the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 25 Mar 2024
Impacts of glacier changes on precipitation in the Tibetan Plateau
Abstract. The Tibetan Plateau (TP) harbors the largest expanse of glaciers at middle and high latitudes globally. Against the backdrop of ongoing global warming, TP glaciers have experienced widespread retreat and significant mass balance alterations in recent decades, raising questions about their impact on regional climate. In this study, we address this knowledge gap by investigating the magnitude and spatial extent of precipitation responses to glacier changes across the TP using four distinct Weather Research and Forecasting (WRF) simulations reflecting different glacier and climate conditions. Our findings reveal that, on average, mean precipitation (except for winter) tends to diminish by approximately 0.6 % to 2.0 % during a cold year and increases by about 0.2 % to 2.5 % during a warm year over most grid cells influenced by glacier alterations. Additionally, glacier changes lead to a reduction (or augmentation) of summer mean precipitation by an average of 0.6 % to 5.2 % (1.2 % to 10.7 %) over different regions of the TP during the cold (warm) years, accompanied by a notable increase of 0.8 % to 19.7 % in summer extreme precipitation, irrespective of climate conditions. In general, glacier changes exert a more pronounced impact on summer extreme precipitation events than mean precipitation, with an average increase of 1.7 % and 4.6 % over the whole TP during the cold and warm years, respectively. Moreover, glacier changes in warmer climate conditions tend to increase summer precipitation amounts in high-altitude areas when the water supply is adequate.
- Preprint
(7612 KB) - Metadata XML
-
Supplement
(10421 KB) - BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2024-826', Anonymous Referee #1, 03 Apr 2024
In the submitted manuscript, the authors seek to understand how glacier changes in the Tibetan Plateau may have affected precipitation amounts. WRF simulations at a 4km resolution are used, nested within a 12km outer domain which extends well beyond the Tibetan Plateau. Glacier inventories using data collected from the 1960’s to the 1980’s are compared with a glacier inventory derived from data collected between 2004-2011 as a proxy for glacier change. The authors then compare precipitation amounts under a warm year or a cold year for the two glacier scenarios. The study promises an interesting compliment to existing studies of glacier-precipitation feedback over the Tibetan plateau, considering marginal changes to glacier extent instead of absolute retreat, as studied by Ren et al., 2020 and Lin et al., 2021.
When analyzing precipitation, the authors report relative difference in precipitation between the two glacier scenarios. This exaggerates the effect of changing the glacial land surface data, where relative differences for areas receiving little precipitation are sensitive to small changes in absolute precipitation amount. The maps of summer extreme precipitation suffer from the same problem of reporting relative differences though, making it difficult to tell how much changes to glacier extents have impacted precipitation amounts. The authors indeed note that “the relative differences in mean daily precipitation between the two glacier conditions are not statistically significant (p<0.1) for almost all grids”. Section 4.3 is then devoted to explaining these differences. This contradicts the previous finding of the daily-mean (or seasonal total) differences not being significant, and the proposed mechanism of WVF is not supported by the plots shown. Furthermore, the authors focus on areas along the crest of the range, while their methodology only altered glacier extents for Chinese glaciers. This methodological quirk rules out the mechanism proposed by Lin et al., 2021 and Ren et al., 2020, which involved the northerly katabatic flows from glaciers to the south of the crest. Because these glaciers remain unchanged in the study, the mechanism proposed by these two earlier studies cannot be responsible for changes to precipitation. The authors still falsely conclude that their results support the findings of these two prior studies. The manuscript thus shows that changing the land surface type for some cells in the domain results in slight changes to precipitation due to a mechanism which has not been rigorously demonstrated.
For these reasons I recommend rejection of the manuscript in its current form. I believe that the study does robustly show that recent glacial retreat/advance in the Tibetan Plateau has resulted in local changes to the surface energy balance. These local changes may cause localized changes to convection. In a follow up, the authors could focus on demonstrating the mechanisms through which glacial retreat and advance affects convective storms, and quantify this effect. For example, section 4.2.3 references non-local changes to precipitation fields as a result of local changes to glacier extent and thus the surface energy balance. Illustrating particular convective cells and extreme precipitation events which occur as a result of recently retreated glaciers would be interesting to see. The dependency on grid resolution, as it impacts convective processes and the representation of glacier retreat, could thus also be investigated. This, in combination with a clearer reporting of changes to precipitation amounts, would improve the manuscript.
