We sincerely thank Professor Nima Zafarmomen for their constructive and thoughtful comments. These suggestions have greatly helped us improve the clarity of the PSR framework’s dynamic interactions and the transparency of our uncertainty analysis. Our point-by-point responses are provided below.

1. Linking the PSR framework with results

In the revised manuscript, we have restructured the Results section to explicitly connect the “Pressure-State-Response” (PSR) framework with the findings. For example, in Section 3.3, we now include:

Within the PSR framework, changes in internal and external irrigation withdrawals constitute pressure factors, which affect the state of the regional water system by modifying water availability and the supply-demand balance. The resulting classification of pressure types then reflects the system’s response.

Subregion 8 is highly sensitive to climate scenario variations during the dry season and exhibits pronounced scenario dependence. Increasing upstream irrigation withdrawals (pressure) interact with climate-driven hydrological changes to influence downstream water availability (state), ultimately leading to shifts in the dominant irrigation pressure regime (response).

2. Justification for reserving 30% of simulated runoff as environmental flow

We have added citations and discussion to justify this threshold:

Methodological basis: Smakhtin et al. (2004) estimated that for the Lancang-Mekong river, an environmental flow requirement of approximately 30% of mean annual runoff is needed to sustain basic ecosystem functions.

Regional consistency: This threshold aligns with the “Good” ecological condition in the Tennant (1976) method.

Sensitivity analysis: While absolute water pressure values vary with the environmental flow threshold, the spatial patterns and timing of the regime shift in Subregion 8 remain robust within a reasonable environmental flow range.

3. Addressing uncertainties in irrigation withdrawal estimation

We have added a new subsection in the Discussion to address uncertainties:

Irrigation efficiency coefficients: We acknowledge that irrigation efficiency varies across countries and crop types in the Lancang–Mekong Basin. Coefficients were adopted from FAO and MRC reports; actual efficiencies may differ due to local management practices and infrastructure age. Sensitivity analysis shows that higher efficiency reduces absolute withdrawal volumes but does not change spatial patterns of water pressure or the timing of regime shifts in Subregion 8.

Irrigation infrastructure mapping: High-resolution (0.5 m) satellite imagery combined with a deep learning–based recognition model was used to minimize mapping uncertainty. This approach provides a more detailed representation than traditional coarse datasets. Validation results indicate over 90% accuracy, supporting the reliability of the identified features.

4. Language and stylistic improvements

We conducted a comprehensive proofreading of the manuscript. Article usage (e.g., “under climate change”) and phrasing consistency (e.g., “30% of the simulated runoff”) have been corrected. Several sections have been restructured to improve readability and professional tone.

5. Addition of missing reference

We thank the reviewer for highlighting this relevant study. The reference: “Assimilation of Sentinel-based leaf area index for modeling surface–groundwater interactions in irrigation districts” has now been added to the Introduction, as it aligns closely with the manuscript’s themes of irrigation modeling.

We sincerely thank the reviewer for the constructive and detailed comments. We have revised the manuscript accordingly, addressing each point with specific improvements to figure clarity, uncertainty analysis, and the discussion of policy implications.

General Responses

1. Improvement of Figure Clarity and Visualization of Uncertainty

We have revised the figures to enhance their self-explanatory power. Legends and captions have been expanded to define all plotted elements, including corresponding time periods, scenarios, and data sources.

Following the reviewer’s suggestion, we quantified the uncertainty associated with the five GCM projections by calculating the deviation of individual GCM outputs from the multi-model ensemble mean for the ratio of irrigation water withdrawal to available water.

Quantified Results: The inter-model spread is relatively limited. At the annual scale, the uncertainty range (maximum deviation) is 0–1% for SSP1-2.6, 0–2% for SSP2-4.5, and 0–3% for SSP5-8.5. During the dry season, the range remains within 0–6% for SSP1-2.6, 0–6% for SSP2-4.5, and 1–6% for SSP5-8.5. Figure 6 and Figure 7 have been updated to display both the ensemble mean and the associated uncertainty range across the five GCMs.

