Spatiotemporal responses of runoff to climate change on the southern Tibetan Plateau

Sun, He; Yao, Tandong; Su, Fengge; Yang, Wei; Chen, Deliang

doi:https://doi.org/10.5194/hess-2024-11

Preprints

https://doi.org/10.5194/hess-2024-11

Preprints

23 Jan 2024

| 23 Jan 2024

Status: a revised version of this preprint is currently under review for the journal HESS.

Spatiotemporal responses of runoff to climate change on the southern Tibetan Plateau

He Sun, Tandong Yao, Fengge Su, Wei Yang, and Deliang Chen

Abstract. A comprehensive understanding of spatiotemporal runoff changes at a sub-basin scale of the Yarlung Zangbo (YZ) basin on the southern Tibetan Plateau (TP), amidst varying climatic and cryospheric conditions, is imperative for effective water resources management. However, spatiotemporal differences of runoff composition, change and the attribution within the YZ basin have not been extensively explored, primarily due to the lack of hydrometeorological observations, especially in the downstream region. In this study, we investigated historical and future evolution of annual and seasonal total water availability, as well as glacier runoff and snowmelt contributions across six sub-basins of the YZ with a particular focus on the comparison between the upstream Nuxia (NX) basin and the downstream Nuxia-Pasighat (NX-BXK) basin, based on a newly generated precipitation dataset and a well-validated model with streamflow, glacier mass, and snow cover observations. Our findings revealed large spatiotemporal differences in changes exist within the YZ basin for 1971–2020. Firstly, runoff generation was dominated by rainfall runoff throughout the YZ basin, with glacier runoff playing more important role in the annual total runoff (19 %) in the NX-BXK sub-basin compared to other sub-basins. Notably, glacier runoff contributed 52 % of the total runoff at the Pasighat outlet of the YZ basin. Secondly, annual runoff exhibited an increasing trend in the NX basin but a decreasing trend in the NX-BXK, primarily attributed to rainfall runoff changes influenced by atmospheric moisture. Glacier runoff enhanced water supply, by offsetting the decreasing contribution from rainfall. Total runoff will consistently increase (27–100 mm/10 yr) across the sub-basins through the 21st century, resulting from increased rainfall runoff and a minor effect of increased snowmelt and glacier runoff.

Received: 11 Jan 2024 – Discussion started: 23 Jan 2024

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Preprint (PDF, 2560 KB)

Supplement (2743 KB)

Download & links

He Sun, Tandong Yao, Fengge Su, Wei Yang, and Deliang Chen

Status: final response (author comments only)

CC1:
'Comment on hess-2024-11', Sivarajah Mylevaganam, 15 Feb 2024
The concept of accounting has been floating around in the business world for many decades. This has been presented in the form of income statements, balance sheets, and cash flow statements to bolster the business operations of many listed and non-listed entities around the world. Considering its potential to resolve many issues, this fascinating concept has been embraced by many disciplines. Water science is one of them, which echoes this concept in the forms of water balance and water availability to address many demanding issues such as upstream/downstream conflicts and interbasin transfers.
In this manuscript, the authors employ an integrated modeling environment to evaluate the runoff composition in the Yarlung Zangbo river basin on the southern Tibetan Plateau, which nourishes around 2 billion human lives.
Most of the equations that are presented in the manuscript need to be rewritten. For example, eq. 3 should be written using a summation notation (see the attached file).The subscripts are confusing (compare eq. 2 and eq. 3). To reflect the band of interest and the grid of interest, the dependent variable should have two subscripts (i.e., i and j).

The unit of the dependent variable in eq.2 is incorrect. Is it mm/day or mm?

The definition of “f” in eq.2 is incorrect. Are you referring to the proportion of glacier area in a particular grid whose total runoff is being calculated?

The variables in eq.4 need units to understand your calculations.

Refer to Part II
Citation: https://doi.org/10.5194/hess-2024-11-CC1
- AC1: 'Reply on CC1', He Sun, 20 May 2024
  
  Thanks for your descriptive review and for highlighting the importance of our study. Your valuable comments will surely improve the quality of our manuscript. Please see below the item-to-item responses to the comments and questions in the attached reply.
  
