the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 08 Apr 2024
Bridging the gap: a new module for human water use in the Community Earth System Model version 2.2.1
Abstract. Water scarcity is often triggered by shifting climate patterns as well as rising water usage, yet state-of-the-art Earth system models typically do not represent human water demand. Here we present an enhancement to the Community Earth System Model (CESM) and its land (CLM) and river (MOSART) components by introducing sectoral water abstractions. The new module enables a better understanding of water demand and supply dynamics across various sectors, including domestic, livestock, thermoelectric, manufacturing, mining, and irrigation. The module conserves water by integrating abstractions from the land component with river component flows, and dynamically calculates daily water scarcity based on local demand and supply. Through land-only simulations spanning 1971–2010, we verify our model against known water scarcity hotspots, historical global water withdrawal trends, and regional variations in water use. Our findings reaffirm the role of irrigation in modulating local surface energy fluxes, while emphasizing the importance of including all sectors for water scarcity assessment capabilities. While the model captures global patterns, it also discerns regional nuances, expanding on the conventional focus on irrigation withdrawals in Earth system models (ESMs). Despite its advancements, the model's limitations, such as its exclusive focus on river water abstractions, highlight areas for potential future refinement. This research paves the way for a more holistic representation of human-water interactions in ESMs, aiming to inform sustainable water management decisions in an evolving global landscape.
-
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
(40277 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(40277 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-362', Anonymous Referee #1, 01 May 2024
Main comments:
I think the manuscript would be improved by focusing on just the emergent behaviors of the new water sector fluxes, and removing peripheral or redundant information, especially in regards to irrigation, which has been discussed in previous studies.Specifically, I would like to see more details of how the prioritization scheme cascades through the different sectors, especially in regards to water scarcity. Â For instance, figure 11 shows the combined days per year of unmet demand, and it would be informative to see similar plots for each sector, highlighting where the occur and the relative magnitudes of each sector. Â This would be similar to appendix D, but I think showing days per year may be more relatable to the reader than the fractions of unmet demand. Â
I encourage the authors to remove most of section 3.1 and figures 3-5. Â Confirmation of conservation is obviously necessary during model development, but a single sentence could indicate that this was successfully performed. Â In CESM mass/energy conservation violations cause the model to abort, therefore a completed simulation is proof enough. Â That being said, I find the authors comment that the errors in regridding the input data were 1-2% (~line 245) to be worrisome. Â Conservative regridding should be much more accurate than that. Â Similarly, I thought figure 4 could be removed and a simple statement indicating that the prioritization algorithm worked properly would provide equivalent information. Â Given that CLM5 and MOSART have previously been coupled in a conservative manner, one would expect that any flux passed between the two would be treated the same way, and thus figure 5 is unnecessary. Â Discussions of the climate impact of irrigation have previously been published for CLM5, so I would remove this type of discussion, and instead focus on how irrigation is changed by the addition of the other sectors. Â For instance, figure 9 is largely a recapitulation of earlier results; I would prefer to know if there are any locations where the other water sectors have a significant impact. Â In that regard, the authors might wish to include irrigation in the CTRL simulation.
Specific comments:
Abstract:
Line 1: Â natural climate variability should be mentioned in addition to shifting climate patterns, e.g. in places like east Africa, where decadal variability has significant impacts historically.Line 10: what is meant by "captures global patterns"?
Introduction:
Line 48: 'spatial' resolution
Line 57: provide a reference to 'demonstrating higher skill' in CMIPMethods:
Line 89: perhaps state that coupler allows components to have grids with different spatial resolutions
Line 104: what happens if demand is not met?
Line 132: MOSART is not aware of the CLM grid; states and fluxes are sent to the coupler
Line 135: again, models are not aware of each others' grid structures in CESM; overlapping areas in each grid are calculated in the coupler, so this should be re-worded to accurately reflect how the coupling is done.
Line 141: 'idealized' may be more appropriate than 'real-world' given the simple prioritization scheme; could just say 'to diagnose instances of water scarcity'.
Line 143: are the Roman numerals used elsewhere?  If not, why are they used?Line 158: a discussion of the dominant modes of water use for each sector would be useful.  Water used for cooling for thermoelectric and  water for mining purposes might be expected to be returned, but why is return flow so large for livestock?  Does figure C1 indicate that ~1/2 of livestock water is return flow?