Citation: https://doi.org/10.5194/egusphere-2024-826-RC1 - AC1: 'Reply on RC1', Qian Lin, 06 Jun 2024
-
RC2: 'Comment on egusphere-2024-826', Anonymous Referee #2, 27 Apr 2024
This paper investigates the feedback from glacier changes on precipitation in the Tibetan Plateau (TP) by using four simulations with the Weather Research and Forecasting (WRF) model. The four different simulations represent a “cold” and “warm” year based on regional mean temperature data from 1960 to 2015, with glacier cover data from two periods (1960s-1980s and 2004-2011), respectively. Specifically, the authors model glacier changes by altering the land surface type in WRF to bare ground based on changes in the glacier inventory data. They find that the highest impact of glacier changes can be seen in extreme precipitation events, specifically in the summer, and that the mean precipitation decreases in a cold year but increases in a warm year.
The manuscript has a valuable objective in discussing the impacts of glacier changes on precipitation patterns in this region. However, in its current form, I believe the manuscript has some shortcomings with regard to its methodological setup and the provided explanations in the text. Comments and suggestions are given in the list below.
Major comments:
Evaluation of the modeled precipitation (Section 4.1): The authors are presenting an evaluation between the WRF-modeled precipitation and GPM. However, to assess the WRF model performance, I believe it is crucial to first evaluate the modeled precipitation in ERA5 (serving as initial and boundary conditions for WRF). Subsequently, the authors can evaluate whether downscaling with WRF improves the representation in comparison to ERA5 or not. Could you please add a comparison between ERA5 and GPM and then evaluate how much the WRF model improves the simulation of precipitation (e.g., NRMSE or MRE per basin) for a “cold” and “warm” year.
Evaluation of absolute precipitation amounts: The authors report relative precipitation changes throughout the manuscript, but don’t focus on absolute amounts. This becomes especially important as the authors state that for most grid cells, the results have been statistically insignificant. Can you please specify and discuss absolute amounts?
Modifications of SNOW/ICE grid cells in WRF: The authors mention that they modified grid cells in WRF (SNOW/ICE or bare ground) in order to model glacier changes in Gl1 and Gl2 (Section 3.2.1). However, from the methodological description it is unclear how this modification has been performed. Have you modified the grid cells in both D1 and D2 (i.e., for D1 within the intersection of D1 with D2) or only D2? Page 6 lines 135-136 refer to D2 (I assume). I believe it is crucial to also change the grid cells in the parent domain (D1).
Moreover, the authors only change glacierized grid cells confined to the TP within China, but they do not change grid cells in WRF in other areas (e.g., in all of D1), which is a physically unrealistic setting. Can the authors please explain in detail their reasoning and possible implications of this choice of model setup on the results?
Selection of WRF physics parameterization schemes and albedo representation (Section 3.2.2): The authors mention that they follow the physics parameterizations of Prein et al. (2023). Please provide an overview of their findings and sensitivity tests for the simulation of precipitation and discuss their results for different seasons in order to justify selecting this specific WRF configuration.
Which albedo values are used in the LSM that the authors have chosen (page 6 line 158)? For example, in Noah-MP, the albedo parameterizations for land ice (variable ALBICE in phys/module_sf_noahmp_glacier.F) is set very high and might have to be changed to a value consistent with bare ice for more realistic simulations and atmospheric feedbacks. Please explain those values.
Minor comments:
Page 1 line 11: Please specify in the abstract what you mean with “different glacier and climate conditions” and add a sentence to better describe the methodology. Please also add some information in the abstract about the area and/or numbers of glaciers covered in the simulations.
Page 1 line 19 and page 14 line 362: Can you please specify what you mean with “adequate”?
