier comments.
Line 225-28: The authors mention "The RGI glacier outlines have been extracted semi-automatically from the Landsat satellite images between 1998 and 2009. However, most of the glaciers (~98%) in the inventory over the study area have been extracted from the images acquired during 1998-2002". How many glaciers were delineated from 1998-2002 data in RGI inventory? See attribute of RGI shapefile. Again comparison seems a problem here; not only due to dates but also the technique used.
Line 244: What do authors mean by "limited field surveys"? How many glaciers (%) were ground surveyed? What kind of data was collected? Need to reflect all that in the MS.
Line 256-57, 262, 267-270: "were verified from the Google EarthTM". The GE data for 2000 is almost not usable for the region. Please explain. Is it a deliberate attempt of misinformation or what?
Line 262-63: "The thin debris layer on the glacier surface, often bearing lower surface temperature". Do the authors mean differential wrt ice or neighboring landscape?
Line 200, 280-282: The authors mention ASTER DEM here having been used to derive glacier-specific topographic parameters. But there is no mention of ASTER DEM in Table 1. Why was ASTER DEM used when CARTO DEM with a similar resolution is available over the region? Seems repetition of (line 201-202) here (280-82).
Line 285-295: When the techniques used for mapping the glacier outlines are different, it is but obvious that there won't be a high overlap. Would it be so? Add a justification.
Remove section 4.3 as explained in the beginning and also uncertainty related to geodetic mass balance.
Line 330: Rp/Ap is a constant. What does it represent?
Line 368: "Jhelum" instead of "Jehlum". Be consistent with spelling.
Line 368: "The glaciers range in size from 0.01 km2". This means ~11 pixels. I wonder if such small-sized glaciers (1 ha as mentioned by authors) could be mapped from 30 m Landsat data? This would be highly uncertain. Could it be that some of them were snowpacks and not glaciers especially when ascertaining from 2000-02 data?
Table 2: The authors mention glacier volume but have not provided any information as to how ice volumes were derived? Did the authors use VA scaling and why if it is known that VA scaling estimates are highly uncertain, even for entire mountain ranges.
Line 383-84: "mean glacier slope in the basin varies between 9° and 50°". The glaciers with a mean slope of 50 degrees are highly unlikely since by the definition such areas (> 30-degree slope) could be avalanche feeding zones. Please speak otherwise it raises concerns about the inventory itself.
Table 3: Why do the authors need to mention glacier area categories from >20->50 in the Jhelum basin. >50 should not be the category. Let it be >50-the highest glacier area in the respective basin. Similarly, if there is a category 1-2, it should be followed with >2-5, >5-10, so on and so forth.
Line 395-397: Delete as it is already mentioned in 275-76.
Table 4: Elevation categories from 5500-7000 are not relevant for Jhelun and as such could be deleted. If the first category starts from <=4000, the next category should be >4000-4500, so on and so forth.
Mention uncertainties about each A (glacier areas) in table 3, 4, 5 and 6.
The uncertainty in TGA for the Jhelum basin is ~13.3% compared to Suru and Chenab (both 3.3%). Why is this so? This should have been the other way around since there are more clean-ice glaciers in Jhelum. Explain.
How different are the estimates of Table 7 different from Scherler et al 2018 (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018GL080158)
Line 474: Delete "regional and global".
Table 8: Should be "ΔN" instead of "N". Percentage difference could be mentioned in brackets.
Line 490-500: Use ICIMOD and GAMDAM rather than ICIMODGI and GAMDAMGI.
Table 9: Above in methods authors mention having used median glacier elevation rather than mean glacier elevation.
Section 5.3 and associated tables/figures should be removed.
Line 675-680: The authors mention large glaciers at high altitudes and then low glacier cover at high altitudes which is a kind of contradiction. Large glaciers as authors suggest in results contribute to glacier cover. Please explain.
Table 14: Remove
The authors do not discuss much about the aspect (neither in results nor discussion) except very qualitatively.
Section 6.2 should be "Inconsistencies in existing glacier inventories"
Figure 5: Mention background image and band combination.
Line 720-724: The overlap ratio could be misleading since the inventories were computed using different methods. Please justify.
Figure 6: SG. On checking from Google Earth, the existing inventories have rightly followed GLIMS definition of glacier delineation and divided the ice into two polygons since the ice fluxes move in different ablation zones. However, the authors have erred here (and maybe in many such cases) by considering it as a single glacier. The authors should know that although the ice masses are connected (in the accumulation zone) the movement of ice in different directions owing to the ridge-topography divides the ice into two polygons and hence two glaciers (refer to GLIMS definition of glaciers). This appears to be a huge flaw with the interpretation by authors as the number of glaciers is massively underestimated in KUGI. This needs to be corrected in the data as well as explanations to the data.
Figure 7: Mention background image and band combination.
Remove section 6.3
Conclusions:
While the authors mention "limited field survey" at many places in the text, they have failed to showcase the data collected, the photographs depicting the glacier environments in these three catchments. They should show field data and photographs from all three catchments and demonstrate its usefulness in inventorying the glaciers in these three respective areas.
Delete sentences about geodetic mass changes.
The authors fail to convince the robustness of KUGI compared to at least RGI and GAMDAM. This needs to come up very in the results, discussion as well as the conclusion.
Citation: https://doi.org/10.5194/essd-2021-28-CC1 -
AC3: 'Reply on CC1', Romshoo Shakil Ahmad, 08 Jun 2021
CC#1
General Comment: The paper by Romshoo et al. is interesting research on evaluating the existing glacier inventories over the north-western Himalayan region of Jammu and Kashmir. However, there are certain loopholes, some of them very gross, which have been pointed out below:
Response: The authors thank the anonymous reviewer/commentator “Glacier Ice” for the useful comments and suggestion on the manuscript. The comments/suggestions have been responded point-by-point in the following sections and the useful suggestions have been incorporated in the revised manuscript which have improved the quality of the revised manuscript.
Comment #1: The title mentions "global glacier inventories", however, the very first sentence of the Abstract section mentions ICIMOD (for Himalaya only) and GAMDAM (for Asia only) inventories which are regional. As such I would suggest the authors modify the text wherever required. Or maybe mention "Evaluation of the existing glacier inventories..."
Response: Agreed, that the ICIMOD and GAMDAM are regional inventories and the RGI is a global glacier inventory. Accordingly, the title and text have been modified in the revised manuscript. The “Global glacier inventories” has been replaced with “Global and Regional glacier inventories” in the title of the revised manuscript.
Comment #2: What do the authors mean by limited field surveys? How many glaciers were actually field surveyed? Specifically, what type of data was collected from the field, and would it qualify as a representative sample for Quality Control?