Line 212: the discussion of spatial and temporal patterns doesn't seem appropriate here, given that this is a hypothetical case for a single location.
Figure 2: I would find it more natural to indicate the events in temporal order, Â since the prioritization works vertically in this case, i.e. a,d,e,b,c
Line 229: as previously indicated, I would include irrigation in CTRL, as irrigation is currently modeled by CLM5. Â Also, it should perhaps be mentioned that sectors that have large return flow to runoff would not affect climate. ÂLine 266: Section 3.2 begins by discussing trends, but mainly discusses irrigation and river-only vs unlimited simulations. Â The first paragraph could be improved by discussing the statistics in more detail rather than trying to compress all the information into 1 or 2 sentences. Â
Line 313: the word 'prioritize' may need context; Â doesn't everyone prioritize domestic water use? Â Would industrial water use ever be prioritized over domestic water use? Â (I am thinking of domestic water for consumption and sanitation).
Line 317: the wording is vague: does 'trends shifted' mean the location of trends changed, or that trends changed in time?Line 331: is there any reason to think that irrigation would act differently in this simulation? Â I would rather see the irrigation included in the CTRL simulation and a discussion of the interactions between irrigation and the other sectors. Â Otherwise, this is not novel and has been reported previously.
Figure 10: I found it hard to compare the two maps due to the scale. Â Perhaps showing the ratio would be more insightful?
Line 345: in section 3.5, be clear about what statements are model results vs observed conditions. Â Also, explain how individual sectors are impacted at the locations mentioned for instance, is the statement that tourists increase demand reflected in a domestic shortfall or other sectors?
Line 366: some details from the referenced studies would be informative here.
Line 382: is it true that confined aquifers are not supported? Â There is a variable in CLM called 'qflx_gw_con_irrig_col' described as 'confined groundwater irrigation flux'.
Line 399: this approach has been attempted previously:
Anderson et al., 2015, Using satellite-based estimates of evapotranspiration and groundwater changes to determine anthropogenic water fluxes in land surface models, GMD, DOI:10.5194/gmdd-8-3565-2015Line 484: what is meant by 'adeptly captures'; isn't the water use an input to the model? Â The interesting thing is not necessarily which sector dominates based on magnitude, but how each is affected by scarcity. Â
Line 497: 'uniquely' is unlikely to continue, perhaps simply say 'is well positioned'
Citation: https://doi.org/10.5194/egusphere-2024-362-RC1 - AC2: 'Reply on RC1', Ioan Sabin Taranu, 03 Jul 2024
-
RC2: 'Comment on egusphere-2024-362', Anonymous Referee #2, 15 May 2024
The authors attempted to develop a new modeling framework to evaluate the global distribution of water availability. The authors tried to quantify the actual water withdrawal based on CLM5’s grid-scale representation with a combined use of river routing module MOSART. While the paper presents a practical approach to quantify the human-water interactions, which can potentially be useful for water resources allocation and projection, some major issues need to be addressed properly for the publication:
- Major comments:
- Perturbations on water and energy balance due to the new sectoral water consumption model:
While significant variations are expected due to the new sectoral module, no evidence for water and energy conservations is listed in the paper. The authors need to show evidences by comparing the water/energy balance between the CLM version with and without the sectoral water module and are also encouraged to add some more explanation regarding the water and energy coupled balance induced by a sectoral water. For example, given that this is a global-scale study, the authors can show how the streamflow at the coastal area is varied after the implementation of the new module.
- Implication:
Given the almost zero correlation between energy balance variations and water consumption (green dots in figure 9), how the reader should understand the importance of the results? The authors just briefly talked about the role of irrigation (which is not a focus of this paper) but none exists for explaining the relationship between the ‘Sectoral cons’ and energy balance variations. More elaborated and process-based explanation should be added in the revised manuscript to get reader more ideas about the implication of this work. Is their approach converting the consumed water into evaporation still effective with this results?
