the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 09 Sep 2026
A method for representing cities in global hydrological models
Abstract. Due to increased computational resources and the availability of fine-scale global data, global hydrological models have begun to run at spatial resolutions that resolve major cities worldwide. Because water supply in cities is achieved by a set of infrastructure, traditional grid-level or basin-level modeling and assessment do not provide reliable and interpretable results for cities. Here, we propose a generic city sub-model for global hydrological models to express the water supply in cities. Using a newly developed global gridded database of city areal extent, we conducted a series of global hydrological simulations focusing on municipal water supply for 11 representative cities. Although the model was uncalibrated and forced with global data, the simulation performed reasonably for these cities. The results improved further when additional data on external water sources were taken into account. The subsequent city-level water assessment provides a single water scarcity score for each city. It is more interpretable than the traditional grid- or basin-level assessment, which typically shows a wide range of scores across grid cells or basins or includes non-urban regions. This study highlights the importance of representing cities in global hydrological models and water scarcity assessments, and the need to develop a comprehensive global database of urban water supply. This study opens a pathway toward city-level global water assessments using generic global hydrological models.
- Preprint
(4035 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 07 Dec 2026)
- RC1: 'Comment on egusphere-2026-5322', Anonymous Referee #1, 22 Oct 2026 reply
-
RC2: 'Comment on egusphere-2026-5322', Anonymous Referee #2, 05 Nov 2026
reply
General comments:
The manuscript presents a new city sub-model for the global hydrological model H08 to improve urban water supply representation and city-level water scarcity assessment. The study is novel and relevant, as urban water supply systems are not adequately represented in many global hydrological models. However, several aspects require further clarification and analysis.
The methodological contribution of the city sub-model needs to be more clearly established. Since H08 already represents municipal water use and multiple water sources, the specific contributions of the new city sub-model beyond the existing H08 model, together with its implementation details, should be more clearly explained.
Furthermore, the model evaluation remains limited. River discharge is compared with observations, whereas the water withdrawal comparison mainly evaluates input data. The CAD analysis illustrates differences among assessment approaches but does not directly validate city-level water scarcity estimates. More evidence is needed to establish the reliability of the proposed approach.
The discussion focuses mainly on uncertainties and challenges, with limited interpretation of the differences among simulations. A clearer explanation of the model's contributions, stronger evaluation, and more detailed interpretation of the results would strengthen the manuscript.
Specific comments
Abstract
Lines 15–17: The statements regarding model performance and improvements should be supported by quantitative results or evaluation metrics.
1 Introduction
The Introduction should provide a clearer overview of previous studies on urban water use and supply in global hydrological models, their limitations, and the specific research gap addressed by this study.
The study objectives should be more precisely defined. The manuscript primarily focuses on representing municipal water abstraction, supply, and return flows for city-level water scarcity assessment, rather than simulating the full range of urban hydrological processes. This distinction should be clarified to define the scope of the proposed sub-model.
2 Methods
2.1.1 The H08 global hydrological model
H08 already represents municipal and industrial water use, reservoirs, groundwater abstraction, and water transfers. The specific modifications introduced by the city sub-model should be clearly distinguished from existing H08 functions.
2.1.2 City sub-model
Lines 93–100; Figure 1: The implementation of the city sub-model requires further explanation. How are water demand and availability aggregated at city inlets? How are water abstraction and return flows allocated among multiple inlets and outlets? The relevant calculations and treatment of water balance should be clarified.
The manuscript should more clearly explain how the spatial aggregation of urban water demand and availability and the reassignment of water abstraction and return flows to city inlets and outlets improve upon the existing H08 representation.
The model assumes that river-entry and river-exit grid cells represent water supply and wastewater discharge locations. The basis and limitations of these assumptions should be clarified, particularly whether the selected locations correspond to actual urban water infrastructure.
Lines 93–95: The authors describe repeated abstraction of return flows within a city as unrealistic. However, downstream reuse of discharged water can occur in practice. A clearer distinction is therefore needed between unrealistic repeated abstraction in grid-based models and actual successive water use within cities.
The authors should also discuss whether the same aggregation method is suitable for cities with different water supply systems, including those relying on groundwater, reservoirs, and inter-basin transfers.
2.2 Data
Line 113: The W5E5 dataset should be defined when first introduced.
2.3 Expression of cities and rivers
Figure 2: The figure is difficult to read because of the numerous small panels. A global map showing the locations of the 11 cities, together with selected regional maps, would improve clarity.
Lines 151–154: The criteria for replacing several of the largest cities with other cities should be explained more clearly, particularly regarding data availability and representativeness.
Lines 170–171: The authors should explain how external water sources and city-specific water-use data were determined, including how the collected data were processed.
Lines 183–186: Since the representation of external water sources relies on the existing explicit-aqueduct function in H08, the contribution of the new city sub-model should be distinguished from that of the original model.
2.5 Validation and assessment
Lines 207–209: Appropriate references and the specific dates used for discharge validation should be provided for the observation data from GRDC and other sources.