Page 3 line 75-76: I suggest removing this sentence as it is a repetition.
Page 4 Figure 1: I am unsure of what the 3 pink rectangles are. Please add an explanation in the caption, and add them to the legend. They are also very hard to see. Please use a different color and/or line width. Please also specify which data set you are using here for the elevation and cite it. Figure 1 should be referenced on page 4 line 90 in the text.
Page 4 line 99: Please specify what you mean with “fundamental parameters”. Which parameters from this data set are you using for this study, in addition to the glacier outlines?
Page 5 line 124: I think it would be helpful to add more information on the accuracy of the GPM data set based on previous studies.
Page 5 line 131: Please introduce the WRF domain set-up and resolution before providing detailed information on the land cover change.
Page 5 line 134: Please specify which land cover data you are using as input to WRF and at which resolution. Can you please also mention how exactly you changed the land cover – you are using TP locations based on shapefiles from Gl1 and Gl2. Which threshold for the intersection of the land cover and glacier shapefiles (i.e., Gl1 or Gl2 glacier percentage cover for the land cover grid cells; e.g. more than 50% of land cover grid cells covered by the glacier shapefiles) did you use when changing the categories? Additionally, can you please provide more details on the two different land use types (SNOW/ICE and bare ground) – what are differences in albedo and roughness length in the land use data set? Are there any other differences?
Page 6 line 150: “Grid spacing” and “resolution” refer to two different length scales and should not be used interchangeably (e.g., Grasso, 2000; Stull, 2015). Please use “grid spacing” here and for similar cases throughout the manuscript.
Page 6 line 151: Standard hourly WRF output is given as instantaneous variable values, and not hourly averages. Please clarify this in the text.
Page 6 lines 153-158: Please include all references for the WRF physics options: https://www2.mmm.ucar.edu/wrf/users/physics/phys_references.html#LS
Are you using the Unified Noah LSM? Please clarify and provide the correct reference.Page 6 Section 3.3.2: Please include the WRF time step and map projection.
Page 6 line 180: Are the authors representing and evaluating the inner WRF domain (D2) here (Fig. 2)? Please clarify.
Page 9 Figure 3: It is almost impossible to see the stippled lines in the plot. Please increase the line size. On page 8 line 10 the authors mention that “the relative differences in mean daily precipitation between the two glacier conditions are not statistically significant (p<0.1) for almost all grids”. Can you please elaborate on how you define a whole region to have statistically significant differences?
Page 9 lines 220-222 and page 8 lines 199-201: Repetitions from Section 3.2.3. I suggest rephrasing in Section 3.2.3 (with regard to a reference data set).
Page 9 line 228: Please provide some information on statistical significance here as well.
Figures 5 and 6: The thick red dots (lines) are overwriting the blue dots (lines). Please make sure to have both visible.
Page 14 lines 314-319: Please explain in detail how you perturbed the initial conditions. Which variables were perturbed? By how much?
Page 14 lines 319-320: It is unclear which three regions the authors are talking about – from the text and the Figure 1. Please explain.
Page 14 Section 4.3: Can you name Regions 1, 2, 3 based on, e.g., basins or subregions, instead of using 1, 2, 3?
Page 14 line 329: Analysis of precipitation changes vs. topographic height: Have you analyzed possible correlations? The authors mention some correlations when discussing the water vapor flux for specific regions, but I am wondering whether the authors have conducted a more systematic analysis over the whole study area?
Figures 7, 8, 9: It is very hard to see the pink dots for glacier-changed grid cells. Please increase the size of the dots. What do you mean by “glacier-changed” grid cells? Do you mean both disappeared and advanced glaciers (from Figure 1)? Please specify.
Page 18 line 404: Two typos
Page 17 Discussion: Can you please explain in a few sentences the modeling of precipitation in the specific WRF physics schemes that you used in comparison to other WRF physics schemes, and possible implications for your results.
Page 19 Conclusion: I believe it is important to mention statistical significance of the results here as well.