Response: In the present study, we have done field surveys/validation on 20 glaciers located across the study area. The locations and the field photographs of these glaciers has been provided as Supplementary figure in the revised manuscript. We have collected the snout positional data of the debris-covered tongues of these glaciers to validate the glacier delineation. The field measurements of these glaciers acted as an interpretation tool for delineation of the debris-covered glaciers in the study area. The terminus of the heavily debris covered glaciers like the Hoksar glacier in Jhelum basin was not easily delineated even using the thermal and google earth imagery. We therefore, mapped the glacier terminus on field and further observed that the debris on the glacier is relatively smooth and aligned in the direction of glacier flow when compared to the debris-cover in the surroundings which was found a useful field-based information for mapping of debris-covered termini of other glaciers when viewed on Google earth. Further, eight of these twelve glaciers have been designated as benchmark glaciers and are continuously studied for mass balance, GPR, debris thickness, Surface mapping temperature profiling and other glaciological studies since the last 5-8 years. In addition, glacier outlines of several other glaciers in the vicinity of these 8 benchmark glaciers have been verified during annual glacier field expeditions during the last 5-8 years.
Additionally, all the heavily debris covered glaciers and a majority of the clean glaciers, numbering more than 850 were qualitatively verified on Google Earth image for the rectification of any delineation error. A similar approach of quality check using Google Earth has been previously adopted by Nagai et al. (2016) and several others and is an accepted method for validation of such a large number of glaciers located in inaccessible complex terrain.
Field photographs of the about 20 selected glaciers in the three basins, which have been visited over the last few years for field measurements/validation along with their GLIMS ID are presented in the revised manuscript (The field photographs of some of these glaciers are provided above). A locational map of these glaciers in the form of a KML file would be provided as a supplemental material in the revised manuscript (The field photographs have also been attached as a supplement file in pdf format).
Comment #3: The uncertainty of mapping is more in Jhelum (~13%) as compared to Suru and Chenab (~3.3%). Since there are a lot of debris-covered glaciers in Chenab and Suru, uncertainty should be more in these basins as compared to Jhelum (where once predominantly finds clean-ice glaciers with few exceptions). Please explain.
Response: For the uncertainty assessment of the glacier area, we have referred to the previous detailed uncertainty analysis by Paul et al. (2013) that reported an error of ~3% for the alpine glaciers. Paul et al. (2013), however, reported a perimeter-area ratio of 5.03 km−1, unlike the perimeter-area ratio of 0.96 km-1 observed for Chenab in this study, 3.9 km-1 for Suru and 6.22 km-1 for the Jhelum basin. We therefore applied a scaling after Braun et al, (2019) to determine the uncertainty in glacier outline delineation in the three basins in this study. As indicated by the value of Perimeter-Area ratio of 6.22 km-1 for the Jhelum basin, it is quite obvious that the basin is expected to have higher uncertainty as the perimeter-area ratio of the basin is higher amongst the three studied basins. This has been specified in the revised manuscript.
Comment #4: When the authors say "most of the glaciers in the study area are <1 km2 in size><1km2", they should mention the number and percentage of these glaciers. Also for the 1-5 km2 category.
Response: The detailed statistics of the glacier number and area in each size category is described in detail in the Results section and already provided in Table 3 of the manuscript. On an average around 91% of the glaciers (glacier number) are in the size category < 1 km2 in the Jhelum basin. Similarly, around 77% of the glaciers (number) are falling in the < 1 km2 area class in the Suru and Chenab basins (Table 3). In terms of the glacier coverage (area), the area class 1-5 km2 harbor ~44% of the glacier area in the Chenab basin, followed by ~55% in the Suru basin and ~90% in the Jhelum basin (Table 3). The statistics are averaged for all the glacier inventories and the numbers vary between the individual inventories (Table 3). This has been specified in the revised manuscript.
Comment #5: "Majority of the glaciers....". Please quantify. Similarly "... Jhelum where the glaciers are mostly...". Again MOSTLY is subjective.
Response: The elevation range 4500-5500 for the Chenab and Suru basins harbor around 85.5% and 91.3% of the glacier number and area respectively. Whereas, the elevation band 4000-5000 harbors 91.8% (number) and 95.8% (area) of glacier number and area respectively in the Jhelum basin (Information already provided in Table 4). However, the information has been specified in the revised manuscript as suggested.
Comment #6: Rather than saying "mainly harbor slopes ranging from 10-30°", it would be better to mention the average slopes of the glaciers in all three basins.
Response: The slope range (10-30°) harbors 78% of the glaciers in terms of number and 84% of the glacier area (Table 4). The numbers have been specified in the revised manuscript.
Comment #7: What is RBA? Mention at first occurrence. If KUGI is "consistent" with RGI and GAMDAM, I wonder then as to what is the purpose of KUGI?
Response: The overlap ratio of individual glaciers is represented by “rov” whereas, the “R” (RBA) has been used to represent the average overlapping ratio of the base and target glacier inventories (overlap ratio averaged for all the glaciers in a particular inventory combination e.g, KUGI-RGI for each basin). This has been described under the methodology section in the revised manuscript. The inventories are consistent in general pattern and distribution which means that a majority of the glaciers in all the inventories are found in a particular area class, slope category etc. however, large differences occur in individual glaciers and overall statistics in terms of the area/size (Table 3-6). However, the specific purpose of the KUGI is provided is provided in the response to the Comment #9 below.
Comment #8: I would suggest removing the portion about geodetic mass changes since the authors have already published geodetic mass changes in Scientific Reports very recently (See the reference below). This would mean plagiarism/redundancy.
Abdullah, T., Romshoo, S. A., & Rashid, I. (2020). The satellite observed glacier mass changes over the Upper Indus Basin during 2000–2012. Scientific Reports, 10(1), 1-9.
Response: The geodetic mass changes of the entire region of Upper Indus Basin comprising of 12000 glaciers, of which Jhelum, Chenab and Suru are a part, has been published in the Scientific Reports which has been referred to in the manuscript and therefore would not in any way amount to plagiarism. However, we wanted to retain the elevation change information in the inventory data base for the ready availability of this valuable database to scientific community as the same provides a valuable additional information about the behavior/dynamics of the glaciers in the study area.
Comment #9: "Evaluation of the glacier inventories and assessment of glacier elevation change in the data-scarce Himalaya, reported in this article, would constitute a reliable database for research particularly in hydrology, glaciology, and climate change". This is not convincing. How will this effort help, given the fact that authors mention that KUGI is "consistent" with RGI and GAMDAM? How is the KUGI more reliable than GAMDAM and RGI?
Response: Please see the response to the comment above where it has been clarified that there is “consistency” in terms of the general pattern/distribution, however there are significant differences in various glacier parameters reported in the three inventories compared to the KUGI. The value addition and novelty of the KUGI over the three inventories evaluated in this study has been prominently mentioned at relevant places in the revised manuscript and the same is consolidated and reproduced here for your perusal is also reproduced as follows:
“Primarily, the motivation for the KUGI is to develop a high-resolution glacier inventory with improved accuracy with visual interpretation and manual delineation of glaciers from Landsat satellite data supported by the limited ground truth and supplementing the glacier outlines with additional data like debris-cover, thickness changes and other glaciological parameters, that are either missing or incorrect in the existing databases so that the database is made available to the large research community for various applications.
