the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
GCAM-GLORY v1.0: Representing Global Reservoir Water Storage in a Multisector Human-Earth System Model
Abstract. Reservoirs play a significant role in modifying the spatiotemporal availability of surface water to meet multi-sector human demands, despite representing a relatively small fraction of the global water budget. Yet the integrated modeling frameworks that explore the interactions among climate, land, energy, water, and socioeconomic systems at a global scale often contain limited representations of water storage dynamics that incorporate feedbacks from other systems. In this study, we implement a representation of water storage in the Global Change Analysis Model (GCAM) to enable exploration of the future role (e.g., expansion) of reservoir water storage globally in meeting demands for, and evolving in response to interactions with, the climate, land, and energy systems. GCAM represents 235 global water basins, operates at 5-year time steps, and uses supply curves to capture economic competition among renewable water (now including reservoirs), non-renewable groundwater, and desalination. Our approach consists of developing the GLObal Reservoir Yield (GLORY) model, which uses a Linear Programming (LP)-based optimization algorithm, and dynamically linking GLORY with GCAM. The new coupled GCAM-GLORY approach improves the representation of reservoir water storage in GCAM in several ways. First, the GLORY model identifies the cost to supply increasing levels of water supply from reservoir storage by considering regional physical and economic factors, such as evolving monthly reservoir inflows and demands, and the levelized cost to construct additional reservoir storage capacity. Second, by passing those costs to GCAM, GLORY enables exploring future regional reservoir expansion pathways and their response to climate and socioeconomic drivers. To guide the model toward reasonable reservoir expansion pathways, GLORY applies a diverse array of feasibility constraints related to protected land, population, water sources, and cropland. Finally, the GLORY-GCAM feedback loop allows evolving water demands from GCAM to inform GLORY, resulting in an updated supply curve at each time step, thus enabling GCAM to establish a more meaningful economic value of water. This study improves our understanding of the sensitivity of reservoir water supply to multiple physical and economic dimensions, such as sub-annual variations in climate conditions and human water demands, especially for basins experiencing socioeconomic droughts.
- Preprint
(4448 KB) - Metadata XML
-
Supplement
(2786 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on gmd-2023-204', Anonymous Referee #1, 12 Feb 2024
The authors Zhao et al submitted a manuscript to GMD entitled with “GCAM-GLORY v1.0: Representing Global Reservoir Water Storage in a Multisector Human-Earth System Model”. They intend to improve the representation of GCAM by developing GLObal Reservoir Yield (GLORY) and coupling to GCAM to enable inclusion of a feedback loop and update of the mechanisms of the reservoir module, specifically the water supply potential of reservoirs for satisfying water use demand. The manuscript is logically structured (esp. in Sect. 2) and well written. The authors clearly describe the rationale of the many decisions that have to be made for developing and running GLORY and coupling it with GCAM. Sometimes this reads by nature a bit lengthy but in fact, this level of detailed description is necessary to understand the structure of the model components. The authors demonstrate the application of the model with 4 scenarios to show the effect of the described model development. Those results are also well described, both by figures (e.g. Fig. 7) and the related text (e.g. around L720). The results show the difference to the original, static reservoir representation. Interpretation and scientific evaluation of the results (e.g. relation to other models) is not focus of the manuscript which is fine for the given manuscript type in GMD that focus on model description. So, I think the authors did a very good job and I have only a few remarks which could make the manuscript even more readable / broaden readership.
Congratulations to the provided meta-repository on GitHub which I enjoyed exploring.
Major
- L 96 Agree, but for readers of communities that are not so familiar with MSD models, it would be really good to read a broad review about the principles and mechanisms of MSD in general, and also the difference to other approaches like GHMs. The authors relate to a nice tabular overview in the Supplement when it comes to the reservoir representation but I have the impression that such a table alone is not too informative and I would wish to see some additional explanatory text. I think both (overview and differences) could be very valuable to embed your work and broaden up readership. It can certainly be a paragraph in the Supplement, if text space in the main manuscript is the limiting point.