  Citation: https://doi.org/10.5194/hess-2024-11-AC1
CC2:
'Comment on hess-2024-11', Sivarajah Mylevaganam, 16 Feb 2024

Sivarajah Mylevaganam

Alumnus, Spatial Sciences Laboratory, Texas A&M University, College Station, USA.
1. Line 164-166
As per the authors, there are around 280 precipitation gages. The spatial extent of the river basin (i.e., YZ) is around 250,000 km². The spatial resolution of the modeling task is 10km. Therefore, the reason for using a machine-learning algorithm to develop a precipitation grid is not understood. Wouldn’t the popular interpolation algorithms that are packaged with GIS products (e.g.,Esri’s ArcGIS) improve the simulation results?
2. Is your DDF (see eq.2) a constant for a particular pixel/grid (10km in spatial extent)? Is your T a constant for a particular pixel/grid? This is what has been understood from your eq.2 and eq.3.If these values are constants for a particular pixel/grid, the value of your Rglac is meaningless. Without considering the temperature profile across your elevation bands, does the value of Rglac computed using eq.2 and eq.3 make sense? Without considering the vertical profile of DDF across your elevation bands, does the value of Rglac make sense? See the attached PDF file.
3. For a particular grid/pixel (10km in spatial extent) of your interest, would you be able to show the values of your Ms (see eq.3)? Is the value of your “n” constant for the study area (i.e., YZ)?
4. Refer to Part III

Citation: https://doi.org/10.5194/hess-2024-11-CC2
- AC2: 'Reply on CC2', He Sun, 20 May 2024
  
  Thanks for your suggestions and questions.
  1.Line 164-166
  As per the authors, there are around 280 precipitation gages. The spatial extent of the river basin (i.e., YZ) is around 250,000 km2. The spatial resolution of the modeling task is 10km. Therefore, the reason for using a machine-learning algorithm to develop a precipitation grid is not understood. Wouldn’t the popular interpolation algorithms that are packaged with GIS products (e.g.,Esri’s ArcGIS) improve the simulation results?
  Reply:
  Thanks for the comments. The gridded data obtained by interpolating observations at low altitudes tend to largely underestimate basin real precipitation due to the strong precipitation gradient created by orographic enhancements and windinduced undercatch of solid precipitation (Sun and Su, 2020). Sun and Su (2020) also compared different interpolation algorithms in the basin, and suggested that a single correction approach may not fit the entire basin. Therefore, daily precipitation with a spatial resolution of 10×10 km was used as the VIC-Glacier model forcing inputs in this work. Historical meteorological data during 1971–2100 was adopted from Sun et al. (2022), which corrected these data by machine learning algorithm based on estimates form meteorological stations and the ERA5. The daily gridded precipitation estimates from the ERA5 was corrected based on 580 rain gauges in the monsoon-dominated TP region (290 rain gauges in the YZ basin, Figure 1), and was inversely evaluated by the VIC-Glacier model, suggesting the good potential utility in model simulation in the YZ basin (Sun et al., 2022).
  References
  Sun, H. Su, F., 2020. Precipitation correction and reconstruction for streamflow simulation based on 262 rain gauges in the upper Brahmaputra of southern Tibetan Plateau. J. Hydrol. 590. https://doi.org/10.1016/j.jhydrol.2020.125484.
  Sun, H., Yao, T., Su, F., He, Z., Tang, G., Li, N., et al., 2022. Corrected ERA5 precipitation by machine learning significantly improved flow simulations for the Third Pole basins. J. Hydrometeorol. 23. https://doi.org/10.1175/JHM-D-22-0015.1.
  
  2. Is your DDF (see eq.2) a constant for a particular pixel/grid (10km in spatial extent)? Is your T a constant for a particular pixel/grid? This is what has been understood from your eq.2 and eq.3.If these values are constants for a particular pixel/grid, the value of your Rglac is meaningless. Without considering the temperature profile across your elevation bands, does the value of Rglac computed using eq.2 and eq.3 make sense? Without considering the vertical profile of DDF across your elevation bands, does the value of Rglac make sense? See the attached PDF file.
  Reply:
  Yes, the DDF is a constant for entire YZ basin, mostly due to the limited direct observations in the glacier area, with the high elevation, complex terrain and inaccessibility. T (°C) is the daily average air temperature above the glacier surface, and it is a daily series in each grid. And in this study, each glacierized grid cell underwent division into various elevation bands with an interval of 100 m, and the T is different in different bands with the temperature lapse rate in each grid.
  3. For a particular grid/pixel (10km in spatial extent) of your interest, would you be able to show the values of your Ms (see eq.3)? Is the value of your “n” constant for the study area (i.e., YZ)?
  Reply:
  Yes, we can show the values of meltwater in each grid, and the “n” is constant for the study area.
  
  Citation: https://doi.org/10.5194/hess-2024-11-AC2
CC3:
'Comment on hess-2024-11', Sivarajah Mylevaganam, 17 Feb 2024
Sivarajah Mylevaganam

Alumnus, Spatial Sciences Laboratory, Texas A&M University, College Station, USA.
As per the authors, the temporal resolution of the modeling work was 3 h (line 270-271). Moreover, as per the authors, the temporal resolution of the precipitation dataset was 24 h (line 165-166). The conversion algorithm from 24 h to 3 h is not found in the current version of the manuscript. How was this conversion carried out in the integrated modeling environment? The computation of R_glac(eq.2 and eq.3) contradicts line 270-271.

As per eq.1 (line 278-282), R_vic is the runoff (surface+baseflow) computed by a model named Variable Infiltration Capacity (VIC). In other words, given the rainfall and the other defining parameters, the model generates the runoff for each pixel/grid (spatial resolution of 10km, see the attached pdf file). As per eq.1, the authors consider only the portion of the runoff (i.e., (1-f)R_vic) generated in a grid/pixel for the rainfall value given for that pixel/grid. What has happened to the other component (i.e., f*Rvic)? As per eq.2, the parameter DDF (i.e., the degree-day factor of glacier or snow melt, see line 284-285) doesn’t account for this component (i.e., f*Rvic).

Refer to Part IV

Acknowledgement and Disclaimer

The author is an alumnus of Texas A&M University, Texas, USA. The views expressed here are solely those of the author in his private capacity and do not in any way represent the views of Texas A&M University, Texas, USA.
Citation: https://doi.org/10.5194/hess-2024-11-CC3
- AC3: 'Reply on CC3', He Sun, 20 May 2024
  
  Thanks for your descriptive review and for highlighting the importance of our study. Your valuable comments will surely improve the quality of our manuscript. Please see below the item-to-item responses to the comments and questions in the attached reply.
  
  Citation: https://doi.org/10.5194/hess-2024-11-AC3
CC4:
'Comment on hess-2024-11', Sivarajah Mylevaganam, 19 Feb 2024
Sivarajah Mylevaganam

Alumnus, Spatial Sciences Laboratory, Texas A&M University, College Station, USA.
As per the authors, a lumped concept has been implemented to calculate the runoff from the glacier (see line 282-292).In other words, a calibration parameter named DDF has been used to calculate the runoff from the glacier (see line 282-292).Moreover, as per the authors, a correlation (an exponential curve) between the glacier volume and glacier area (see eq.4) has been used to compute the glacier volume given the glacier area. The initial glacier area is obtained from the first Glacier Inventory of China (CGI V1.0; see line 299-300).In the current version of the manuscript, the actual pipeline, if there exists one, between the DDF and the glacier volume(S, see eq.4) is not found. What is the relationship between DDF and S? Isn’t the glacier volume melted to get R_glac?

I am trying to understand the relationship expressed in eq.2. Assume that the value of DDF is 10 units. I guess this value is reasonable considering the calibrated value that has been presented in the manuscript (see line 338-339). Moreover, assume that the number of elevation bands (n) is set to 1. If the air temperature is 10 units, as per eq.2 and eq.3, the calculated value of R_glac will be 100 units. Similarly, if the air temperature is 20 units, as per eq.2 and eq.3, the calculated value of R_glac will be 200 units. Do the values govern the underlying principles? Since the value of T_base is set to 0, basically, eq.2 is a linear relationship that goes through the origin. The gradient of the line is the defining parameter (i.e., DDF). What is the physical meaning of DDF? Is it the glacier volume that could be melted given the temperature values?