References:
Grasso, LD. (2000). The Differentiation between grid spacing and resolution and their application to numerical modeling. Bulletin of the American Meteorological Society. 81 (3). 579-580. 10.1175/1520-0477(2000)081<0579:CAA>2.3.CO;2.
Stull, R. B. (2015). Practical meteorology: An algebra-based survey of atmospheric science. Department of Earth, Ocean & Atmospheric Sciences, University of British Columbia, Vancouver, BC. https://doi.org/10.14288/1.0300441.
Citation: https://doi.org/10.5194/egusphere-2024-826-RC2 - AC2: 'Reply on RC2', Qian Lin, 06 Jun 2024
Status: closed
-
RC1: 'Comment on egusphere-2024-826', Anonymous Referee #1, 03 Apr 2024
In the submitted manuscript, the authors seek to understand how glacier changes in the Tibetan Plateau may have affected precipitation amounts. WRF simulations at a 4km resolution are used, nested within a 12km outer domain which extends well beyond the Tibetan Plateau. Glacier inventories using data collected from the 1960’s to the 1980’s are compared with a glacier inventory derived from data collected between 2004-2011 as a proxy for glacier change. The authors then compare precipitation amounts under a warm year or a cold year for the two glacier scenarios. The study promises an interesting compliment to existing studies of glacier-precipitation feedback over the Tibetan plateau, considering marginal changes to glacier extent instead of absolute retreat, as studied by Ren et al., 2020 and Lin et al., 2021.
When analyzing precipitation, the authors report relative difference in precipitation between the two glacier scenarios. This exaggerates the effect of changing the glacial land surface data, where relative differences for areas receiving little precipitation are sensitive to small changes in absolute precipitation amount. The maps of summer extreme precipitation suffer from the same problem of reporting relative differences though, making it difficult to tell how much changes to glacier extents have impacted precipitation amounts. The authors indeed note that “the relative differences in mean daily precipitation between the two glacier conditions are not statistically significant (p<0.1) for almost all grids”. Section 4.3 is then devoted to explaining these differences. This contradicts the previous finding of the daily-mean (or seasonal total) differences not being significant, and the proposed mechanism of WVF is not supported by the plots shown. Furthermore, the authors focus on areas along the crest of the range, while their methodology only altered glacier extents for Chinese glaciers. This methodological quirk rules out the mechanism proposed by Lin et al., 2021 and Ren et al., 2020, which involved the northerly katabatic flows from glaciers to the south of the crest. Because these glaciers remain unchanged in the study, the mechanism proposed by these two earlier studies cannot be responsible for changes to precipitation. The authors still falsely conclude that their results support the findings of these two prior studies. The manuscript thus shows that changing the land surface type for some cells in the domain results in slight changes to precipitation due to a mechanism which has not been rigorously demonstrated.
For these reasons I recommend rejection of the manuscript in its current form. I believe that the study does robustly show that recent glacial retreat/advance in the Tibetan Plateau has resulted in local changes to the surface energy balance. These local changes may cause localized changes to convection. In a follow up, the authors could focus on demonstrating the mechanisms through which glacial retreat and advance affects convective storms, and quantify this effect. For example, section 4.2.3 references non-local changes to precipitation fields as a result of local changes to glacier extent and thus the surface energy balance. Illustrating particular convective cells and extreme precipitation events which occur as a result of recently retreated glaciers would be interesting to see. The dependency on grid resolution, as it impacts convective processes and the representation of glacier retreat, could thus also be investigated. This, in combination with a clearer reporting of changes to precipitation amounts, would improve the manuscript.
Citation: https://doi.org/10.5194/egusphere-2024-826-RC1 - AC1: 'Reply on RC1', Qian Lin, 06 Jun 2024
-
RC2: 'Comment on egusphere-2024-826', Anonymous Referee #2, 27 Apr 2024
This paper investigates the feedback from glacier changes on precipitation in the Tibetan Plateau (TP) by using four simulations with the Weather Research and Forecasting (WRF) model. The four different simulations represent a “cold” and “warm” year based on regional mean temperature data from 1960 to 2015, with glacier cover data from two periods (1960s-1980s and 2004-2011), respectively. Specifically, the authors model glacier changes by altering the land surface type in WRF to bare ground based on changes in the glacier inventory data. They find that the highest impact of glacier changes can be seen in extreme precipitation events, specifically in the summer, and that the mean precipitation decreases in a cold year but increases in a warm year.