- Sometimes (e.g. the overview section within Sect. 2.1) I read several aspects of review, motivation and application possibilities which I would not necessarily see well fitting in a methods section. Therefore, I would suggest to concentrate to methodological description in the methods section and put other aspects to other parts of the manuscript. Specificly, I was a bit surprised to read in around L 150 two (additional) research questions; would have loved to read those in the Intro section. That could help to streamline the manuscript and avoid repetition. On the other hand, I really enjoy reading Sect. 2.2.1. So, it is a bit a question of the right amount of overview (nicely done in 2.2.1) or going sometimes a bit beyond (e.g. 2.1.1) and of course it is subjective to find the right balance. I nevertheless would suggest to go through and try to improve the balance of the components.
- In some sections, e.g. 2.3.4 it is not always clear which time period is meant. E.g. Fig 4 vaguely expresses “historical SEDI levels” – but it would be good to have a statement in the figure caption which time period is meant. Also, as some of the input data are depending on Xanthos input it would be good to read which climate input the GHM is using (specifically to generate the diagrams in Fig. 4). This certainly would help the reader to digest such diagrams.
Technical comments
- L 46 Text in brackets should be within the previous sentence
- L510 GranD should be read as GRanD
- L 256 I cannot find the two references (with the indicated year) in the reference list. Please check carefully the coherence of references within the manuscript and the reference list.
- Fig 8: Please write in the figure caption the thick black line.
- Fig 9: suggest to extend the figure caption to better grasp the content solely from figure and figure caption
Citation: https://doi.org/10.5194/gmd-2023-204-RC1 - AC1: 'Reply on RC1', Mengqi Zhao, 03 Apr 2024
-
RC2: 'Comment on gmd-2023-204', Anonymous Referee #2, 20 Feb 2024
General comments
The authors have developed a reservoir model, GLORY, that can be connected to the global integrated assessment model GCAM. This paper describes GLORY and reports the results of several preliminary experiments when connected to GCAM.
Water use has both physical and economic aspects. To capture water use from an economic aspect, a demand curve and a supply curve are necessary. For water without an explicit market, defining and estimating these is fraught with great difficulty. This study is extremely interesting because it attempts to break through this problem by focusing on the function of reservoirs.
Although the paper is long, it is basically well organized. However, I think there is a lack of description in terms of how the water balance and water use of the vast watershed was modeled. I also think there is a lack of definition of fundamental concepts such as “yield.” Consequently, the results, such as producing a "yield" that is more than one order of magnitude larger than the water storage capacity, are not immediately convincing to me, who have studied water use from a physical aspect. I think an additional explanation and information needs to be added.
Specific comments
Line 36 “Oki and Kanae, 2006; Vorosmarty et al. 2000; Organization and Fund 2000”: These studies pioneered the global water resources assessment at an annual time frame, but little was mentioned on a sub-annual scale. A sub-annual time frame was first introduced by the work of Hanasaki et al. (2008), followed by Wada et al. (2011) and Hoekstra et al. (2012).
Line 255: “water users can readily access the mean flow at excess cost. The high cost associated with accessing this upper limit reflects the likely high cost associated with extensive reservoir deployment”: This part sounds curious to me. It sounds like water users in the vast basin can access water resources anywhere, anytime. To achieve this, first, one needs to build a reservoir at the outlet (the most downstream point) of a basin to control the runoff. Moreover, a pressured aqueduct network is needed to deliver the outflow to the upstream users (i.e., the reservoir is at the lowest point in the basin; hence, pumping is needed for delivery). I fully understand simplification is unavoidable in modeling, but the treatment within the basin should be clearly mentioned.
Line 263, “The second data point”: I guess the authors are discussing the inflection points of the supply curve here. Better to provide a schematic here. Particularly, the second data point is hard to understand.
Line 269 Equation 1: Better to add a unit of each term. I think the unit of QA is km3 year-1. What I am wondering is the unit for RS. It must be in km3, but then the unit becomes different from the other terms.
Line 277 “(Note that… from groundwater depletion observed over a historical calibration period: withdrawal – depletion/runoff)”: I am puzzled here. Equation 1 doesn’t include withdrawal. Why do the authors mention water withdrawal here? A further explanation is needed.