Purportedly the global and regional glacier databases that were chosen for comparison in this study have been generated using a semi-automated method (manual for GAMDAMGI) allowing less human error, quick delivery, and high accuracy. However, it was found in this study that there are significant errors in the evaluated databases due to the misinterpretation of seasonal snow cover particularly on the glacier headwalls at high altitudes, shadow-covered glaciers and debris-cover. Against the reports/claims of the overall accuracy of the global/regional glacier databases, ~3% for ICIMOD (Bajracharya er al, 2014), ~5% for RGI (Pfeffer et al. 2014; RGI Consortium, 2017) and 15% for GAMDAM (Nuimura et al. 2015), it was found that, compared to the KUGI, the ICIMOD is overestimating glacier area by 12.2%, RGI underestimates the glacier area by 2.4% and the GAMDAM inventory by 1.5%. However, the three databases overestimate the glacier numbers in the three basins considerably; RGI by ~45%, ICMOD by ~68% and GAMDAM by ~56%. Gardelle et al. (2013) found that in the southeastern Tibet, RGI 2.0 database has glacier extent 88% greater than their estimate. Similarly, Nuimura et al. (2015) while comparing GAMDAM with ICIMOD found a significant discrepancy between the two inventories. Frey et al. (2014) and Mölg et al. (2018), have highlighted the presence of debris-cover, seasonal snow and cloud cover as the main source of uncertainty in the Himalayan region. Mohammad et al. (2019) has also highlighted the differences between the existing glacier inventories in Indus basin.
Keeping in view the worldwide use and applications of global and regional databases, it is important that a rigorous evaluation of these global and regional inventories is undertaken for the continued refinement of the methodology which is a fundamental requirement for any meaningful application of the global or regional database. It is hoped that the future releases of the databases will improve these and other shortcomings identified in this manuscript. Although this cross-checking improved the quality of the data, the mapped glacier outlines are also affected by various other types of obscurities, which are mostly dependent on image resolution which is the also the case with KUGI.
KUGI improved the mis-mapped glacier outlines/boundaries from existing global and regional inventories and any mismatches of the glacier geometry due to the seasonal/temporal snow cover and shadows were manually corrected using additional Landsat images and Google earth images. Further the mapped glaciers with better georeferencing were overlaid with high resolution images in Google Earth environment for validation wherever available. Though, the mis-mapped/mis-located outlines, observed on the global/regional inventories, may have only limited effect on measurements of glacier area, but can introduce serious errors into applications that rely on absolute positioning (e.g. co-registration to other datasets such as DEMs). The only realistic way to correct them is to provide more accurate outlines as done in the KUGI which would serve as source of improved outlines for the scientific community interested in conducting various application studies using the glacier outlines.
The analysis of the debris-cover (>19%)-the criterion we used to classify the debris-covered glaciers (Brun et al., 2019) showed that the RGI, ICIMOD and GAMDAM glacier inventories have underestimated the debris-covered glaciers by ~15%, ~25%, 8% respectively. Debris cover, present on 44% of Earth’s glaciers, significantly influences glacier melt. Despite its significant importance, the debris cover has not been mapped with accuracy in the three global/regional glacier databases evaluated in this study. Due to lack of debris-cover map at the global level (Sam Herreid & Francesca Pellicciotti, 2020), debris cover has been omitted from global glacier models and forecasts of their response to a changing climate. Therefore, the KUGI, has fundamentally resolved this omission and provided improved debris-cover outlines of the three basins in the Northwest Himalaya. KUGI has added a separate debris-cover database in the three basins which is missing in the global/regional databases. This is a major improvement and correction to the existing global/regional databases evaluated in this study. Use of the KUGI outlines of the debris-covered parts of glaciers in glacier-melt models will enable improved estimates of melt over the three basins.
Other than the debris-covered glaciers, the discrepancy related to the shadowed glaciers is another major error with the glacier inventories. Though, KUGI has not generated a database of these glaciers, but the same is under preparation for inclusion in the revised version of the KUGI database being submitted after this revision”
Comment #10: Line 70: Azam et al., 212 should be Azam et al. 2012
Response: The typo has been corrected in the revised manuscript.
Comment #11: Line 73: "Indicated" should be "indicates".
Response: The word “indicated” has been replaced by “indicates” in the revised manuscript.
Comment #12: I do not find the introduction section very convincing. Especially Line 60-105 appears more of discontinuous sentences where authors jump between various glaciological assessments (area changes, frontal retreat, geodetic and glaciological mass changes). This becomes irrelevant since the focus of the paper is the "evaluation" of glacier inventories over 3 river basins. Besides, I find certain sentences over-referenced and others poorly referenced. In many places, the authors have not even bothered to cite the recent literature (See details after the comment. Although I do not know whether they will be relevant if the MS is revised and contextualized for comparison of glacier inventories). For example, the authors say that using the freely available glacier datasets for glacier change assessment and future projections is not recommended as the glacier inventories have inconsistencies in terms of different glacier variables. Does it mean the regional glacier-related assessments (cited by authors in first paragraphs) are not imprecise and not reliable? The authors further go on to say that "the glacier inventory database by Shukla et al., (2020), restricted to the Suru basin, is primarily based on the automatic approach (normalized-difference snow index) unlike the present study where the glaciers are mapped manually using on-screen digitization." Do they mean the inventory by Shukla et al is not credible? Since on-screen digitization is highly subjective and dependent on the cognition/skill of analysts, the approach could be contested especially when it comes to inventory mapping over large areas. How can/have the uncertainties about cognition been addressed by authors?
Response: The introduction section provides a broader overview of the research under consideration and therefore, the authors felt that the reference to geodetic mass balance and other glacial studies is relevant. However, as suggested, a few relevant references suggested by the anonymous commentator have been incorporated in the revised manuscript.
We never said that the existing inventories are unreliable. Instead, we reiterate that a thorough evaluation of the glacier boundaries, as reported in this paper, is required before using them for impact assessment or any other climatic and hydrological application especially when the spatial domain of investigation is small (basins or sub-basins). The discrepancy in the glacier area might result in significant uncertainty in the use of the data for various applications.
The commentator is again insinuating when he emphasizes in italics something that is not meant by the authors-Not our words. We did not discredit the glacier inventory by Shukla et al, in fact, a part of the study area in Suru overlaps with the glaciers studied by Shukla et al. and therefore it is necessary to refer the paper. However, we observed a difference in glacier coverage reported between for a few glaciers in the two inventories and it is therefore important to explain this difference which we have attributed to the different techniques used in the present study.
Regarding minimizing the error due to the onscreen digitization by the analyst, it was made sure that analysts (authors) use the same criteria in terms of pre-processing, mapping scale etc. for the delineation of glacier boundaries. Furthermore, all the glacier boundaries were checked for quality control and corrected by the Lead author before finalization. This minimized the uncertainty due to the skill/interpretation of the analyst. The uncertainty approach used in the present study is also based on detailed analysis of glacier uncertainties mapped by multiple analysts (Paul et 2013) and therefore we believe that this approach is quite valid and addresses the uncertainties well.
Comment #13: Merely saying Google Earth was used for validation will not have many takers among the remote sensing glaciology community. I tried to dig into Google Earth data of the 2000s for the three basins but found massively snow/cloud-covered data for the assessment period. The authors need to come up clean on this and say precisely where Google Earth data was used for correcting the glacier outlines. And also since Google Earth and Landsat data do not have an exact overlap, how was coregistration achieved.