The manuscript has a valuable objective in discussing the impacts of glacier changes on precipitation patterns in this region. However, in its current form, I believe the manuscript has some shortcomings with regard to its methodological setup and the provided explanations in the text. Comments and suggestions are given in the list below.
Major comments:
Evaluation of the modeled precipitation (Section 4.1): The authors are presenting an evaluation between the WRF-modeled precipitation and GPM. However, to assess the WRF model performance, I believe it is crucial to first evaluate the modeled precipitation in ERA5 (serving as initial and boundary conditions for WRF). Subsequently, the authors can evaluate whether downscaling with WRF improves the representation in comparison to ERA5 or not. Could you please add a comparison between ERA5 and GPM and then evaluate how much the WRF model improves the simulation of precipitation (e.g., NRMSE or MRE per basin) for a “cold” and “warm” year.
Evaluation of absolute precipitation amounts: The authors report relative precipitation changes throughout the manuscript, but don’t focus on absolute amounts. This becomes especially important as the authors state that for most grid cells, the results have been statistically insignificant. Can you please specify and discuss absolute amounts?
Modifications of SNOW/ICE grid cells in WRF: The authors mention that they modified grid cells in WRF (SNOW/ICE or bare ground) in order to model glacier changes in Gl1 and Gl2 (Section 3.2.1). However, from the methodological description it is unclear how this modification has been performed. Have you modified the grid cells in both D1 and D2 (i.e., for D1 within the intersection of D1 with D2) or only D2? Page 6 lines 135-136 refer to D2 (I assume). I believe it is crucial to also change the grid cells in the parent domain (D1).
Moreover, the authors only change glacierized grid cells confined to the TP within China, but they do not change grid cells in WRF in other areas (e.g., in all of D1), which is a physically unrealistic setting. Can the authors please explain in detail their reasoning and possible implications of this choice of model setup on the results?
Selection of WRF physics parameterization schemes and albedo representation (Section 3.2.2): The authors mention that they follow the physics parameterizations of Prein et al. (2023). Please provide an overview of their findings and sensitivity tests for the simulation of precipitation and discuss their results for different seasons in order to justify selecting this specific WRF configuration.
Which albedo values are used in the LSM that the authors have chosen (page 6 line 158)? For example, in Noah-MP, the albedo parameterizations for land ice (variable ALBICE in phys/module_sf_noahmp_glacier.F) is set very high and might have to be changed to a value consistent with bare ice for more realistic simulations and atmospheric feedbacks. Please explain those values.
Minor comments:
Page 1 line 11: Please specify in the abstract what you mean with “different glacier and climate conditions” and add a sentence to better describe the methodology. Please also add some information in the abstract about the area and/or numbers of glaciers covered in the simulations.
Page 1 line 19 and page 14 line 362: Can you please specify what you mean with “adequate”?
Page 3 line 75-76: I suggest removing this sentence as it is a repetition.
Page 4 Figure 1: I am unsure of what the 3 pink rectangles are. Please add an explanation in the caption, and add them to the legend. They are also very hard to see. Please use a different color and/or line width. Please also specify which data set you are using here for the elevation and cite it. Figure 1 should be referenced on page 4 line 90 in the text.
Page 4 line 99: Please specify what you mean with “fundamental parameters”. Which parameters from this data set are you using for this study, in addition to the glacier outlines?
Page 5 line 124: I think it would be helpful to add more information on the accuracy of the GPM data set based on previous studies.
Page 5 line 131: Please introduce the WRF domain set-up and resolution before providing detailed information on the land cover change.