- Values:
While the values for annual actual withdrawal is presented, I wonder if the authors are really certain about the values. The values ranges up to 700 mm/year. Even for some wet area with annual precipitation around 1,500 mm/y – 2,000 mm/y, if we apply relatively higher runoff ratio such as 0.7, the streamflow could ranges up to 1400 mm/y. I would think this is an extreme case having abundant water resources and even for this case the 700 mm/year consumption of water is the half of the total streamflow. Furthermore, if we sum the (a), (b), (c), ... (f), what is the total consumption of water at the grid-scale? Can you provide the spatial distribution of the total sum of withdrawal?
-Specific comments:
Line 39-40:
Line 100-104: Is the main channel grid cell-scale?
Line 125 mentions that the spatial scales of CLM5 and MOSART are different. From what I read from the paper, the MOSART seems to be a macro grid-scale river routing module relying on the grids defined in running CLM5. So I think there shouldn’t be a scale mismatch between them?
Line 127-130: These sentences need a clarification. Does land unit mean the same with spatial unit? It is mentioned that the sub-grid variability is handled by snow/soil column and PFTs. Are these the same spatial concepts with lat/lon grid cell or watershed?
Line 139: What is VOLR here?
Line 138: Why is the total water storage decided only by the channel water? Can you verify how the total water storage of a grid in CLM5 is estimated? It is need to check if soil water and groundwater are considered in the total water storage.
Line 144: Are there any reasons for this kind of order in allocating water?
Line 169: This sentence needs more clarifications.
Line 205: A sudden change in ‘expected’ or ‘actual’ withdrawal?
Line 229: Is 2 year spin-up enough? Can you show how the steady-state of the model, especially concerning about the new sectoral water module, is confirmed (e.g., criteria)? What (water-related) variables were considered to confirm the model’s steady-state?
Figure 3: How the withdrawal is differ from the consumption? Is it explained before? Are these expected withdrawal or actual? All these information must be clarified. Also, how does this figure explain the sentence (line 245) saying that remapping procedure is found to be conservative?
Figure 4: It is very hard to tell the differences among the plots. Moreover, there is no color bar so the color gradients in the plots do not make any sense.
Line 338-340: This sentence seems to be unnecessary.
While I understand the concept of two-way interactions between the land and river in this model, this is just a simple water balance and, moreover, the watersheds’ hydraulic gradient-driven nature is not explicitly considered. For example, the withdrawal of water from river should have an effect on surface/subsurface runoff. While I am saying this to change the structure of the model, but the authors should indicate the limitation of their two-way method applied in this work.
Citation: https://doi.org/10.5194/egusphere-2024-362-RC2 - AC1: 'Reply on RC2', Ioan Sabin Taranu, 03 Jul 2024
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-362', Anonymous Referee #1, 01 May 2024
Main comments:
I think the manuscript would be improved by focusing on just the emergent behaviors of the new water sector fluxes, and removing peripheral or redundant information, especially in regards to irrigation, which has been discussed in previous studies.Specifically, I would like to see more details of how the prioritization scheme cascades through the different sectors, especially in regards to water scarcity. Â For instance, figure 11 shows the combined days per year of unmet demand, and it would be informative to see similar plots for each sector, highlighting where the occur and the relative magnitudes of each sector. Â This would be similar to appendix D, but I think showing days per year may be more relatable to the reader than the fractions of unmet demand. Â
I encourage the authors to remove most of section 3.1 and figures 3-5. Â Confirmation of conservation is obviously necessary during model development, but a single sentence could indicate that this was successfully performed. Â In CESM mass/energy conservation violations cause the model to abort, therefore a completed simulation is proof enough. Â That being said, I find the authors comment that the errors in regridding the input data were 1-2% (~line 245) to be worrisome. Â Conservative regridding should be much more accurate than that. Â Similarly, I thought figure 4 could be removed and a simple statement indicating that the prioritization algorithm worked properly would provide equivalent information. Â Given that CLM5 and MOSART have previously been coupled in a conservative manner, one would expect that any flux passed between the two would be treated the same way, and thus figure 5 is unnecessary. Â Discussions of the climate impact of irrigation have previously been published for CLM5, so I would remove this type of discussion, and instead focus on how irrigation is changed by the addition of the other sectors. Â For instance, figure 9 is largely a recapitulation of earlier results; I would prefer to know if there are any locations where the other water sectors have a significant impact. Â In that regard, the authors might wish to include irrigation in the CTRL simulation.
