the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 27 Jun 2025
A multiple spatial scales water use simulation for capturing its spatial heterogeneity through cellular automata model
Abstract. Reliable water use simulation is essential for sustainable water resource planning, especially under intensifying pressures from climate change, population growth, and socio-economic transitions. While previous studies have extensively explored water availability as supply side modeling across multiple spatial scales for its spatial heterogeneity, the water demand side remains relatively underdeveloped—often constrained by fixed spatial scales and coarse statistical data that assume spatial homogeneity. This mismatch between supply side and demand side limits the ability of existing models to accurately represent spatial heterogeneity in water use and brings uncertainty into water resource allocation strategies. To address this mismatch, we propose a novel multi-scale water use simulation framework by integrating cellular automata (CA) model with Generalized Likelihood Uncertainty Estimation (GLUE). The CA model captures the spatial heterogeneity of water use through the grid-based update rules. Two update rules are adopted—probability rule (i.e., capturing stochastic transitions via distribution fitting) and linear rule (i.e., modeling neighborhood-weighted evolution). To evaluate the impacts of spatial scale on water use heterogeneity, simulations are conducted at three spatial scales: 1 km, appropriate scale, and prefecture scale across 341 prefectures in China. Results show that both the update rule and spatial scale significantly affect spatial heterogeneity and uncertainty of water use. The probability rule can capture the broader variability but results in higher Root Mean Squared Error (RMSE) and Relative Error (RE) while the linear rule brings more stable performance with lower errors. While the 1 km scale increases uncertainty due to sensitivity to local fluctuations, and the prefecture scale suppresses spatial details, the appropriate scale offers the best trade-off between stability and spatial heterogeneity. The uncertainty quantified by GLUE, expresses as confidence intervals, varies across prefectures and spatial scales. Overall, the proposed framework offers a flexible tool for multi-scale water use simulation and highlights the critical role of spatial heterogeneity, thereby supporting adaptive water resource planning and management.
-
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
(5541 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(5541 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-2734', Anonymous Referee #1, 26 Jul 2025
-
AC1: 'Reply on RC1', Jiayu Zhang, 17 Aug 2025
We sincerely thank the reviewers for their constructive and insightful comments, which have greatly helped us to improve the quality and clarity of our manuscript. In response, we have carefully addressed each comment point by point and made the corresponding revisions in the main text. A detailed reply to the reviewers’ comments, along with explanations of the modifications, is provided in the supplementary PDF file. We hope that the revised manuscript meets the expectations of the reviewers and the editor.
-
AC1: 'Reply on RC1', Jiayu Zhang, 17 Aug 2025
-
RC2: 'Comment on egusphere-2025-2734', Anonymous Referee #2, 09 Oct 2025
This study proposes a multi-scale water use simulation framework that integrates a Cellular Automata (CA) model with Generalized Likelihood Uncertainty Estimation (GLUE) to address the impact of spatial scale on the spatial heterogeneity and uncertainty of water use. The topic is highly relevant, given the growing mismatch between water supply and demand under climate and socio-economic changes. The methodology is somehow innovative and the case study across 341 Chinese prefectures is comprehensive. The proposed method and dataset can offer potential values for water resources planning. However, many issues still exist, including methods’ details, result precenting, and unclear findings. They should be fully addressed before can be accepted.
- How the water use in this study is defined? Including all human-used water? Irrigation, industrial, rural and urban use? From river and lake water? And underground water?
- What is the values of the appropriate spatial scale? Can the model show the appropriate spatial scale in unit of km for each region?
- What is the difference between water use grid maps and water use simulation results? It seems that water use grid maps from CNN model has relatively good performances, why we need continue to estimate water use at grid-scales using CA model?
- The method part shows that CV, Moran’s I, AIC are used for analyzing the model performance, but I did not find their resulting values in the Result or other part. Please provide more details about the analysis on spatial Heterogeneity of water use.
- Figure 4 and Figure 6, it seems not necessary to show all plots from 1998 to 2013, and it is not easy to recognize differences between different spatial scales. Here, in Figure 4 and 6, please only show plots from 2010-2013 which are from the validation mode. Please provide data availability statement presenting data links for water use simulation results over 1998-2013.
- At the same spatial scale, what is difference between the probability rule and the linear rule? Can you calculate the difference between Figure 4 and Figure 6, providing more statistical information?
- What is the best estimation of water use simulation in this study, and what is the overall accuracy for China and each province?
Citation: https://doi.org/10.5194/egusphere-2025-2734-RC2 -
AC2: 'Reply on RC2', Jiayu Zhang, 27 Oct 2025
We sincerely thank the reviewers for their constructive and insightful comments, which have greatly helped us to improve the quality and clarity of our manuscript. In response, we have carefully addressed each comment point by point and made the corresponding revisions in the main text. A detailed reply to the reviewers’ comments, along with explanations of the modifications, is provided in the supplementary PDF file. We hope that the revised manuscript meets the expectations of the reviewers and the editor.
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-2734', Anonymous Referee #1, 26 Jul 2025
The research on "A multiple spatial scales water use simulation for capturing its spatial heterogeneity through cellular automata model" provides a novel multi-scale water use simulation framework by integrating cellular automata (CA) model with Generalized Likelihood Uncertainty Estimation (GLUE) over China. This study fills the gap of multi-scale water use estimation and offers a flexible tool. The paper is well organized and the topic is suitable for HESS. However, I have some suggestions before publication.
- The main objective of this study is to address the impact of spatial scale on the spatial heterogeneity of water use. However, this study treats water use as a whole, without separating the spatial heterogeneity of water use by different sectors (e.g., irrigation, industrial and domestic). Actually, water use by different sectors shows large spatial variation which I think is important in water use modelling.
- In the model framework, water use grid maps at different spatial scales are first prepared as inputs to the CA model. To generate the water use grid maps, the water use data at administrative survey scale is processed and downscaled to grid-based formats. Here, the iterative input selection algorithm is used to select the most relevant variables for water use, while the CNN model captures the relationships between input variables and water use. This step is important for the model performance, So I am wondering what are the most relevant variables for sectoral water use? This result should appear in this paper. Commonly, irrigation water use is mostly related to irrigated cropland area, and industrial/domestic water use are relevant to GDP/population density, as well as night light intensity. Whether the result of this study aligns with previous results?
- L145: “the appropriate spatial scale for water use simulation is identified using an end-to-end deep learning-based spatial scale adaptive selection model”. What is the definition of “appropriate spatial scale”? In my view, the appropriate spatial scale should be a clear spatial resolution (e.g., 5km, 10km), and may vary across water use sectors or limited water use dataset. It is suggested to clarify the result of the appropriate spatial scale in gridded water use simulation.
- Water use simulation from the probability rule CA model (Section 4.1.1) are not validated. This part uses the Akaike Information Criterion (AIC) to determine the most suitable probability distributions for water use grids across various prefectures. However, the optimal probability distributions also rely on the input data (e.g., the long-term gridded water use data). As water use in China shows significant spatial and temporal variation between different periods, it is doubtable that the probability rule CA model can used for water use prediction.
- This study calibrates the parameters in the linear rule CA model for the 1998–2009 while the dataset from 2010–2013 is for its validation. However, the calibration and validation processes are not clear. Which datasets are used for model evaluation, the prefecture water use data or the gridded water use maps?
- The main objective of the model framework is to generate water use data at multiple spatial scale. There are many gridded water use products at both global or country scale for China (e.g., Hou et al., 2024, ESSD; Huang et al., 2018, HESS; Zhang et al., 2025, Scientific Data),a s well as the high-resolution hydrological model simulations. It is necessary to compare the water use simulation with previous products, which helps to evaluate the reliability of the model framework of this study.
- Figure 4 & 6: water use is high in many irrigated areas. However, water use in the North China Plain which is marked with intensive irrigation and population, shows moderate level of water use, lower than that of the northeastern China. This result is contrary with previous estimates.