Lie 294 “This yield may far exceed the physical storage capacity of the reservoirs in a basin.”: Additional explanation is needed here because the background of this claim is unclear. As shown in Figure 4, the world regions typically have one dry and one wet season in a year. Reservoirs can transfer water from the wet season to the dry season for their storage capacity. First, I am puzzled why the authors claim that the “yield” (same as seasonal water transfer, if I understood correctly) may far exceed the storage capacity. There is another concern: water reuse within a basin. Once water is used, the unconsumed fraction of water is drained, and it can be withdrawn again downstream. Therefore, the volume of water withdrawal can be greater than that of water transferred by reservoirs. If the authors mainly refer to this (i.e., the reuse of water within a basin), it should be clearly mentioned. Note that reuse is constrained by the consumption-to-withdrawal ratio. Hence, I guess it cannot “far exceed” the physical storage capacity.
Line 295 “the existing approach”: Specify what approach (i.e., studies) the authors refer to. Many physical models can simulate this (e.g., Masaki et al. 2017). Do you mean the earlier versions of GCAM?
Line 341 “by creating a single pool of water storage for each basin”: Again, it is better to mention where this aggregated reservoir was located clearly. I guess that the authors assumed that the reservoir is located at the lowermost point of each basin and that redistribution of water within a basin is costless and possible even in the large river basins in the world (e.g., the Nile, the Amazon, the Congo in Figure 1).
Line 344 “how much yield could be achieved if the system was operated to maximize yield”: Same as my comment above. I guess one important assumption of this study is that all runoff generated in a basin flows into the aggregated reservoir, and the outflow from the reservoir can be delivered to the entire basin. Because this assumption is not obvious, it is better to clarify. The concepts of “yield” and “maximization” also require this big assumption.
Line 377 Equation 8: I guess Et is evaporation from the water surface of the reservoir, but the equation looks like the basin's total evapotranspiration. I guess this becomes problematic in arid regions where Ig << Eg. Under this condition, Equation 3 will become negative. Clarify this point.
Line 388: “fp, pt and zt”: As a hydrologist, it is hard to imagine that these parameters can be objectively determined. The location, consequently, the catchment area and climate of the reservoir have been changed from reality. Indeed, the authors imply in Line 403 that these parameters were not calibrated against observations. Further elaboration is required on how the authors determined these parameters for 235 basins in the world at a monthly interval.
Line 407: “possible for that same level of storage capacity”: Same to what? What does “possible” mean?
Line 410: “to check the level of reservoir annual release from existing reservoir storage capacity is within the range of the annual release (same as yield) that GLORY produces at the same level of storage capacity.”: This part is particularly hard to understand. Fist, what does “the level” indicate? Second, I believe storage capacity and release are independent. For instance, the storage capacity of the Big Bend Dam (2.2 km3) and the Oahe Dam (28.5 km3) in the Missouri River is different, but their annual release is almost the same because these two dams are cascading. Simply, the mean annual release of a reservoir is the same as its inflow. A further explanation is needed here. Third, again, what does the “same level of storage capacity” mean? What does “level” indicate? Same to what?
Line 433 “A basin’s optimization for a particular GCAM period (e.g. 2050) is executed in the absence of any carryover of information about reservoir storage levels…”: Then, how was the initial storage of reservoir determined at the beginning of a given GCAM period?
Line 453 “via a monthly water demand fraction profile that shifts close in appearance to the monthly irrigation demand profile”: Hard to read. Do you mean that the profile of the total water demand becomes closer to the new irrigation demand? The expression “shifts close in appearance” sounds highly subjective and arbitrary.
Line 490 “we also filter out grid cells that do not contain existing water bodies suitable for siting of reservoirs (i.e., rivers)”: What do you mean “water bodies”? What kind of information the grid cells include? What data did you use?
Line 600 Equations 14-17: It is better to show the units for variables.
Line 633 “a nuanced analysis”: What does this mean?
Figure 7: I expected the unit of Yield is km3 year-1. As mentioned above, I still couldn’t understand why the yield could exceed the storage capacity. To begin with, add a concrete definition of reservoir yield at the beginning of this paper.