3 Results and Discussion
3.1.1 Discharge
The discharge evaluation is presented only for URB. To assess the hydrological effects of the city sub-model, simulated discharge and KGE metrics should also be compared between CTR and URB across cities.
Several cities show negative KGE values. The main sources of error should be discussed, particularly the negative correlation and earlier simulated discharge peak in Tokyo and the weak seasonal variation in Los Angeles.
Figure 3: The caption should specify the observation sources and periods, define the reported KGE components (r, α, and β), and clarify the river and station names. Adjusting the axis scales or panel layout could also facilitate comparison of seasonal variations among cities.
Lines 238–239: Discharge was evaluated for only 10 of the 11 cities because observations were unavailable for Chennai. The authors should explain how the absence of discharge validation for Chennai affects the interpretation of its water scarcity assessment. In addition, the rationale for retaining Chennai as a study city despite the lack of discharge observations should be clarified.
3.1.2 Water withdrawal
The comparison of municipal water withdrawal in Figure 4 primarily evaluates the reliability of the withdrawal data used to construct the model inputs, rather than independently validating a newly simulated output of the city sub-model. This distinction should be made explicit.
The authors should clarify how gridded municipal water withdrawal data are aggregated into city-level water use in the new city sub-model, including the spatial aggregation method and relevant calculations. The treatment of industrial water use, consumptive use, and return flows should also be clarified.
Several cities show substantial overestimation or underestimation of water withdrawal. The extent to which these discrepancies arise from the spatial interpolation of national-level data should be further examined. The manuscript should also explain why the available city-specific municipal water withdrawal observations were not used to correct or refine the input data.
3.2 Water scarcity assessment
Figure 5 and Table 4: The comparison among grid-level, city-level, and basin-level assessments is central to the manuscript. However, the discussion focuses mainly on interpreting the city-level CAD scores, while the differences among assessment approaches are not sufficiently examined.
For several cities, including Cape Town, Los Angeles, Mexico City, and São Paulo, the city-level CAD scores are substantially lower than the corresponding basin-level scores or mean grid-level scores. The processes and differences in spatial aggregation responsible for these discrepancies should be explained in greater detail, rather than only describing the resulting water scarcity classifications.
A single city-level CAD score may improve interpretability, but this alone does not establish the reliability of the resulting water scarcity estimates. The authors should clarify this distinction and provide stronger evidence supporting the reliability of the city-level assessment.
Figure 5: The four assessment approaches should be more clearly distinguished using row labels or panel identifiers.
Table 4; Lines 289–292: The text indicates that grid-level and basin-level results are obtained from CTR, while city-level results are obtained from URB or EXT. However, the grouping of results in Table 4 appears inconsistent with this description. The table should be revised to clarify the corresponding simulations.
The authors should also explain the purpose and limitations of comparing city-level and basin-level CAD values, given that the latter incorporate water availability and demand outside the city boundary. The comparison is useful for illustrating differences in assessment scale, but it should not be interpreted as a direct validation of the city-level index.
3.3 External water resources
The increase in CAD after adding external water sources is expected when water availability increases while demand remains unchanged. The differences in the magnitude of these increases among cities should be explained, including whether the results are consistent with available observations.
The interpretation should also consider uncertainties in simulated river discharge and municipal water withdrawal, which may lead to overestimated or underestimated CAD scores.
3.4 Uncertainties and challenges
The manuscript discusses several important uncertainties, but more analysis is needed to assess their effects on the main results. In particular, the influence of discharge and water withdrawal errors on CAD should be examined, potentially through a sensitivity analysis for selected cities.
The differences among CTR, URB, and EXT should also be interpreted more clearly to distinguish the effects of the city sub-model from those of additional external water source data.
4 Conclusions
The Conclusions should emphasize the new representation of cities as integrated water supply and assessment units, rather than implying that urban water use and transfer processes are introduced into global hydrological models for the first time.
The distinction between improved interpretability and improved accuracy should be made clear, together with a more balanced assessment of the demonstrated contributions of the proposed method. Although the city-level method improves the interpretability of water scarcity assessments, the accuracy and reliability of the resulting CAD estimates remain uncertain because of limitations in discharge simulation, water withdrawal data, and the representation of urban water supply infrastructure. These limitations, together with the limited direct validation of city-level water scarcity estimates, should be acknowledged when discussing the potential for global application.
Citation: https://doi.org/10.5194/egusphere-2026-5322-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 348 | 172 | 70 | 590 | 73 | 67 |
- HTML: 348
- PDF: 172
- XML: 70
- Total: 590
- BibTeX: 73
- EndNote: 67
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The manuscript addresses an important research question concerning the representation of cities and their water supply networks within a global hydrological model. Although the research problem is relevant and clearly identified, the manuscript lacks sufficient methodological detail, quantitative evaluation, and qualitative analysis to adequately demonstrate the performance and reliability of the proposed approach. In particular, the model outputs require more rigorous statistical comparisons and a clearer discussion of the assumptions and limitations.
I recommend 'reject and resubmit', as substantial improvements are needed to strengthen the scientific rigor, methodological transparency, and overall quality of the manuscript.
The following concerns should be addressed: