Climate sensitivity of the summer runoff of two glacierised Himalayan catchments with contrasting climate
- 1Earth and Climate Science, Indian Institute of Science Education and Research (IISER) Pune, Pune-411008, India
- 2National Centre for Polar and Ocean Research (NCPOR), Ministry of Earth Sciences, Vasco-da-Gama, Goa-403804, India
- 1Earth and Climate Science, Indian Institute of Science Education and Research (IISER) Pune, Pune-411008, India
- 2National Centre for Polar and Ocean Research (NCPOR), Ministry of Earth Sciences, Vasco-da-Gama, Goa-403804, India
Abstract. The future changes in runoff of Himalayan glacierised catchments will be determined by the local climate forcing and the climate sensitivity of the runoff. Here, we investigate the sensitivity of summer runoff to precipitation and temperature changes in winter-snow dominated Chandra (the western Himalaya) and summer-rain dominated upper Dudhkoshi (the eastern Himalaya) catchments. We analyse the interannual variability of summer runoff in these catchments during 1980–2018 using a semi-distributed glacio-hydrological model, which is calibrated with the available runoff and glacier mass balance observations. Our results indicate that despite the contrasting precipitation regimes, the catchments have a similar runoff response: The summer runoff from the glacierised parts of both the catchments is sensitive to temperature changes and is insensitive to precipitation changes; the summer runoff from the non-glacierised parts has an exactly opposite pattern of sensitivity for both the catchments. The precipitation-independent glacier contribution stabilises the catchment runoff against precipitation variability to some degree. The estimated sensitivities capture the characteristic ‘peak water’ in the long-term mean summer runoff, which is caused by the excess meltwater released by the shrinking ice reserve. As the glacier cover depletes, the summer runoff is expected to become more sensitive to precipitation forcing in these catchments. However, The net impact of the glacier loss on the catchment runoff may not be detectable, given the relatively large interannual runoff variability in these catchments.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(1475 KB)
-
Supplement
(927 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1475 KB) -
Supplement
(927 KB) - BibTeX
- EndNote
Journal article(s) based on this preprint
Sourav Laha et al.
Interactive discussion
Status: closed
-
RC1: 'Comment on hess-2022-205', Anonymous Referee #1, 16 Aug 2022
In this paper, the authors quantified the sensitivity of summer runoff to precipitation and temperature changes in two glacierized Himalayan catchments with contrasting climate based on a hydrological model. This study is well prepared, and the results are meaningful. However, there are still some questions needing to be clarified. Here are the details:
- Lines 48-49: This expression should be careful. In my opinion, the simulation is not as same as the observations. The observed data can represent the reality at a point scale but is hard to obtain, especially in the high mountain regions, while the simulation can systematically analyze for a basin-wide scale over a long period.
- Figure 1. The boundaries of the two study basins should be highlighted on the map.
- Lines 65-66: This sentence is inaccurate. As I see, the annual temperature of Chandra (-55â) is lower than Upper Dudhkoshi (-4.7â). In addition, the glacierized fraction of Chandra (0.25) is higher than Upper Dudhkoshi (0.20), and the former glacier area is more than two times the latter. These differences are significant and have a large impact on the glaciohydrology. So please revise it.
- Lines 165-166: How to calculate the glacier area change? There is only the glacier mass balance change data in the supplementary Figure S3. In addition, have the model considered the compensation of snow and transforms into ice?
- Supplementary Figure S1: How to deal with the observed data gaps?
- Line 220-225: I think the temperature before the ablation season can also influence the glacier melt and snowmelt since it controls the distribution of rainfall and snowfall in the accumulation season. Especially in the Chandra basin, where most precipitation occurs in the winter. Thus, I suggest that the authors should add a temperature sensitivity experiment before the ablation season.
- Lines 312-317: How to define the glacier runoff in this paper and what is the difference between it and the glacier ice loss? Moreover, the results show that the glacier runoff contribution to the total summer runoff in upper Duhkoshi is higher than that in Chandra basin while the glacier cover in upper Duhkoshi is lower than that in Chandra and the former summer precipitation is much higher than the latter, which seems contradictory. Please show the other contributions of the summer runoff and clarify this contradiction.
- Lines 351-354: How does the glacier hypsometry affect the mass-balance sensitivity, and how is this factor considered in the model?
- Lines 377-379: In my opinion, the results vary from different studies, I think the authors should discuss the reason for the difference among the studies at different basins rather than describe it as “largely in line with”.
