the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 26 May 2025
Understanding the Gangotri glacier dynamics: Implications from a fully distributed inversion of equivalent water-volume change
Abstract. The Gangotri Glacier is scientifically controversial regarding its dynamics, ice thickness, volume, and mass balance due to the lack of field data. Evidence of rapidly increasing temperatures with climate change is clearly visible in the concerning mass changes of Himalayan glaciers. Subsequently, monitoring glacier volumes is critical for managing regional water resources and predicting glacier dynamics. The Gangotri glacier, a significant water resource for northern India, is experiencing significant changes due to climate change. This study emphasizes its dynamic nature from 2016 to 2023. Ice thickness distribution of Gangotri glacier estimated using velocity and shear stress-based approach. Sentinel-2 multi-spectral imagery is used to estimate glacier velocity with three different approaches for comparative assessment of the ice dynamics based on pixel-wise cross-correlation. A laminar flow-based approach is applied to determine the thickness of the Gangotri Glacier. The thickness change of the study period is used to estimate the mass balance and equivalent water volume change of the glacier. The observed velocity ranged from 31 ± 5.8 - 81 ± 15.12 ma-1 in the accumulation area to 15 ± 2.8 - 28 ± 5.23 ma-1 near the snout, and the thickness varied from 580 ± 74.47 m in the upper reaches to 70 ± 9 - 115 ± 14.77 m near the snout. Through this study, we found that the mass wastage of the glaciated ice during the study period was -1.01 ± 0.403 m w.e. a-1 (meter water equivalent), and the mean glaciated ice volume was 19.70 ± 2.64 km3. We observed the volumetric change is a declining pattern of the study period 2017 to 2023 gradually. The climatic parameters observed an increasing trend over the last two decades. We also found that the Apparent Thermal Inertia (ATI) increased which determined the debris accumulation over the ablation zone significantly from the side wall of the glacier due to fluctuation of the temperature differences (Thaw-freezing). These changes denote a significant reduction in the water storage capacity of the Gangotri Glacier.
- Preprint
(3434 KB) - Metadata XML
-
Supplement
(2032 KB) - BibTeX
- EndNote
Status: open (until 02 Oct 2025)
-
RC1: 'Comment on egusphere-2025-1614', Anonymous Referee #1, 07 Jul 2025
reply
Review of
Understanding the Gangotri glacier dynamics: Implications from a fully distributed inversion of equivalent water-volume change
by Anikul Islam et al.This manuscript uses a diverse compilation of remote sensing data sets to investigate the recent evolution of Gangotri Glacier in the Himalaya. A special focus is laid on volume changes and implied annual mass balances. The authors used optical satellite images for updating available glacier boundaries. Glacier surface velocities were determined by feature tracking, using different software solutions, from optical and NIR satellite imagery, while published digital elevation models, also from remote sensing sources, were used for glacier geometry and the long term mass change analysis. Additional data sets were used as a control of the analysis, as well as for relating the detected glacier changes to environmental conditions. Inferring the ice thickness across the glacier by different ice dynamic methods, basically using the shallow ice approximation, represents the core of the manuscript. Annual ice velocity fields were determined for this purpose and combined with additional information, like glacier surface slope, ice temperature and assumptions about geometrical constraints and basal sliding.