Response: This is very unfortunate that the anonymous commentator is again using very strong words like “need to come clean” in his review comments which sound like personal and unprofessional. We do agree that the Google Earth data for 2000s for the study area is cloud and snow covered. However, it is pertinent mention here that the Google Earth image of post-dating 2000s were used for quality check and verification of ambiguous glacier outlines. For example, glacier headwalls of many glaciers covered with thin layer of snow appeared smooth on the Landsat imagery, however, when checked on the high-resolution Google EarthTM imagery, it usually turned out an undulated non glaciated surface or a rock surface covered with thin snow. Besides all the heavily debris covered glaciers were qualitatively verified on Google Earth for rectification of any delineation error. A similar approach of quality check using Google Earth has been previously adopted by Nagai et al. (2016) and several others and is an accepted method for such a large number of glaciers located in inaccessible complex terrain. Further, we did not use the Google Earth data for the quantification of glacier area, and therefore an exact-overlap was not required. All the statistic quantifications and analyses in the present study is based on the Landsat images. This has been specified in the revised manuscript. The Google Earth images were used for the verification and validation of the glacier outlines only. The location of the 850 glaciers verified from the snow- and cloud-free Google Earth images would be provided as supplemental KML file in the revised manuscript.
RECENT LITERATURE:
Nie, Y., Pritchard, H. D., Liu, Q., Hennig, T., Wang, W., Wang, X., ... & Chen, X. (2021). Glacial change and hydrological implications in the Himalaya and Karakoram. Nature Reviews Earth & Environment, 1-16.
Farinotti, D., Immerzeel, W. W., de Kok, R. J., Quincey, D. J., & Dehecq, A. (2020). Manifestations and mechanisms of the Karakoram glacier Anomaly. Nature geoscience, 13(1), 8-16.
Shean, D. E., Bhushan, S., Montesano, P., Rounce, D. R., Arendt, A., & Osmanoglu, B. (2020). A systematic, regional assessment of high mountain Asia glacier mass balance. Frontiers in Earth Science, 7, 363.
Soheb, M., Ramanathan, A., Angchuk, T., Mandal, A., Kumar, N., & Lotus, S. (2020). Mass-balance observation, reconstruction and sensitivity of Stok glacier, Ladakh region, India, between 1978 and 2019. Journal of Glaciology, 66(258), 627-642.
Mehta, M., Kumar, V., Garg, S., & Shukla, A. (2021). Little Ice Age glacier extent and temporal changes in annual mass balance (2016–2019) of Pensilungpa Glacier, Zanskar Himalaya. Regional Environmental Change, 21(2), 1-18.
Farinotti, D., Immerzeel, W. W., de Kok, R. J., Quincey, D. J., & Dehecq, A. (2020). Manifestations and mechanisms of the Karakoram glacier Anomaly. Nature geoscience, 13(1), 8-16.
Nie, Y., Pritchard, H. D., Liu, Q., Hennig, T., Wang, W., Wang, X., ... & Chen, X. (2021). Glacial change and hydrological implications in the Himalaya and Karakoram. Nature Reviews Earth & Environment, 1-16.
Pritchard, H. D. (2019). Asia’s shrinking glaciers protect large populations from drought stress. Nature, 569(7758), 649-654.
Immerzeel, W. W., Lutz, A. F., Andrade, M., Bahl, A., Biemans, H., Bolch, T., ... & Baillie, J. E. M. (2020). Importance and vulnerability of the world’s water towers. Nature, 577(7790), 364-369.
Hugonnet, R., McNabb, R., Berthier, E. et al. Accelerated global glacier mass loss in the early twenty-first century. Nature 592, 726–731 (2021). https://doi.org/10.1038/s41586-021-03436-z
Response: The relevant literature has been cited at appropriate places in the revised manuscript.
Comment #14: The authors suggest having used satellite data of 2000±3 years to delineate inventories whereas ICIMOD glacier inventory has used satellite data of 2005±5 years. Wouldn't it be comparing apples with oranges? This becomes important especially in the case of small glaciers and needs to be factored for.
Response: The difference in the dates of source images does not explain the significant glacier area over-estimation observed in the Jhelum and Chenab basins. Using 2005±3 data (compared to 2000±2 in case of KUGI), normally an under estimation in glacier area is expected in ICIMOD but the comparison of the ICIMOD database with KUGI shows over-estimation of the glacier area which has been attributed to inclusion snow covered glacier headwalls and at places some season snowpacks also. Further, the area underestimation of 10.97% in case of the Suru basin is not fully explained by the expected area change between 2000±2 and 2005±3. Such comparisons using data gap of 2-3 years have also been carried out in previous studies (Nuimura et al. 2015) and therefore, the comparison is valid. However, the data difference has been explained in the revised manuscript.
Comment #15: "It is hoped that the KU glacier inventory and elevation change databases presented in this paper shall further help in promoting research in fields like climate change, hydrology, and other allied fields." This is common for any inventory. The rationale should be how KUGI will help to further it. This needs to be mentioned.
Response: Please see the response to the similar comment above. The novelty, value addition and usefulness of the KUGI over other inventories has been discussed in the revised manuscript and is reproduced in the authors response against the Comment#9.
Comment #16: Line 132: Need to place "and" between latitude and longitude values.
Response: “and” placed between the between latitude and longitude in the revised manuscript.
Comment #17: Line 135-36: "The area above 3600 m asl in general remains covered with perennial snow and glaciers". Not true mostly. This has to be ~4000 m asl for the J&K region.
Response: In general, areas above 3600 masl remain snow covered with snow for the entire year in the study area. However, there might be some exceptions. Therefore, as suggested the sentence has been modified in the revised manuscript to show that areas ~4000 m asl remain covered with snow.
Comment #18: Line 136-37: Why have authors quoted numbers from RGI inventory and not GAMDAM?
Response: Here, we are providing an overall estimate of the glacier cover over the entire Jammu, Kashmir and Ladakh region extracted from RGI and used in the previous study (Abdullah et al. 2020).
Comment #19: Line 138: "thus making the study area the most glaciated terrain". Reframe.
Response: The sentence has been rephrased as suggested in the revised manuscript.
Comment #20: Line 139-40: The authors quote Kamp et al. (2011) and suggest that glaciers are cirque-type in Ladakh which is not true with all the glaciers in the Suru basin and also neighboring Zanskar region. Glaciers in North Ladakh (Siachin area, Rimo group) aren’t cirque-type either. Please reframe.
Response: Agreed, there are a few exceptions. Therefore, the sentence has been modified in the revised manuscript.
Comment #21: Line 140-41: "All the major tributaries" instead of "Most of the major tributaries"
Response: The sentence has been modified as suggested in the revised manuscript.
Comment #22: Fig. 1: The caption should mention the following: What does the inset map represent? What is the background image (a DEM or what)? Have a legend for elevation if it is so. Mention may be "GA" for glacier area instead of "A" since "A" also represents Jhelum Basin. The text could be made bold and a little larger for histograms in the study area map. What is the source of number of glaciers, glacier area, debris cover, and glacier volume? Maybe plot glacier volume and debris cover on the secondary axis since the associated values are small.