- Figure 7: The color scheme is too blurry to distinguish.
- Line 389: The precipitation increased while the rainfall on glacier did not change, why? Please clarify it.
- Lines 464-466: The RCP2.6 scenario data has been used in this paper, but has no introduction (e.g. which general circulation model has been selected and the evaluation of the projected data).
- AC1: 'Reply on RC1', Argha Banerjee, 13 Sep 2022
- AC3: 'Reply on RC1', Argha Banerjee, 13 Sep 2022
-
RC2: 'Review of hess-2022-205', Anonymous Referee #2, 16 Aug 2022
This is a well written manuscript on a very relevant topic regarding the contribution of streamflow generated in the glacier-covered part of a catchment to catchment-scale water resources. It uses two contrasting glacierised Himalayan catchments, one of which is winter-precipitation dominated, Chandra (the western Himalaya), and the other one summer-precipitation dominated, upper Dudhkoshi (the eastern Himalaya). For these catchments, climate sensitivities of simulated streamflow is obtained by regressing the simulated variability of streamflow to the one its meteorological drivers. The used model is a the Variable Infiltration Capacity (VIC) model, augmented with a glacier melt module.
The analysis is model-based; , the used precipitation-glacier-melt-streamflow model is very simple for the glacier-covered catchment part; as far as I see, it sums up the ice melt and the snowmelt (and rainfall) and routes it through a single (or perhaps two, unclear) linear reservoir, i.e. the corresponding streamflow response has a single time scale stemming from icemelt and snowmelt and no baseflow, thus the model can most likely not simulate a water carry-over effect from month to month for the glacier part. This model structure might have a different impact on the estimated sensitivities for the different analysed catchments. Furthermore, we do not have information on how large the (ignored) debris cover is nor on how important snow redistribution is, we simply know that it is ignored.
Only two parameters of the hydrological model are calibrated, the ones that affect the water balance the most strongly (melt factor for ice and precipitation scaling factor). The calibration is on streamflow and glacier mass balance; there is an empirical weight factor to combine the performance with respect to both quantities; despite a clear lack of giving any formal statistical framework, the parameter estimation approach is called a Bayesian inference.
Accordingly, I am rather skeptical about the added value of this model study; I think that this is essentially a modelling exercise without clear indications that it actually corresponds to how nature reacts; moreover, the conclusion is very general with new insights that can be inferred from general process knowledge such as e.g. the sentence “the temperature sensitivity of the glacier runoff and the precipitation sensitivity of the off-glacier runoff are critical determinants of the future changes of summer runoff and its variability in these two catchments”.
I therefore recommend rejection of this version. The work could become more valuable if it was more critical about the value of the model, if it discussed what we miss with the simplifications and if it provided more insights in what we can learn from the two different types of catchments.
Detailed comment:
- what do you mean by runoff? there are usages of this term where it does not include groundwater-fed baseflow; accordingly: if we mean total flow leaving a catchment, we might want to use streamflow;
- it needs to be very clear also what is meant by “glacier runoff”: runoff generated in the glacier-covered part of the catchment? including baseflow? Including runoff from lateral moraines that are not glacier covered?
- Methods, calibration: I do not think that 5% or 10% error on summer streamflow observations is a realistic value; since summer is the period of high flow, this values is certainly, much higher; furthermore, in the chosen formulation, the error should correspond to the total model error and not just to the observational error (see e.g. an Bayesian inference paper by Dmitri Kavetski); the Bayesian formulation for the mass balance is based on very few obs. values; what assumption do you make about the distribution of the residuals for streamflow and mass balance? i.e. what motivates the chosen form of the likelihood? How did you compute the posterior (you did not use any sampling method that would yield a sample for the posterior; I guess you did some kind of rescaling?)
- Methods, other: i) what is the used temporal time step of the VIC model on the glacier part ? why is it reasonable to keep the bias correction constant in space? I would expect that biases depend on elevation? Why does the glacier melt model not use the energy-balance approach? Is the glacier melt coded by the authors of the study or someone else?
- Methods: the computation of glacier mass balance sensitivity is not clear to me; did you run the model with modified precipitation input?
- Methods: how did you compute the deltaP and deltaT values (anomalies)?
- Results: please reword “the present calibration strategy resolved the equifinality problem that is usually encountered while calibrating glacio-hydrological models using only discharge data”; using two data sets does not remove equifinality; you built a single performance metric with an empirical factor to sum up two performance measures and then you report only the best value; it does not mean that there is no equifinality
- Fig. 1: the legend (not the caption) should also include what the dashed lines are
- Fig. 3: the glacier scheme is probably wrong, the text states that there are two linear reservoirs, rainfall is missing
- Fig. 6: what is the y-axis (equation 1 is the calibration equation, something is wrong?)?