The basic results agree rather well with existing global data sets, demonstrating the correct application of the widely used algorithms for surface velocity determination and ice thicknesses. This is a comprehensive study of combining remote sensing data for inferring knowledge about the status of one of the best-studied glaciers in the Indian Himalaya. The data sets provide the opportunity to analyse the decadal changes of the glacier since 2000, which however is not fully exploited. Even though the manuscript presents rather interesting data about ice velocities and ice volume change, there are several problematic issues, which require a critical evaluation. The authors calculate ice thicknesses on an annual basis for determining annual thickness changes, due to the lack of available annual elevation models. Unfortunately, these results are used in a rather uncritical approach for discussing mass balance and climate related reasons for such thickness changes, even though the volume variations are not significant with respect to the error calculations provided. In addition. the error estimates only cover the uncertainties originating from the inherent data characteristics, but not from the glacier conditions in space and time. Surface velocities can be determined to a satisfying degree of accuracy in the ablation region of glaciers, but it becomes difficult in the accumulation region, where suitable features are rare and contrast is very low. The manuscript completely lacks a discussion about the quality of surface velocities across the glacier. The shape factor is assumed to be constant for the entire glacier, which is clearly not applicable in such a complicated environment. The assumption of the basal sliding component is a rather crude estimate based on a general assessment of Himalayan glaciers, while seasonal variations from the surface velocity analysis were not utilised for a more site specific approximation. The slope of the glacier was considered constant for the analysis period, based on the DEM from 2018 and there is no information about the spatial choice for slope calculations, which is crucial for velocity determinations. Even though it can be assumed that the surface slope is not changing to a large degree within the seven years, this implies an additional error for the final results. In the end a critical assessment of the determined volume variations and their limitations is not provided, especially about the validity in slow moving regions like the small, high elevation accumulation basins with poor velocity information. The small variations in glacier volume between 2018 and 2023 are not significant at all, while potential reasons for the large deviations, especially in 2017, are not discussed in this framework. It is surprising that the DEM from 2018 is not used to evaluate the volume changes from 2015 to 2018 in comparison with the volume changes from 2016 to 2018 derived from the velocity analysis. This could provide a general idea about the validity of the analysis.In the following, I provide more detailed remarks about the sections 1-5. The further sections probably require a serious reconsideration, after implementing a realistic assessment of the methods and results.
Specific remarks:
L. 42/43: Himalayas vs HKH, this is not the same.
L. 47: the three basins covert he entire region
L. 51: what does glaciers below 5700 m mean? The entire glacier below this elevation, or only the terminus?
L. 54/55: This statement is not supported by the references provided. Prasad et al (2009) onls considers a small sample of glaciers and Bhambri & Bolch (2009) state that the length changes are highly heterogeneous, but the data basis does not provide evidence for centennial changes of the majority of the glaciers. Indeed, the largest glacier class is the class of small glaciers, which would already have disappeared with the cited retreat rates.
L. 56: What does that mean “ongoing melting and stagnation”? Stagnant glaciers, very likely do not experience exceptional melt.
L. 59: This sentence has no connection to the text above. What do you want to express with this statement?
L. 83/84: Is the inversion of ice thickness distribution a major aim of the paper? It is one application of glacier surface velocities among others, which then also could be named. Otherwise elaborate, why you specifically mention thickness inversion and connect it to the following paragraph.
L. 98-107: This reads like a loose compilation of methods and experiments without a clear goal. There is a large difference between global estimates and detailed reconstructions of individual glaciers. The last sentence is out of context. What is the reference time for the reduction of ice volume in High Mountain Asia?
L. 108-117: This is again a loose collection of methods missing a clear context. The existence of different methods for ice thickness inversion is already mentioned at lines 85 ff. It would be better to prepare a concise paragraph about the different methods and the application of these methods in a structured way.
L. 144-152: The general description of the Ganngotri Glacier setting is rather unclear. ELA estimation is based on observations at Dokrani Glacier, missing information how this was done. The temperature range of the Gangotri Glacier region is not plausible. There is a seasonal variation and an altitude variation. It is unlikely that there are positive temperatures at 7000 m elevation. What are the given values, daily means, monthly means? The same is true for the humidity level, which rather likely varies considerably during the seasons (dry winter and humid monsoon). The precipitation magnitudes are very likely far too low, as measurements at high altitudes do not exist. What do the two values represent, seasonal or altitudinal variability? What is the period?