Response: The suggested modifications in the figure have been incorporated in the revised manuscript.
Comment #23: Line 150-180: Could be better represented as a table. For climatology of Jhelum Basin use: Zaz et al 2019 (ACP). For Suru met data are improper. Authors used Schmidt and Nüsser, 2012 (which mentions a different area of Ladakh, Kang Yatze massif, and not Suru) while Chevuturi et al. (2018) report climate of Leh and not Suru. Need to correct it. Similarly for Chenab, the authors quote Azam et al. (for Chotta Shigri area). Why not cite Bhutiyani et al. 2007 (Climatic change) and Bhutiuyani et al 2010 (Int J of Climatology)?
Response: We agree that the meteorological data in Schmidt and Nüsser, (2012) is for the neighboring area, however, due to lack of meteorological data observations in the study region, we used the data from the referred paper assuming that the climatology of the Ladakh region does not vary much spatially from the Zanaskar region, as both being the part of the cold desert climatic zone. Again, due to non-availability of the meteorological data specific to Chenab basin, we cited Azam et al. which is located in the upper Chenab basin. Like, Chevuturi et al. (2018), Bhutiyani et al. (2007) also reported the climate data for the Ladakh Range based on the data available at the Leh station as there is no climate data available for the Zanaskar region. However, the time series in case of Chevuturi et al. (2018) is longer and as such, we cited the same in the manuscript. However, we have cited the relevant literature cited by the commentator in the revised manuscript.
Comment #24: Line 192: "The of moderate resolution". Please correct.
Response: The sentence have been corrected to “The use of ….” in the revised manuscript.
Comment #25: Line 193-96: Some of the references have been quoted above and do not necessarily need to be mentioned here.
Response: As suggested some of the references has been removed from here in the revised manuscript.
Comment #26: Table 1:
The authors mentioned wrong dates for imagery used for ICIMOD inventory. It is 2005±3 years (Weblink: https://lib.icimod.org/record/9419. See page 7>Section Satellite images> the second paragraph). I would again repeat my above question: Can 2000 data (in RGI, GAMDAM, and KUGI) be compared with 2005 data (as used in ICIMOD inventory)? Definitely not. Please justify. Also modify the respective entry in the Table. Details of all satellite scenes should be provided as supplementary data.
Response: Sorry for inadvertent typo and the same has been corrected in the table. The rest of the comment has already been responded to (Please see the response to Comment#14 above.)
Comment #27: Google Earth: Mention the date of Google Earth images, if at all they were used for correcting/validating glacier outlines. Maybe have a supplementary file for mentioning which glaciers were validated using Google Earth imagery.
Response: This comment has been already responded to above. All the debris-covered glacier and several clean numbering more than 850 have been validated/corrected using the Google Earth data. Google earth images are dated, 2009-2011 for the Jhelum basin: 2006-2013 for the Suru basin and 2000-2006 for Chenab basin have been used to verify the glacier outlines. The locational location of the glaciers verified on Google Earth is provided as KML file in the supplemental material of the revised manuscript.
Comment #28: Line 213-14: Delete " hereafter named Kashmir University Glacial Inventory" as it has already been defined as KUGI in the abstract.
Response: The sentence has been modified, as suggested, in the revised manuscript.
Comment #29: Line 214: Delete "global".
Response: The modifications, as suggested, have been incorporated throughout the revised manuscript.
Comment #30: Line 129: "acquired during 2002 to 2008". See my earlier comments.
Response: This has been explained in the response to Comment#14 above and specified in the revised manuscript.
Comment #31: Line 225-28: The authors mention "The RGI glacier outlines have been extracted semi-automatically from the Landsat satellite images between 1998 and 2009. However, most of the glaciers (~98%) in the inventory over the study area have been extracted from the images acquired during 1998-2002". How many glaciers were delineated from 1998-2002 data in RGI inventory? See attribute of RGI shapefile. Again comparison seems a problem here; not only due to dates but also the technique used.
Response: The information in the text is relevant for the entire Jammu, Kashmir and Ladakh region which has been specifically mentioned in the revised manuscript. Out of 15064 glaciers in RGI, 14894 glaciers were delineated from source images acquired between 1998-2002 (1998: 2228; 1999:2065; 2000:4034; 2001: 2117; 20002:3789; 2006:17; 2009:153). For the three basins under consideration, all the glaciers except one in the Suru basin, have been delineated from source images acquired between 2000-2002. This has been specified in the revised manuscript.
Comment #32: Line 244: What do authors mean by "limited field surveys"? How many glaciers (%) were ground surveyed? What kind of data was collected? Need to reflect all that in the MS.
Response: This is again a repeated comment and has been responded to above (Please see response to the Comment#2 above)
Comment #33: Line 256-57, 262, 267-270: "were verified from the Google EarthTM". The GE data for 2000 is almost not usable for the region. Please explain. Is it a deliberate attempt of misinformation or what?
Response: This is very unfortunate that the commentator is repeating the use of strong language like “Misinformation or what”. This is comment is repeated 4th time in the review. However, the comment stands already responded above. We in general appreciate the suggestions of the anonymous reviewer but the repeated use of the personal and unscientific language is unnecessary and therefore very unfortunate.
Comment #34: Line 262-63: "The thin debris layer on the glacier surface, often bearing lower surface temperature". Do the authors mean differential wrt ice or neighboring landscape?
Response: Yes, it is with reference to the neighboring landscape. The same has been mentioned in the revised manuscript.
Comment #35: Line 200, 280-282: The authors mention ASTER DEM here having been used to derive glacier-specific topographic parameters. But there is no mention of ASTER DEM in Table 1. Why was ASTER DEM used when CARTO DEM with a similar resolution is available over the region? Seems repetition of (line 201-202) here (280-82).
Response: The information pertaining GDEM2 has been incorporated in the revised manuscript. We preferred ASTER GDEM2 as its use for glacier studies is well established as mentioned in section 3.2. Furthermore, since both the DEMs have same spatial resolution and we did not find any study reporting any specific advantage of using CartoDEM for glacier inventory studies.
The repeating sentence at line 201-202 has been deleted in the revised manuscript.
Comment #36: Line 285-295: When the techniques used for mapping the glacier outlines are different, it is but obvious that there won't be a high overlap. Would it be so? Add a justification.
Response: Of course, there won’t be a high overlap and in fact we have already mentioned this (different techniques used for glacier mapping) as one of the reasons for the discrepancy observed in glacier outlines in terms of the overlap ratio
Comment #37: Remove section 4.3 as explained in the beginning and also uncertainty related to geodetic mass balance.
Response: As mentioned in response to Comment #8 above, we believe that the elevation change information will be a value addition to the database as such there is merit in retaining this section.
Comment #38: Line 330: Rp/Ap is a constant. What does it represent?