- Fig. 7: should be improve, I cannot see much about the circles
- Table 1: is there snowfall occurring in summer and if yes, what is the amount of sommer snowfall? Caption could say how summer is defined (it is in the text though); how high is ET?
- The following reference which is certainly relevant is missing: van Tiel et al : https://hess.copernicus.org/articles/25/3245/2021/; probably their review on glacier modelling is also relevant: https://wires.onlinelibrary.wiley.com/doi/10.1002/wat2.1483
- AC2: 'Reply on RC2', Argha Banerjee, 13 Sep 2022
Peer review completion










Interactive discussion
Status: closed
-
RC1: 'Comment on hess-2022-205', Anonymous Referee #1, 16 Aug 2022
In this paper, the authors quantified the sensitivity of summer runoff to precipitation and temperature changes in two glacierized Himalayan catchments with contrasting climate based on a hydrological model. This study is well prepared, and the results are meaningful. However, there are still some questions needing to be clarified. Here are the details:
- Lines 48-49: This expression should be careful. In my opinion, the simulation is not as same as the observations. The observed data can represent the reality at a point scale but is hard to obtain, especially in the high mountain regions, while the simulation can systematically analyze for a basin-wide scale over a long period.
- Figure 1. The boundaries of the two study basins should be highlighted on the map.
- Lines 65-66: This sentence is inaccurate. As I see, the annual temperature of Chandra (-55â) is lower than Upper Dudhkoshi (-4.7â). In addition, the glacierized fraction of Chandra (0.25) is higher than Upper Dudhkoshi (0.20), and the former glacier area is more than two times the latter. These differences are significant and have a large impact on the glaciohydrology. So please revise it.
- Lines 165-166: How to calculate the glacier area change? There is only the glacier mass balance change data in the supplementary Figure S3. In addition, have the model considered the compensation of snow and transforms into ice?
- Supplementary Figure S1: How to deal with the observed data gaps?
- Line 220-225: I think the temperature before the ablation season can also influence the glacier melt and snowmelt since it controls the distribution of rainfall and snowfall in the accumulation season. Especially in the Chandra basin, where most precipitation occurs in the winter. Thus, I suggest that the authors should add a temperature sensitivity experiment before the ablation season.
- Lines 312-317: How to define the glacier runoff in this paper and what is the difference between it and the glacier ice loss? Moreover, the results show that the glacier runoff contribution to the total summer runoff in upper Duhkoshi is higher than that in Chandra basin while the glacier cover in upper Duhkoshi is lower than that in Chandra and the former summer precipitation is much higher than the latter, which seems contradictory. Please show the other contributions of the summer runoff and clarify this contradiction.
- Lines 351-354: How does the glacier hypsometry affect the mass-balance sensitivity, and how is this factor considered in the model?
- Lines 377-379: In my opinion, the results vary from different studies, I think the authors should discuss the reason for the difference among the studies at different basins rather than describe it as “largely in line with”.
- Figure 7: The color scheme is too blurry to distinguish.
- Line 389: The precipitation increased while the rainfall on glacier did not change, why? Please clarify it.
- Lines 464-466: The RCP2.6 scenario data has been used in this paper, but has no introduction (e.g. which general circulation model has been selected and the evaluation of the projected data).
- AC1: 'Reply on RC1', Argha Banerjee, 13 Sep 2022
- AC3: 'Reply on RC1', Argha Banerjee, 13 Sep 2022
-
RC2: 'Review of hess-2022-205', Anonymous Referee #2, 16 Aug 2022
This is a well written manuscript on a very relevant topic regarding the contribution of streamflow generated in the glacier-covered part of a catchment to catchment-scale water resources. It uses two contrasting glacierised Himalayan catchments, one of which is winter-precipitation dominated, Chandra (the western Himalaya), and the other one summer-precipitation dominated, upper Dudhkoshi (the eastern Himalaya). For these catchments, climate sensitivities of simulated streamflow is obtained by regressing the simulated variability of streamflow to the one its meteorological drivers. The used model is a the Variable Infiltration Capacity (VIC) model, augmented with a glacier melt module.