Page 5 line 134: Please specify which land cover data you are using as input to WRF and at which resolution. Can you please also mention how exactly you changed the land cover – you are using TP locations based on shapefiles from Gl1 and Gl2. Which threshold for the intersection of the land cover and glacier shapefiles (i.e., Gl1 or Gl2 glacier percentage cover for the land cover grid cells; e.g. more than 50% of land cover grid cells covered by the glacier shapefiles) did you use when changing the categories? Additionally, can you please provide more details on the two different land use types (SNOW/ICE and bare ground) – what are differences in albedo and roughness length in the land use data set? Are there any other differences?
Page 6 line 150: “Grid spacing” and “resolution” refer to two different length scales and should not be used interchangeably (e.g., Grasso, 2000; Stull, 2015). Please use “grid spacing” here and for similar cases throughout the manuscript.
Page 6 line 151: Standard hourly WRF output is given as instantaneous variable values, and not hourly averages. Please clarify this in the text.
Page 6 lines 153-158: Please include all references for the WRF physics options: https://www2.mmm.ucar.edu/wrf/users/physics/phys_references.html#LS
Are you using the Unified Noah LSM? Please clarify and provide the correct reference.Page 6 Section 3.3.2: Please include the WRF time step and map projection.
Page 6 line 180: Are the authors representing and evaluating the inner WRF domain (D2) here (Fig. 2)? Please clarify.
Page 9 Figure 3: It is almost impossible to see the stippled lines in the plot. Please increase the line size. On page 8 line 10 the authors mention that “the relative differences in mean daily precipitation between the two glacier conditions are not statistically significant (p<0.1) for almost all grids”. Can you please elaborate on how you define a whole region to have statistically significant differences?
Page 9 lines 220-222 and page 8 lines 199-201: Repetitions from Section 3.2.3. I suggest rephrasing in Section 3.2.3 (with regard to a reference data set).
Page 9 line 228: Please provide some information on statistical significance here as well.
Figures 5 and 6: The thick red dots (lines) are overwriting the blue dots (lines). Please make sure to have both visible.
Page 14 lines 314-319: Please explain in detail how you perturbed the initial conditions. Which variables were perturbed? By how much?
Page 14 lines 319-320: It is unclear which three regions the authors are talking about – from the text and the Figure 1. Please explain.
Page 14 Section 4.3: Can you name Regions 1, 2, 3 based on, e.g., basins or subregions, instead of using 1, 2, 3?
Page 14 line 329: Analysis of precipitation changes vs. topographic height: Have you analyzed possible correlations? The authors mention some correlations when discussing the water vapor flux for specific regions, but I am wondering whether the authors have conducted a more systematic analysis over the whole study area?
Figures 7, 8, 9: It is very hard to see the pink dots for glacier-changed grid cells. Please increase the size of the dots. What do you mean by “glacier-changed” grid cells? Do you mean both disappeared and advanced glaciers (from Figure 1)? Please specify.
Page 18 line 404: Two typos
Page 17 Discussion: Can you please explain in a few sentences the modeling of precipitation in the specific WRF physics schemes that you used in comparison to other WRF physics schemes, and possible implications for your results.
Page 19 Conclusion: I believe it is important to mention statistical significance of the results here as well.
References:
Grasso, LD. (2000). The Differentiation between grid spacing and resolution and their application to numerical modeling. Bulletin of the American Meteorological Society. 81 (3). 579-580. 10.1175/1520-0477(2000)081<0579:CAA>2.3.CO;2.
Stull, R. B. (2015). Practical meteorology: An algebra-based survey of atmospheric science. Department of Earth, Ocean & Atmospheric Sciences, University of British Columbia, Vancouver, BC. https://doi.org/10.14288/1.0300441.
Citation: https://doi.org/10.5194/egusphere-2024-826-RC2 - AC2: 'Reply on RC2', Qian Lin, 06 Jun 2024
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 420 | 118 | 33 | 571 | 35 | 18 | 22 |
- HTML: 420
- PDF: 118
- XML: 33
- Total: 571
- Supplement: 35
- BibTeX: 18
- EndNote: 22
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1