Line 721: “For example, the Indus basin is relying on 9.2km3 of storage capacity to provide about 135km3 of annual yield in 2050”: Again and again, I still cannot understand this. How one can serve 135 liters of water by 9.2 liters of a tank? Also, I am totally lost in how the authors treat the reuse (cascading use) of water within a basin. The authors’ model seems theoretically sound from the economic perspective. I am wondering whether the parameters and output numbers can be supported from the hydrological (physical) perspective.
Line 746 “informally validating the value offered by the new methodology we present here”: What does “informally” mean? How one can “validate” values?
Line 766 “The complex relationships between.. due to difficulties in meeting increased demand during irrigation seasons”: I guess what the authors report here is that the GCAM-GLORY can tell that irrigation in California should be decreased because irrigation water is unavailable (too costly) for a limited number of months. This is a significant advancement in global hydro-economic modeling.
Line 776 “a single point on each basin’s supply curve”: What is “a single point”?
Line 776 “Affordable water prices can indicate a basin has not invested and expanded reservoirs much, or the basin has well-expanded storage capacity, but the LOSC is relatively low.”: Hard to understand. What do you mean here?
Line 792 “firm water yield”: what’s this?
Figure 10: Surface water withdrawal in the Indus River exceeds 300 km3 year-1 for the Reference and Climate Impact scenarios, which is greater than the mean annual runoff (140 km3 year-1). How could this happen? For the No Feedback and Feedback scenarios, the usage of groundwater will sharply decrease after 2030. How could this be achieved? All in all, what will be the global total water withdrawal in 2050 for each scenario? What will be the global total reservoir capacity for each scenario?
References
Hanasaki, N., Kanae, S., Oki, T., Masuda, K., Motoya, K., Shirakawa, N., Shen, Y., and Tanaka, K.: An integrated model for the assessment of global water resources - Part 2: Applications and assessments, Hydrol. Earth Syst. Sci., 12, 1027-1037, 10.5194/hess-12-1027-2008, 2008.
Hoekstra, A. Y., Mekonnen, M. M., Chapagain, A. K., Mathews, R. E., and Richter, B. D.: Global Monthly Water Scarcity: Blue Water Footprints versus Blue Water Availability, PLoS One, 7, e32688, 10.1371/journal.pone.0032688, 2012.
Masaki, Y., Hanasaki, N., Biemans, H., Müller Schmied, H., Tang, Q., Wada, Y., Gosling, S. N., Takahashi, K., and Hijioka, Y.: Intercomparison of global river discharge simulations focusing on dam operation—multiple models analysis in two case-study river basins, Missouri–Mississippi and Green–Colorado, Environmental Research Letters, 12, 055002, 10.1088/1748-9326/aa57a8, 2017.
Wada, Y., van Beek, L. P. H., Viviroli, D., Dürr, H. H., Weingartner, R., and Bierkens, M. F. P.: Global monthly water stress: 2. Water demand and severity of water stress, Water Resources Research, 47, W07518, 10.1029/2010wr009792, 2011.
Citation: https://doi.org/10.5194/gmd-2023-204-RC2 - AC2: 'Reply on RC2', Mengqi Zhao, 03 Apr 2024
Data sets
Meta-Repository: Representing Reservoir Water Storage in the Global Change Analysis Model (GCAM) M. Zhao, T. B. Wild, N. T. Graham, S. Kim, M. Binsted, K. Chowdhury, S. Msangi, P. Patel, C. R. Vernon, H. Niazi, H. Li, and G. Abeshu https://doi.org/10.5281/zenodo.10211057
Model code and software
Meta-Repository: Representing Reservoir Water Storage in the Global Change Analysis Model (GCAM) M. Zhao, T. B. Wild, N. T. Graham, S. Kim, M. Binsted, K. Chowdhury, S. Msangi, P. Patel, C. R. Vernon, H. Niazi, H. Li, and G. Abeshu https://doi.org/10.5281/zenodo.10211057
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
285 | 94 | 22 | 401 | 31 | 15 | 15 |
- HTML: 285
- PDF: 94
- XML: 22
- Total: 401
- Supplement: 31
- BibTeX: 15
- EndNote: 15
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1