- L39: Key words: This study is all about water demand/water use, and “water resources management” and “water scarcity assessment” are not suitable for the keywords.
- L145：I don’t find the reference for Liu et al., 2022.
Citation: https://doi.org/10.5194/egusphere-2025-2734-RC1 -
AC1: 'Reply on RC1', Jiayu Zhang, 17 Aug 2025
We sincerely thank the reviewers for their constructive and insightful comments, which have greatly helped us to improve the quality and clarity of our manuscript. In response, we have carefully addressed each comment point by point and made the corresponding revisions in the main text. A detailed reply to the reviewers’ comments, along with explanations of the modifications, is provided in the supplementary PDF file. We hope that the revised manuscript meets the expectations of the reviewers and the editor.
-
RC2: 'Comment on egusphere-2025-2734', Anonymous Referee #2, 09 Oct 2025
This study proposes a multi-scale water use simulation framework that integrates a Cellular Automata (CA) model with Generalized Likelihood Uncertainty Estimation (GLUE) to address the impact of spatial scale on the spatial heterogeneity and uncertainty of water use. The topic is highly relevant, given the growing mismatch between water supply and demand under climate and socio-economic changes. The methodology is somehow innovative and the case study across 341 Chinese prefectures is comprehensive. The proposed method and dataset can offer potential values for water resources planning. However, many issues still exist, including methods’ details, result precenting, and unclear findings. They should be fully addressed before can be accepted.
- How the water use in this study is defined? Including all human-used water? Irrigation, industrial, rural and urban use? From river and lake water? And underground water?
- What is the values of the appropriate spatial scale? Can the model show the appropriate spatial scale in unit of km for each region?
- What is the difference between water use grid maps and water use simulation results? It seems that water use grid maps from CNN model has relatively good performances, why we need continue to estimate water use at grid-scales using CA model?
- The method part shows that CV, Moran’s I, AIC are used for analyzing the model performance, but I did not find their resulting values in the Result or other part. Please provide more details about the analysis on spatial Heterogeneity of water use.
- Figure 4 and Figure 6, it seems not necessary to show all plots from 1998 to 2013, and it is not easy to recognize differences between different spatial scales. Here, in Figure 4 and 6, please only show plots from 2010-2013 which are from the validation mode. Please provide data availability statement presenting data links for water use simulation results over 1998-2013.
- At the same spatial scale, what is difference between the probability rule and the linear rule? Can you calculate the difference between Figure 4 and Figure 6, providing more statistical information?
- What is the best estimation of water use simulation in this study, and what is the overall accuracy for China and each province?
Citation: https://doi.org/10.5194/egusphere-2025-2734-RC2 -
AC2: 'Reply on RC2', Jiayu Zhang, 27 Oct 2025
We sincerely thank the reviewers for their constructive and insightful comments, which have greatly helped us to improve the quality and clarity of our manuscript. In response, we have carefully addressed each comment point by point and made the corresponding revisions in the main text. A detailed reply to the reviewers’ comments, along with explanations of the modifications, is provided in the supplementary PDF file. We hope that the revised manuscript meets the expectations of the reviewers and the editor.
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 693 | 55 | 25 | 773 | 22 | 32 |
- HTML: 693
- PDF: 55
- XML: 25
- Total: 773
- BibTeX: 22
- EndNote: 32
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Jiayu Zhang
Dedi Liu
Jiaoyang Wang
Feng Yue
Hanxu Liang
Zhengbo Peng
Wei Guan
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(5541 KB) - Metadata XML
The research on "A multiple spatial scales water use simulation for capturing its spatial heterogeneity through cellular automata model" provides a novel multi-scale water use simulation framework by integrating cellular automata (CA) model with Generalized Likelihood Uncertainty Estimation (GLUE) over China. This study fills the gap of multi-scale water use estimation and offers a flexible tool. The paper is well organized and the topic is suitable for HESS. However, I have some suggestions before publication.