Refer to Part V

Acknowledgement and Disclaimer
The author is an alumnus of Texas A&M University, Texas, USA. The views expressed here are solely those of the author in his private capacity and do not in any way represent the views of Texas A&M University, Texas, USA.
Citation: https://doi.org/10.5194/hess-2024-11-CC4
- AC4: 'Reply on CC4', He Sun, 20 May 2024
  
  1. As per the authors, a lumped concept has been implemented to calculate the runoff from the glacier (see line 282-292). In other words, a calibration parameter named DDF has been used to calculate the runoff from the glacier (see line 282-292). Moreover, as per the authors, a correlation (an exponential curve) between the glacier volume and glacier area (see eq.4) has been used to compute the glacier volume given the glacier area. The initial glacier area is obtained from the first Glacier Inventory of China (CGI V1.0; see line 299-300). In the current version of the manuscript, the actual pipeline, if there exists one, between the DDF and the glacier volume(S, see eq.4) is not found. What is the relationship between DDF and S? Isn’t the glacier volume melted to get Rglac?
  Reply:
  The DDF was used to calculate glacier melt (glacier volume, V), and the calculated glacier area (S) was updated every year in the model by the volume-area (V-S) scaling approach.
  
  2. I am trying to understand the relationship expressed in eq.2. Assume that the value of DDF is 10 units. I guess this value is reasonable considering the calibrated value that has been presented in the manuscript (see line 338-339). Moreover, assume that the number of elevation bands (n) is set to 1. If the air temperature is 10 units, as per eq.2 and eq.3, the calculated value of Rglac will be 100 units. Similarly, if the air temperature is 20 units, as per eq.2 and eq.3, the calculated value of Rglac will be 200 units. Do the values govern the underlying principles? Since the value of Tbase is set to 0, basically, eq.2 is a linear relationship that goes through the origin. The gradient of the line is the defining parameter (i.e., DDF). What is the physical meaning of DDF? Is it the glacier volume that could be melted given the temperature values?
  Reply:
  Sorry for the confusion. T (°C) is the daily average air temperature above the glacier surface, and it is a daily series in each grid. It is not a constant for each grid. It will be positive in summer and autumn, and negative in other seasons. When daily T is above 0°C, the glacier will melt; otherwise, it was accumulated. In addition, precipitation will also influence the accumulation in glacier area. Therefore, it is not a linear relationship.
  
  Citation: https://doi.org/10.5194/hess-2024-11-AC4
RC1:
'Comment on hess-2024-11', Anonymous Referee #1, 25 Mar 2024

This resubmitted version of the paper has addressed my previous concerns effectively, demonstrating the authors' great efforts. The manuscript now provides a clear explanation of the novel motivation behind the work, and the research results are presented in an informative manner with well-crafted plots. The first comparisons of sub-basin runoff changes in the YZ river are particularly valuable, as they contribute to a better understanding of runoff changes at the downstream basin outlet, which will likely be of great interest to other researchers.

Before accepting this version, I have three minor suggestions:

1. Including additional statements on the calculation of runoff composition contributions in sub-basins would enhance reader understanding of differences among the sub-basins. For instance, providing formulas such as rainfall contribution = rainfall in the sub-basin / (rainfall + snowmelt + glacier melt) generated in the sub-basin area could clarify these calculations.

2. It would be beneficial to include a table summarizing the model calibration and performance to provide a more straightforward description of the calibration procedure. This table could include details such as the calibration step, model parameters calibrated in each step, data used for evaluation, objective function, and performance metrics.

3. Adding an additional discussion section to explore the underlying reasons for the different runoff change trends (both historical and future) in the sub-basins would enrich the results. In this section, quantitative comparisons of changes in total precipitation, temperature, snow fraction in precipitation, evapotranspiration, and glacier mass among the sub-basins could be included to provide more insight into the observed runoff change trends.

Citation: https://doi.org/10.5194/hess-2024-11-RC1
- AC5: 'Reply on RC1', He Sun, 20 May 2024
  
  Dear reviewers:
  On behalf of my co-authors, thank you very much for your attention to our paper “Spatiotemporal responses of runoff to climate change on the southern Tibetan Plateau”. My co-authors and I think that all the comments are valuable and very helpful for us to improve the manuscript. We have carefully revised the manuscript according to your comments.
  Please see below the item-to-item responses to the reviewers’ comments and questions in the attached reply. We hope this manuscript will satisfy the requirement of the Hydrology and Earth System Sciences.
  
  Citation: https://doi.org/10.5194/hess-2024-11-AC5
RC2:
'Comment on hess-2024-11', Anonymous Referee #2, 07 Jun 2024
This manuscript investigates the spatiotemporal responses of runoff to climate change across six sub-basins of the Yarlung Zangbo (YZ) river basin, with a particular focus on differences between the upstream Nuxia and the downstream Nuxia-Pasighat basin. The manuscript is well-written and is of interest to Hydrology and Earth System Sciences. However, some improvements are necessary before publication.
The authors focus on six sub-basins of the YZ basin. Therefore, they should explain the changes in runoff and their possible causes. Additionally, it would be beneficial to elucidate why there is a negative trend in the RKZ sub-basin.

The authors provide a table showing the parameters and performance of the VIC-Glacier model during the calibration and validation periods. It is recommended that they provide more details about the observed data used for each step.

The authors could expand the discussion on hydrologic modeling, considering the glacier melt component, in other high mountainous basins. They should also explore the possible reasons for variations in glacier contribution within the same basin.
Citation: https://doi.org/10.5194/hess-2024-11-RC2
- AC6: 'Reply on RC2', He Sun, 23 Jun 2024
  
  Dear reviewers:
  On behalf of my co-authors, thank you very much for your attention to our paper “Spatiotemporal responses of runoff to climate change on the southern Tibetan Plateau”. My co-authors and I think that all the comments are valuable and very helpful for us to improve the manuscript. We have carefully revised the manuscript according to your comments.
  Please see below the item-to-item responses to the reviewers’ comments and questions in the attached reply. We hope this manuscript will satisfy the requirement of the Hydrology and Earth System Sciences.
  
  Citation: https://doi.org/10.5194/hess-2024-11-AC6
RC3:
'Comment on hess-2024-11', Anonymous Referee #3, 19 Jun 2024
The study uses the VIC-Glacier hydrological model to examine historical and future runoff changes across six sub-basins, highlighting significant differences in rainfall, snowmelt, and glacier runoff contributions. The findings provide critical insights for water resource management and adaptation strategies in this important region. The manuscript is well-organized and provides a comprehensive analysis of research question. I believe the manuscript is suitable for publication on HESS after some minor revisions.
The manuscript will benefit from a more detailed introduction to the forcing data and the accuracy of the datasets from the authors’ previous studies. Especially, how snowfall is estimated and whether the undercatch is corrected.

I suggest adding more discussions about the generalizability of the methods, results, and conclusions in this study.

Please discuss the potential impact of land cover and land use change on the conclusions in this study

Section 5.2 provides very general implications. I wonder whether this part is necessary since the water management recommendations are just loosely and conceptually linked to the findings in this study.

Please add more details in the “Data availability” section.
Citation: https://doi.org/10.5194/hess-2024-11-RC3
- AC7: 'Reply on RC3', He Sun, 23 Jun 2024
  
  Thanks for your descriptive review and for highlighting the importance of our study. Your valuable comments will surely improve the quality of our manuscript. Please see below the item-to-item responses to the comments and questions in the attached reply.
  
  Citation: https://doi.org/10.5194/hess-2024-11-AC7

He Sun, Tandong Yao, Fengge Su, Wei Yang, and Deliang Chen

Supplement

https://doi.org/10.5194/hess-2024-11-supplement

He Sun, Tandong Yao, Fengge Su, Wei Yang, and Deliang Chen

Viewed

Total article views: 663 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
464	144	55	663	43	27	22

HTML: 464
PDF: 144
XML: 55
Total: 663
Supplement: 43
BibTeX: 27
EndNote: 22

Views and downloads (calculated since 23 Jan 2024)

Month	HTML	PDF	XML	Total
Jan 2024	130	38	3	171
Feb 2024	85	27	11	123
Mar 2024	65	23	6	94
Apr 2024	35	14	9	58
May 2024	91	25	14	130
Jun 2024	58	17	12	87

Cumulative views and downloads (calculated since 23 Jan 2024)

Month	HTML	PDF	XML	Total
Jan 2024	130	38	3	171
Feb 2024	85	27	11	123
Mar 2024	65	23	6	94
Apr 2024	35	14	9	58
May 2024	91	25	14	130
Jun 2024	58	17	12	87

Viewed (geographical distribution)

Total article views: 636 (including HTML, PDF, and XML) Thereof 636 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 29 Jun 2024

Short summary

Our findings revealed runoff generation is dominated by rainfall runoff in the YZ, and the largest glacier runoff contribution is in the downstream sub-basin. Annual runoff trends indicate an increase in the NX but a decrease in the NX-BXK for 1971–2020, due to contrasting precipitation changes. Total runoff across the sub-basins will consistently increase through the 21st century, mostly resulting from increased rainfall runoff.


Total:	0
HTML:	0
PDF:	0
XML:	0