Response: Please refer to Paul et al, (2013). is the perimeter-area ration reported by Paul et al, (2013) in a detailed analysis aimed at the uncertainty assessment of glacier mapping which is equal to 5.03 km -1. In the present study the perimeter-area ratio however, varied from 0.96 – 6.22 km-1, we therefore applied a scaling after Braun et al, (2019) to determine the uncertainty in glacier delineation over the study region. We have incorporated the description in the revised manuscript.
Comment #39: Line 368: "Jhelum" instead of "Jehlum". Be consistent with spelling.
Response: The typo has been corrected in the revised manuscript.
Comment #40: Line 368: "The glaciers range in size from 0.01 km2". This means ~11 pixels. I wonder if such small-sized glaciers (1 ha as mentioned by authors) could be mapped from 30 m Landsat data? This would be highly uncertain. Could it be that some of them were snowpacks and not glaciers especially when ascertaining from 2000-02 data?
Response: Glaciers of the size (0.01 km2) have been previously mapped by Paul et al. (2002); Paul et al. (2009); Pfeffer et al. (2014); Abdullah et al (2020). Shukla et al. (2020) have also mapped glaciers with minimum size of 0.01 km2 in a recent study comprising a part of the study region using 30 m Landsat data. Also, the glaciers of the same size have been mapped in the GAMDAM inventory using Landsat 2000 data (Nuimura et al. 2015). So, mapping such glaciers from 30 m data is not a problem, however, to ensure that snow-packs are not misinterpreted for glaciers, we specifically checked satellite images dating before and after the satellite image under consideration. We also used Google earth imagery (post-dating) to verify that snow-packs are not misinterpreted as glaciers as already explained in response to your previous comment.
Comment #41: Table 2: The authors mention glacier volume but have not provided any information as to how ice volumes were derived? Did the authors use VA scaling and why if it is known that VA scaling estimates are highly uncertain, even for entire mountain ranges.
Response: The glacier volume was estimated using the slope-dependent volume estimation approach (Haeberli and Hoelzle, 1995), this has been mentioned in the manuscript.
Comment #42: Line 383-84: "mean glacier slope in the basin varies between 9° and 50°". The glaciers with a mean slope of 50 degrees are highly unlikely since by the definition such areas (> 30-degree slope) could be avalanche feeding zones. Please speak otherwise it raises concerns about the inventory itself.
Response: In the Jhelum basin, there is only one glacier with the mean slope of 49.7° and similarly there are a few more in the other basins. We rechecked these glaciers and found them situated at higher altitudes with very little exposed headwall area, and therefore, there is not enough terrain to form avalanche zone.
Comment #43: Table 3: Why do the authors need to mention glacier area categories from >20->50 in the Jhelum basin. >50 should not be the category. Let it be >50-the highest glacier area in the respective basin. Similarly, if there is a category 1-2, it should be followed with >2-5, >5-10, so on and so forth.
Response: As suggested, the non-relevant categories have been removed from the tables and the tables have been modified in the revised manuscript.
Comment #44: Line 395-397: Delete as it is already mentioned in 275-76.
Response: As suggested, the line is deleted in the revised manuscript.
Comment #45: Table 4: Elevation categories from 5500-7000 are not relevant for Jhelun and as such could be deleted. If the first category starts from <=4000, the next category should be >4000-4500, so on and so forth.
Response: The non-relevant categories have been removed from the tables in the revised manuscript.
Comment #46: Mention uncertainties about each A (glacier areas) in table 3, 4, 5 and 6.
Response: Thanks for the suggestion. The uncertainties for each area, elevation, slope and aspect class has been provided in the revised manuscript.
Comment #48: The uncertainty in TGA for the Jhelum basin is ~13.3% compared to Suru and Chenab (both 3.3%). Why is this so? This should have been the other way around since there are more clean-ice glaciers in Jhelum. Explain.
Response: This is again a repeating comment and has been already responded to (Please see the response to Comment#3 above)
Comment #49: How different are the estimates of Table 7 different from Scherler et al 2018 (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2018GL080158)
Response: On comparison of KUGI with Scherler et al. (2018), We found that the DC estimates are in general higher compared to the Scherler et al. (2018) for all the three basins. For example, for the Jhelum basin, the DC estimate of 7.2 km2 is around 8% higher than Scherler et al. (6.6 km2).
Comment #50: Line 474: Delete "regional and global".
Response: Deleted in the revised manuscript.
Comment #51: Table 8: Should be "ΔN" instead of "N". Percentage difference could be mentioned in brackets.
Response: The table is modified as suggested in the revised manuscript with percentage (%) difference provided in brackets and the difference in number is represented by “ΔN” instead of “N”
Comment #51: Line 490-500: Use ICIMOD and GAMDAM rather than ICIMODGI and GAMDAMGI.
Response: GI was added to reduce the word count and also it becomes monotonous to append the phrase “glacier inventory” each time an inventory name is mentioned. Therefore, we are retaining GIs after global, and regional inventories.
Comment #53: Table 9: Above in methods authors mention having used median glacier elevation rather than mean glacier elevation.
Response: Sorry, we have used the mean rather than median and the same been corrected in the methods section of the revised manuscript.
Comment #54: Section 5.3 and associated tables/figures should be removed.
Response: As mentioned in response to Comment#8 and Comment#37, we find merit in retaining the elevation change information in the revised manuscript.
Comment #55: Line 675-680: The authors mention large glaciers at high altitudes and then low glacier cover at high altitudes which is a kind of contradiction. Large glaciers as authors suggest in results contribute to glacier cover. Please explain.
Response: The glacier cover observed in the study region is in general concentrated at higher altitudes for example above 4500 m asl in the Suru and Chenab basins which is justified by the elevation and temperature relationship for glacier growth. However, glacier coverage starts to decrease considerably with the further increase in the altitude e.g., above 5500 m asl in the Suru and Chenab which has been attributed to the steeper headwalls above this altitude facilitating snow/ice avalanches, thereby precluding the glacier formation. The sentences have been rephrased in the revised manuscript for more clarity and better understanding.
Comment #56: Table 14: Remove
Response: Please see our response to the suggestion above (Comment #37 and Comment#54)
Comment #57: The authors do not discuss much about the aspect (neither in results nor discussion) except very qualitatively.
Response: More details regarding aspect have been added in the revised manuscript.
Comment #58: Section 6.2 should be "Inconsistencies in existing glacier inventories"
Response: The heading has been modified, as suggested, in the revised manuscript.
Comment #59: Figure 5: Mention background image and band combination.
Response: The background images is FCC (7,4,2) of Landsat ETM+ dated 04-09-2000. The information has been added to the figure caption.
Comment #60: Line 720-724: The overlap ratio could be misleading since the inventories were computed using different methods. Please justify.
Response: Nagai et al. (2016) have demonstrated usefulness of the overlap ratio to assess the consistency of glacier outlines including location shifts which would be difficult to assess using the absolute value of delineated areas. Furthermore, the results of the overlap ratio observed in the present study are corroborated by the comparison of the glacier inventories in absolute terms (number and area). For example, the overlap ratio between KUGI-ICIMOD combination is relatively poor and the same is reflected by relatively larger differences in glacier number and area. Therefore, overlap ratio is quite useful indicator to assess the consistency of glacier inventory irrespective of the methodology used to delineate glacier boundaries.