Specific comments:
Abstract:
Line 1: Â natural climate variability should be mentioned in addition to shifting climate patterns, e.g. in places like east Africa, where decadal variability has significant impacts historically.Line 10: what is meant by "captures global patterns"?
Introduction:
Line 48: 'spatial' resolution
Line 57: provide a reference to 'demonstrating higher skill' in CMIPMethods:
Line 89: perhaps state that coupler allows components to have grids with different spatial resolutions
Line 104: what happens if demand is not met?
Line 132: MOSART is not aware of the CLM grid; states and fluxes are sent to the coupler
Line 135: again, models are not aware of each others' grid structures in CESM; overlapping areas in each grid are calculated in the coupler, so this should be re-worded to accurately reflect how the coupling is done.
Line 141: 'idealized' may be more appropriate than 'real-world' given the simple prioritization scheme; could just say 'to diagnose instances of water scarcity'.
Line 143: are the Roman numerals used elsewhere?  If not, why are they used?Line 158: a discussion of the dominant modes of water use for each sector would be useful.  Water used for cooling for thermoelectric and  water for mining purposes might be expected to be returned, but why is return flow so large for livestock?  Does figure C1 indicate that ~1/2 of livestock water is return flow?
Line 212: the discussion of spatial and temporal patterns doesn't seem appropriate here, given that this is a hypothetical case for a single location.
Figure 2: I would find it more natural to indicate the events in temporal order, Â since the prioritization works vertically in this case, i.e. a,d,e,b,c
Line 229: as previously indicated, I would include irrigation in CTRL, as irrigation is currently modeled by CLM5. Â Also, it should perhaps be mentioned that sectors that have large return flow to runoff would not affect climate. ÂLine 266: Section 3.2 begins by discussing trends, but mainly discusses irrigation and river-only vs unlimited simulations. Â The first paragraph could be improved by discussing the statistics in more detail rather than trying to compress all the information into 1 or 2 sentences. Â
Line 313: the word 'prioritize' may need context; Â doesn't everyone prioritize domestic water use? Â Would industrial water use ever be prioritized over domestic water use? Â (I am thinking of domestic water for consumption and sanitation).
Line 317: the wording is vague: does 'trends shifted' mean the location of trends changed, or that trends changed in time?Line 331: is there any reason to think that irrigation would act differently in this simulation? Â I would rather see the irrigation included in the CTRL simulation and a discussion of the interactions between irrigation and the other sectors. Â Otherwise, this is not novel and has been reported previously.
Figure 10: I found it hard to compare the two maps due to the scale. Â Perhaps showing the ratio would be more insightful?
Line 345: in section 3.5, be clear about what statements are model results vs observed conditions. Â Also, explain how individual sectors are impacted at the locations mentioned for instance, is the statement that tourists increase demand reflected in a domestic shortfall or other sectors?
Line 366: some details from the referenced studies would be informative here.
Line 382: is it true that confined aquifers are not supported? Â There is a variable in CLM called 'qflx_gw_con_irrig_col' described as 'confined groundwater irrigation flux'.
Line 399: this approach has been attempted previously:
Anderson et al., 2015, Using satellite-based estimates of evapotranspiration and groundwater changes to determine anthropogenic water fluxes in land surface models, GMD, DOI:10.5194/gmdd-8-3565-2015Line 484: what is meant by 'adeptly captures'; isn't the water use an input to the model? Â The interesting thing is not necessarily which sector dominates based on magnitude, but how each is affected by scarcity. Â
Line 497: 'uniquely' is unlikely to continue, perhaps simply say 'is well positioned'
Citation: https://doi.org/10.5194/egusphere-2024-362-RC1 - AC2: 'Reply on RC1', Ioan Sabin Taranu, 03 Jul 2024
-
RC2: 'Comment on egusphere-2024-362', Anonymous Referee #2, 15 May 2024
The authors attempted to develop a new modeling framework to evaluate the global distribution of water availability. The authors tried to quantify the actual water withdrawal based on CLM5’s grid-scale representation with a combined use of river routing module MOSART. While the paper presents a practical approach to quantify the human-water interactions, which can potentially be useful for water resources allocation and projection, some major issues need to be addressed properly for the publication:
- Major comments:
- Perturbations on water and energy balance due to the new sectoral water consumption model:
While significant variations are expected due to the new sectoral module, no evidence for water and energy conservations is listed in the paper. The authors need to show evidences by comparing the water/energy balance between the CLM version with and without the sectoral water module and are also encouraged to add some more explanation regarding the water and energy coupled balance induced by a sectoral water. For example, given that this is a global-scale study, the authors can show how the streamflow at the coastal area is varied after the implementation of the new module.