The analysis is model-based; , the used precipitation-glacier-melt-streamflow model is very simple for the glacier-covered catchment part; as far as I see, it sums up the ice melt and the snowmelt (and rainfall) and routes it through a single (or perhaps two, unclear) linear reservoir, i.e. the corresponding streamflow response has a single time scale stemming from icemelt and snowmelt and no baseflow, thus the model can most likely not simulate a water carry-over effect from month to month for the glacier part. This model structure might have a different impact on the estimated sensitivities for the different analysed catchments. Furthermore, we do not have information on how large the (ignored) debris cover is nor on how important snow redistribution is, we simply know that it is ignored.
Only two parameters of the hydrological model are calibrated, the ones that affect the water balance the most strongly (melt factor for ice and precipitation scaling factor). The calibration is on streamflow and glacier mass balance; there is an empirical weight factor to combine the performance with respect to both quantities; despite a clear lack of giving any formal statistical framework, the parameter estimation approach is called a Bayesian inference.
Accordingly, I am rather skeptical about the added value of this model study; I think that this is essentially a modelling exercise without clear indications that it actually corresponds to how nature reacts; moreover, the conclusion is very general with new insights that can be inferred from general process knowledge such as e.g. the sentence “the temperature sensitivity of the glacier runoff and the precipitation sensitivity of the off-glacier runoff are critical determinants of the future changes of summer runoff and its variability in these two catchments”.
I therefore recommend rejection of this version. The work could become more valuable if it was more critical about the value of the model, if it discussed what we miss with the simplifications and if it provided more insights in what we can learn from the two different types of catchments.
Detailed comment:
- what do you mean by runoff? there are usages of this term where it does not include groundwater-fed baseflow; accordingly: if we mean total flow leaving a catchment, we might want to use streamflow;
- it needs to be very clear also what is meant by “glacier runoff”: runoff generated in the glacier-covered part of the catchment? including baseflow? Including runoff from lateral moraines that are not glacier covered?
- Methods, calibration: I do not think that 5% or 10% error on summer streamflow observations is a realistic value; since summer is the period of high flow, this values is certainly, much higher; furthermore, in the chosen formulation, the error should correspond to the total model error and not just to the observational error (see e.g. an Bayesian inference paper by Dmitri Kavetski); the Bayesian formulation for the mass balance is based on very few obs. values; what assumption do you make about the distribution of the residuals for streamflow and mass balance? i.e. what motivates the chosen form of the likelihood? How did you compute the posterior (you did not use any sampling method that would yield a sample for the posterior; I guess you did some kind of rescaling?)
- Methods, other: i) what is the used temporal time step of the VIC model on the glacier part ? why is it reasonable to keep the bias correction constant in space? I would expect that biases depend on elevation? Why does the glacier melt model not use the energy-balance approach? Is the glacier melt coded by the authors of the study or someone else?
- Methods: the computation of glacier mass balance sensitivity is not clear to me; did you run the model with modified precipitation input?
- Methods: how did you compute the deltaP and deltaT values (anomalies)?
- Results: please reword “the present calibration strategy resolved the equifinality problem that is usually encountered while calibrating glacio-hydrological models using only discharge data”; using two data sets does not remove equifinality; you built a single performance metric with an empirical factor to sum up two performance measures and then you report only the best value; it does not mean that there is no equifinality
- Fig. 1: the legend (not the caption) should also include what the dashed lines are
- Fig. 3: the glacier scheme is probably wrong, the text states that there are two linear reservoirs, rainfall is missing
- Fig. 6: what is the y-axis (equation 1 is the calibration equation, something is wrong?)?
- Fig. 7: should be improve, I cannot see much about the circles
- Table 1: is there snowfall occurring in summer and if yes, what is the amount of sommer snowfall? Caption could say how summer is defined (it is in the text though); how high is ET?
- The following reference which is certainly relevant is missing: van Tiel et al : https://hess.copernicus.org/articles/25/3245/2021/; probably their review on glacier modelling is also relevant: https://wires.onlinelibrary.wiley.com/doi/10.1002/wat2.1483
- AC2: 'Reply on RC2', Argha Banerjee, 13 Sep 2022
Peer review completion










Journal article(s) based on this preprint
Sourav Laha et al.
Sourav Laha et al.
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
501 | 160 | 26 | 687 | 46 | 4 | 7 |
- HTML: 501
- PDF: 160
- XML: 26
- Total: 687
- Supplement: 46
- BibTeX: 4
- EndNote: 7
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1475 KB) - Metadata XML
-
Supplement
(927 KB) - BibTeX
- EndNote