Comment #61: Figure 6: SG. On checking from Google Earth, the existing inventories have rightly followed GLIMS definition of glacier delineation and divided the ice into two polygons since the ice fluxes move in different ablation zones. However, the authors have erred here (and maybe in many such cases) by considering it as a single glacier. The authors should know that although the ice masses are connected (in the accumulation zone) the movement of ice in different directions owing to the ridge-topography divides the ice into two polygons and hence two glaciers (refer to GLIMS definition of glaciers). This appears to be a huge flaw with the interpretation by authors as the number of glaciers is massively underestimated in KUGI. This needs to be corrected in the data as well as explanations to the data.
Response: We have followed the GLIMS criteria, by dividing a glacier into two or more polygons as determined by the underlying ridge topography during the inventory. It is fact that the latest Google Earth imagery shows that the glacier has fragmented, however, when we closely look at the Google Earth data, the glacier has fragmented or the under lying ridge divides the glacier far away from the position it was previously divided in case of GAMDAM and ICIMOD glacier inventories. Neither, GAMDAM nor the ICIMOD has divided the glacier where it appears to have fragmented in the recent years. Also, when we drape the 2000 satellite data over DEM, the ridge-topography is not prominent enough to manifest the flow direction/ridge divide on the satellite images especially in the accumulation zone. Therefore, there is no question of dividing a glacier into multiple polygons.
Comment #62: Figure 7: Mention background image and band combination.
Response: The image source and the band combination has been specified for the figure in the revised manuscript.
Comment #63: Remove section 6.3
Response: Please see our response to the similar comments above.
Comment #64: Conclusions:
While the authors mention "limited field survey" at many places in the text, they have failed to showcase the data collected, the photographs depicting the glacier environments in these three catchments. They should show field data and photographs from all three catchments and demonstrate its usefulness in inventorying the glaciers in these three respective areas.
Response: This is again repeating comment. Please see our response to Comment #2; Comment #9.
Comment #65: Delete sentences about geodetic mass changes.
Response: Again, as explained above, we want to retain this section in the revised manuscript.
Comment #66: The authors fail to convince the robustness of KUGI compared to at least RGI and GAMDAM. This needs to come up very in the results, discussion as well as the conclusion.
Response: This is a repeating comment. Please see the response to the similar comments above, particularly the author response to Comment #9.
-
AC3: 'Reply on CC1', Romshoo Shakil Ahmad, 08 Jun 2021
-
CC2: 'Comment on essd-2021-28', Sher Muhammad, 04 May 2021
The authors derive a new glacier inventory for selected Himalayan river basins using manual delineation and various data sources. The authors also highlight the strength of their inventory through the field data. The derived inventory is compared with RGI, ICIMOD, and GAMDAM inventories and highlighted the limitations in the mentioned inventories. In addition to the comparison of inventories, the authors estimated the surface elevation changes of glaciers in the basin between 2000 and 2012. It is important and interesting to see the comparison of various inventories (e.g., Muhammad et al., 2019a) to support the glaciological community to use the most appropriate inventory for their research. I only review part of the manuscript and suggest few comments to incorporate in the revision to strengthen their manuscript.
- Interesting to see that ICIMOD inventory is not only underestimating (as in the Karakoram (Muhammad et al., 2019a) but also overestimating. The main reason for underestimation in the Karakoram by ICIMOD inventory is the slope criteria. Most of the glaciers are avalanche-fed in the Karakoram and the accumulation falls on the steep slopes which is mostly not considered. However, I found that the inventory here shows that there is overestimation as well in the ICIMOD inventory. The authors are suggested to discuss the overestimation in ICIMOD inventory and its potential reasons and also discuss the results in comparison with Muhammad et al., 2019a.
- The authors manually digitize the glaciers which is extremely inconvenient in the presence of state of the art automatic techniques considering the >2000 glaciers. Mapping only a single (medium to large) glacier with manual digitization takes several hours. Usually, automatically derived extents are improved using manual digitization but the approach is different here. The authors may explain why they selected manual digitization.
- Also, it is unclear why the authors use topographic parameters if they use manual digitization? These parameters are useful when the glaciers are automatically mapped.
- The authors indicate field surveys data for glacier inventory validation but did not show the results of the survey anywhere (in any figure or text). The authors are suggested to add detailed information of the field survey including 1) the number of glaciers surveyed in the field, 2) what kind of information/data collected in the field, 3) how the survey information/data improved/validated the remote sensing results?
References
Farhan, SB, Zhang, Y, Ma, Y, Guo, Y and Ma, N (2015) Hydrological regimes under the conjunction of westerly and monsoon climates: a case investigation in the Astore Basin, Northwestern Himalaya. Clim. Dynam., 44(11–12), 3015–3032
Muhammad, Sher, Lide Tian, and Asif Khan. "Early twenty-first century glacier mass losses in the Indus Basin constrained by density assumptions." Journal of Hydrology 574 (2019a): 467-475.
Muhammad, S., Tian, L., & Nüsser, M. (2019b). No significant mass loss in the glaciers of Astore Basin (North-Western Himalaya), between 1999 and 2016. Journal of Glaciology, 65(250), 270-278. doi:10.1017/jog.2019.5
Citation: https://doi.org/10.5194/essd-2021-28-CC2 -
AC4: 'Reply on CC2', Romshoo Shakil Ahmad, 08 Jun 2021
CC#2
General Comment: The authors derive a new glacier inventory for selected Himalayan river basins using manual delineation and various data sources. The authors also highlight the strength of their inventory through the field data. The derived inventory is compared with RGI, ICIMOD, and GAMDAM inventories and highlighted the limitations in the mentioned inventories. In addition to the comparison of inventories, the authors estimated the surface elevation changes of glaciers in the basin between 2000 and 2012. It is important and interesting to see the comparison of various inventories (e.g., Muhammad et al., 2019a) to support the glaciological community to use the most appropriate inventory for their research. I only review part of the manuscript and suggest few comments to incorporate in the revision to strengthen their manuscript.
Response: We express our gratitude to the reviewer for suggestions and comments on the manuscript. The comparison of the present study with Muhammad et al., (2019) has been incorporated in the revised manuscript. However, the point by point response to the detailed comments and suggestions raised by the reviewer are provided as follows:
Comment#1: Interesting to see that ICIMOD inventory is not only underestimating (as in the Karakoram (Muhammad et al., 2019a) but also overestimating. The main reason for underestimation in the Karakoram by ICIMOD inventory is the slope criteria. Most of the glaciers are avalanche-fed in the Karakoram and the accumulation falls on the steep slopes which is mostly not considered. However, I found that the inventory here shows that there is overestimation as well in the ICIMOD inventory. The authors are suggested to discuss the overestimation in ICIMOD inventory and its potential reasons and also discuss the results in comparison with Muhammad et al., 2019a.
Response: The overestimation observed in case of the ICIMOD inventory is largely attributed to the misinterpretation of snowpacks as glaciers as demonstrated in the Fig. 5. Furthermore, the description of the results with respect to Muhammad et al. (2019) is provided in the revised manuscript. Both these points have been discussed in details in the revised manuscript.