L. 161-163: This sentence is not comprehendible.
L. 166: This is not the vertical accuracy, it is the mean vertical error which has a rather large RMSE of almost 15 m. Therefore, it depends very much on the individual case, how well the true elevation is represented.
L. 168/169 This is not comprehendible.
L. 170/171: What is the reason for using the Cartosat DEM for the slope, but the Copernicus DEM for the DEM differencing?
L. 171-174: Why did you have to adapt the glacier boundaries?
Table 1: The table presents more data sets as are described in the data section. What were the missing data used for?
L. 188: What does “basal shear stress of glacier surface” mean?
Section 3.1 is rather cumbersome to read, as you predominantly write about the future steps (GIV analysis) instead of presenting details about the pre-processing of the images and the techniques used for glacier delineation.
L. 231/232: repetition of the first sentence in the section.
L. 232: What is overlapped? Please be specific about the method. What is the iterative method?
L. 237/240: This is not comprehensible. Please report which filters were used (for comparison? For which purpose?) in which step of the processing.
L. 240-243: Why do you need this upper velocity threshold? Please explain. I guess there was not only one velocity map, but maps with annual time spans?
L. 2447/245: In lines 2010/2011 you mention that you use the NIR band images from October and November. Here you state that you use the images from July-October?
L. 251-264: It seems a bit strange that you provide so many details about the COSI-Corr algorithm, while ImGraft and GIV is presented rather briefly. It would be much better to shortly describe the core principle of all three algorithms and then provide some details about the specific application to your data set (sample size, iteration steps, overlaps, multi-band samples, etc.) followed by a short description of the resulting products.
L. 284/295: and the potential bed deformation. The formula in Cuffey and Paterson assume a hard bed.
L. 300/301: Equation 2 shows the glacier surface velocity as a combination of the contribution from internal deformation and basal sliding.
Eq. 4: this equation is only correct for beta=0. Therefore I assume some mistake in the formulation.
L. 313/314: I am not convinced that you can apply a general estimate for a specific study, as the Gangotri Glacier does not represent a “mean” glacier of the Himalaya. It would be much better to try and discriminate between annual velocities and winter velocities, in order to estimate the magnitude of sliding. In addition, basal sliding is not constant across the entire glacier. Therefore it very much depends on the purpose of your thickness inversion if a constant value is reasonable.
L. 314/315: How do you discriminate between driving stress and basal shear stress, based on your available data?
L. 316/317: This sentence is not comprehendible. Where do you get the ice temperature from?
Eq. 6: f is not “measureable” and you do not have information about H.
L. 326: Why do you use a density of 850 kg/m³ for ice?
Eq. 7: Can you explain how you get rid of u_b? Refer to the formulation of beta in eq. 4.
Section 3.3.2: This is a rather crude estimate of ice thicknesses for calculating glacier volumes.
L. 355: It seems a bit odd that you state a study period of 2000-2015, while the velocity fields are determined for 2016-2023. This is probably not the study period, but the period of thickness change analysis.
L. 364: What do you want to say with this sentence? What is CRS?
L. 370: How can a publication from 2000 reference a publication from 2011? Or do you refer to a method, developed in 2000 and used be the authors from in the 2011 publication?
Eq. 10: this equation relates to the correction of the DEM during the rectification process. This should be made clear.
L. 381: Why do you not discriminate between the penetration depth of snow and ice?
L. 394: Can you specifiy, why you need a Monte Carlo approach for the mean glacier-wide ice thickness? It this included in your approach according to section 3.3.2?
L. 397: what is the meaning of the mean ice thickness for a single pixel?
L. 404/405: Why did you not utilize the DEM from 2018 for an elevation change analysis? This you enable you to check the validity of your annual elevation changes at least for the period 2016-2018.
Eq. 13: You cannot use the ice density for the density conversion to water equivalent, because you neglect the effect of the firn body. Either r is the pixel area, or you need to write r².
L. 421: The method is not limited by cloud boundaries, but but cloud cover.
L. 425: what about the accumulation area?
L. 428: the 25° is the maximum slope of the stable ground? It is not clear how you define the velocity uncertainty in the end.
L. 438-442: Several of the values used seem rather crude estimates (e.g. for beta). Some more justifications are required for their choice.
Fig. 4: It is not clear what it the time basis for the individual sub-plots. I guess that a) is the mean over the 2016-2023 period. The thinning rate in c) is the mean of the annual thickness changes, d) represents the total thickness change between 2000 and 2015, e) is the mean like in c) but for the central flow line, while d) is equal to c) but scaled for w.eq.
L. 523: positive thinning rate relating to thickening is rather unfortunate. Please consider to reformulate.
L. 527/528: How do you calculate a total mean thinning of -8.39 m, while the mean thinning rate is 0.5 m/a? I would expect that the mean thinning rate is in fact the mean thinning divided by the time.
Table 3: These results do not reflect variations in the glaciers´ mass balance, they just reflect the uncertainty inherent of the approach. The variability is considerably smaller than the error. In column 3 it is not ice volume, but glacier volume, as this volume also contains firn and snow.
L. 536-538: These numbers seem much too small to be detectable. How would you resolve 500 m³/a?
L. 544: This is not in line with your results in Table 3, which rather shows a very variable year to year potential mass balance. Even though, as noted above, this is not a significant result due to the involved errors.
Section 6.1.1. This is the wrong discussion. If you compare remote sensing products between each other, it does not tell anything about the reliability of the results, but only about the comparability of the methods. It is common knowledge that velocity inversions are especially affected by errors in the accumulation zone, which is also demonstrated by the strong variability of the individual results in Fig. 6 c. The discussion should therefore focus on the potential reliability of the results for different regions of the glacier. The velocities might represent the real velocities on the glacier tongue reasonably well, while the velocities in the accumulation zone rather likely need to be considered with care. By the way, which periods are compared in this analysis? Cover the ITS-Live date the same period as in your analysis?
L. 620-622: Why do you use such a weird unit, m/d? Please be consistent.
L. 620-626: Comparing mean glacier wide velocities has no real value, as dynamic changes can happen, even if the glacier wide mean velocity remains almost constant. If you want to compare your results to other investigations, you should focus on specific regions, the main flow line or other representative units.
Section 6.1.2: again this analysis only shows how well the different algorithms agree, but not if this results in a more reliable ice thickness distribution. It would be rather interesting to discuss the influence of the different parameters on the ice thickness distribution, like the unknown ice temperature for different parts of the glacier, or the uncertainties in the slope, especially for steep areas, where remote sensing methods are affected by larger errors.
Section 6.2.2: The long term mass loss of Gangotri Glaciers seems a robust result of this analysis. This seems to be in accordance with other studies. However, the comparison could be more focussed on the temporal evolution (still taking the errors in consideration), as most of the studies cover different periods. Also, the comparison of the DEM-differencing results from 2000-2015 and the thickness changes from 2016-2023 could be compared in a more detailed way. Again there raises the question, why the DEM from 2018 is not used to combine the two periods.
Fig. 8: considering the inherent errors of the method, this evolution of the ice volume rather likely cannot be used for any analysis of volume change. It is also necessary to show the errors in the graph.
L. 716-727: This is an interesting attempt to relate the volume changes to climatic variations. But I doubt that the results are robust enough. At least this needs to be discussed in the manuscript.
Fig. 9: The source of the data is only mentioned in the text, but not in the caption. Maximum temperature is not a valuable parameter, as it does not provide any information about the seasonal melt potential. How is the runoff calculated in this data set?Citation: https://doi.org/10.5194/egusphere-2025-1614-RC1
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 449 | 42 | 12 | 503 | 39 | 11 | 29 |
- HTML: 449
- PDF: 42
- XML: 12
- Total: 503
- Supplement: 39
- BibTeX: 11
- EndNote: 29
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1