- Implication:
Given the almost zero correlation between energy balance variations and water consumption (green dots in figure 9), how the reader should understand the importance of the results? The authors just briefly talked about the role of irrigation (which is not a focus of this paper) but none exists for explaining the relationship between the ‘Sectoral cons’ and energy balance variations. More elaborated and process-based explanation should be added in the revised manuscript to get reader more ideas about the implication of this work. Is their approach converting the consumed water into evaporation still effective with this results?
- Values:
While the values for annual actual withdrawal is presented, I wonder if the authors are really certain about the values. The values ranges up to 700 mm/year. Even for some wet area with annual precipitation around 1,500 mm/y – 2,000 mm/y, if we apply relatively higher runoff ratio such as 0.7, the streamflow could ranges up to 1400 mm/y. I would think this is an extreme case having abundant water resources and even for this case the 700 mm/year consumption of water is the half of the total streamflow. Furthermore, if we sum the (a), (b), (c), ... (f), what is the total consumption of water at the grid-scale? Can you provide the spatial distribution of the total sum of withdrawal?
-Specific comments:
Line 39-40:
Line 100-104: Is the main channel grid cell-scale?
Line 125 mentions that the spatial scales of CLM5 and MOSART are different. From what I read from the paper, the MOSART seems to be a macro grid-scale river routing module relying on the grids defined in running CLM5. So I think there shouldn’t be a scale mismatch between them?
Line 127-130: These sentences need a clarification. Does land unit mean the same with spatial unit? It is mentioned that the sub-grid variability is handled by snow/soil column and PFTs. Are these the same spatial concepts with lat/lon grid cell or watershed?
Line 139: What is VOLR here?
Line 138: Why is the total water storage decided only by the channel water? Can you verify how the total water storage of a grid in CLM5 is estimated? It is need to check if soil water and groundwater are considered in the total water storage.
Line 144: Are there any reasons for this kind of order in allocating water?
Line 169: This sentence needs more clarifications.
Line 205: A sudden change in ‘expected’ or ‘actual’ withdrawal?
Line 229: Is 2 year spin-up enough? Can you show how the steady-state of the model, especially concerning about the new sectoral water module, is confirmed (e.g., criteria)? What (water-related) variables were considered to confirm the model’s steady-state?
Figure 3: How the withdrawal is differ from the consumption? Is it explained before? Are these expected withdrawal or actual? All these information must be clarified. Also, how does this figure explain the sentence (line 245) saying that remapping procedure is found to be conservative?
Figure 4: It is very hard to tell the differences among the plots. Moreover, there is no color bar so the color gradients in the plots do not make any sense.
Line 338-340: This sentence seems to be unnecessary.
While I understand the concept of two-way interactions between the land and river in this model, this is just a simple water balance and, moreover, the watersheds’ hydraulic gradient-driven nature is not explicitly considered. For example, the withdrawal of water from river should have an effect on surface/subsurface runoff. While I am saying this to change the structure of the model, but the authors should indicate the limitation of their two-way method applied in this work.
Citation: https://doi.org/10.5194/egusphere-2024-362-RC2 - AC1: 'Reply on RC2', Ioan Sabin Taranu, 03 Jul 2024
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 503 | 138 | 36 | 677 | 21 | 34 |
- HTML: 503
- PDF: 138
- XML: 36
- Total: 677
- BibTeX: 21
- EndNote: 34
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
David M. Lawrence
Yoshihide Wada
Ting Tang
Erik Kluzek
Sam Rabin
Yi Yao
Steven J. De Hertog
Inne Vanderkelen
Wim Thiery
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(40277 KB) - Metadata XML