Comment#2: The authors manually digitize the glaciers which is extremely inconvenient in the presence of state of the art automatic techniques considering the >2000 glaciers. Mapping only a single (medium to large) glacier with manual digitization takes several hours. Usually, automatically derived extents are improved using manual digitization but the approach is different here. The authors may explain why they selected manual digitization.
Response: Agreed that the automatic glacier boundary delineation followed by manual correction is one of the preferred approach for glacier mapping from satellite images especially over large regions because of the less time required when compared to the manual digitization techniques. However, automatic glacier delineation technique poses a significant challenge for mapping debris covered glaciers particularly the glacier terminus in the Himalaya. In fact, the reflectance of the supra-glacial debris cover is similar to the surroundings which results in the exclusion of such areas from the glacier extents. Furthermore, seasonal snow, cloud cover and shadow also pose a significant challenge in mapping Himalayan glaciers using automatic image delineation techniques. Therefore, to overcome these challenges in mapping glaciers in the Himalaya, we used multiple data sets including thermal data, high resolution imagery, time series of satellite data etc. which is not possible to use in the automatic approach. Furthermore, the local knowledge/field experience of an analyst also proves very useful in precise glacier delineation which is again not possible in the automatic approach. Since a considerable number of glaciers in the present study have debris-covered termini and we found it appropriate and necessary to map the glaciers manually to minimize the errors and uncertainties in the glacier inventory. Furthermore, advantage of the manual approach over the automatic approach for mapping glaciers when debris covered, shadowed and seasonal snow-covered area has been previously demonstrated by Paul et al. (2013); Nuimura et al. (2015).
KUGI improved the mis-mapped glacier outlines/boundaries from the automatic approach and any mismatches of the glacier geometry due to the seasonal/temporal snow cover and shadows were manually corrected using additional Landsat images and Google earth images. Further the mapped glaciers with better geo-referencing were overlaid with high resolution images in Google Earth environment for validation wherever available. Though, the mis-mapped/mis-located outlines, observed on the global/regional inventories, may have only limited effect on measurements of glacier area, but can introduce serious errors into applications that rely on absolute positioning (e.g. co-registration to other datasets such as DEMs). The only realistic way to correct them is to provide more accurate outlines using manual approach as was done in the KUGI which would serve as source of improved outlines for the scientific community interested in conducting various application studies using the glacier outlines.
Comment#3: Also, it is unclear why the authors use topographic parameters if they use manual digitization? These parameters are useful when the glaciers are automatically mapped.
Response: As specified in the methodology section (section 4.1), under the surface conditions on headwalls with slopes (topographic parameter) exceeding 40˚, we specifically verified such glaciers from the Google Earth image for accurate delineation of glacier extents. Further, the satellite images draped on DEM (hillshade) was found useful in demarcating glacier outlines when the ridges particularly in the accumulation zone were covered with seasonal snow (Paul et al. 2004; Paul et al. 2009). The overall visualization of a glacier in 3D helped in the precise mapping of glacier outlines in KUGI.
Comment#4: The authors indicate field surveys data for glacier inventory validation but did not show the results of the survey anywhere (in any figure or text). The authors are suggested to add detailed information of the field survey including 1) the number of glaciers surveyed in the field, 2) what kind of information/data collected in the field, 3) how the survey information/data improved/validated the remote sensing results?
Response: In the present study, we have done field surveys on 20 glaciers located across the study area, which we visit almost annually for other glaciological studies. The locations and the field photographs of these glaciers has been provided as Supplementary figure in the revised manuscript. We have collected the snout positional data of the debris-covered tongues of these glaciers to validate the glacier delineation. The field measurements of these glaciers acted as an interpretation tool for delineation of the debris-covered glaciers in the study area. The terminus of the heavily debris covered glaciers like the Hoksar glacier in Jhelum basin was not easily mappable even using the thermal and google earth imagery. We therefore, mapped the glacier terminus in the field and further observed that the debris on the glacier is relatively smooth and aligned in the direction of glacier flow when compared to the debris-cover in the surroundings which was found a useful field-based information for mapping debris-covered termini of other glaciers when viewed on Google earth images. Further, eight of these twenty glaciers have been designated as benchmark glaciers and are continuously studied for mass balance, GPR, debris thickness, Surface mapping temperature profiling and other glaciological studies since the last 5-8 years. In addition, the glacier outlines of several other glaciers in the vicinity of these 8 glaciers in the three basins have been verified during annual glacier field expeditions during the last 5-8 years.
Additionally, all the heavily debris covered glaciers and a majority of the clean glaciers, numbering more than 850, were qualitatively verified on the Google Earth image for the rectification of any delineation error. A similar approach of quality check using Google Earth has been previously adopted by Nagai et al. (2016) and several others and is an accepted method for validation of such a large number of glaciers located in inaccessible complex terrain.
Field photographs of the about 20 selected glaciers in the three basins, which have been visited over the last few years for field measurements/validation along with their GLIMS ID are presented in the revised manuscript (The field photographs of some of these glaciers are provided below). A locational map of these glaciers in the form of a KML file would be provided as a supplemental material in the revised manuscript ((The field photographs have also been attached as a supplement file in pdf format)).
Comment#5: References
Farhan, SB, Zhang, Y, Ma, Y, Guo, Y and Ma, N (2015) Hydrological regimes under the conjunction of westerly and monsoon climates: a case investigation in the Astore Basin, Northwestern Himalaya. Clim. Dynam., 44(11–12), 3015–3032
Muhammad, Sher, Lide Tian, and Asif Khan. "Early twenty-first century glacier mass losses in the Indus Basin constrained by density assumptions." Journal of Hydrology 574 (2019a): 467-475.
Muhammad, S., Tian, L., & Nüsser, M. (2019b). No significant mass loss in the glaciers of Astore Basin (North-Western Himalaya), between 1999 and 2016. Journal of Glaciology, 65(250), 270-278. doi:10.1017/jog.2019.5
Response: The suggested literatures references have been incorporated in the revised manuscript.
Data sets
Evaluation of the Global Glacier Inventories and Assessment of Glacier Thickness Changes over North-western Himalaya Shakil Ahmad Romshoo, Tariq Abdullah, and Mustafa Hameed Bhat http://doi.org/10.5281/zenodo.4461799
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,462 | 795 | 90 | 2,347 | 62 | 77 |
- HTML: 1,462
- PDF: 795
- XML: 90
- Total: 2,347
- BibTeX: 62
- EndNote: 77
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Cited
3 citations as recorded by crossref.
- An Assessment of Glacier Inventories for the Third Pole Region X. He & S. Zhou 10.3389/feart.2022.848007
- Mass balance estimation of Mulkila glacier, Western Himalayas, using glacier melt model G. M. et al. 10.1007/s10661-022-10458-1
- Glacier Volume Estimation Using Laminar-Flow and Volume–Area Scaling Techniques in the Chenab Basin J. Gopika et al. 10.1007/s12524-023-01744-7