2. Expanded Discussion of Uncertainty Sources

We have added a section discussing three primary sources of uncertainty:

Model Spread: We quantified uncertainty intervals for projected irrigation demand and runoff across the five GCMs to identify areas where projections are most robust.

Irrigation-demand Assumptions: We discussed the limitations of using Mekong River Commission (MRC) irrigation expansion plans. While authoritative, actual implementation may vary due to policy shifts, economic development, or changing cropping patterns.

Climate-scenario Variability: We expanded the analysis to compare how differences in future climate forcing (SSP1-2.6,SSP2-4.5,SSP5-8.5) propagate into uncertainties in runoff and irrigation pressure.

3. Strengthening the Link Between Scientific Findings and Water Governance

To improve policy relevance, we added a new subsection titled “Policy implications and transboundary cooperation.” Specifically, we:

Linked spatial vulnerability and seasonal pressure shifts to management interventions (e.g., targeted demand management, seasonally adaptive allocation, and reservoir adjustments).

Discussed transboundary governance mechanisms, including enhanced data sharing, joint monitoring systems, and institutional frameworks.

Identified priority actions for vulnerable subregions, such as irrigation-efficiency improvements and optimized upstream storage operations.

Detailed Point-by-Point Responses

1. The sentence has been revised to: “The analysis identifies vulnerable components of the basin system, clarifies the spatial distribution of transboundary water pressures, and provides a basis for differentiated water governance strategies across the basin.”

2. We have replaced the previous phrasing with a more direct and active structure to improve clarity. The sentence now reads:

"Accordingly, numerous studies have investigated water stress and its driving mechanisms in transboundary river basins (Chen et al., 2020; Do et al., 2020; Tian et al., 2020)."

3. National boundaries are retained as they are essential for illustrating the spatial relationship between subregions and countries in a transboundary context. The caption now explicitly states that the spatial map represents the distribution of irrigation areas for 2020 and includes the data source.

4. We have revised the Methods section to provide a more detailed description of the datasets. Irrigated area data were obtained from the FAO Global Map of Irrigation Areas  dataset, which integrates national and subnational irrigation statistics with geospatial information on irrigation locations and extents. The spatial distribution of irrigated land was further combined with the global equipped irrigated area dataset and the Spatial Production Allocation Model (SPAM) dataset to represent irrigated crop distribution across the basin during the period 1980–2020.

5. We apologize for the error. The streamflow data used in this study were obtained from the Mekong River Commission (MRC) database (https://portal.mrcmekong.org/home). This has been corrected.

6. We have expanded the description of data in the Methods section:

Irrigation canal network data were derived from high-resolution Google satellite imagery with a spatial resolution of 0.5 m, which was downloaded through the MapGIS software platform. Irrigation canals were identified using a deep learning–based intelligent remote sensing recognition model. Canal headworks were extracted and treated as irrigation water abstraction points for the irrigation water-withdrawal analysis (Zhao et al., 2025). Reservoir data were obtained from the Mekong River Commission (MRC) and the Mekong Region Futures Institute (2024). The dataset includes reservoir locations, storage capacities, and years of operation, covering both existing and planned reservoirs during the period 1965–2035, and was used to represent basin-scale water storage and regulation capacity in the analysis.

Soil data were obtained from the global soil database of the Food and Agriculture Organization of the United Nations (FAO). The dataset has a spatial resolution of 10 × 10 km and provides key soil parameters used in the hydrological model. The normalized difference vegetation index (NDVI), leaf area index (LAI), and snow cover fraction were derived from MODIS products, with a spatial resolution of 500 × 500 m and a temporal resolution of 16 days. These datasets were used as input variables to drive the hydrological model, representing vegetation and land-surface conditions affecting evapotranspiration and runoff processes.

Irrigated area data were obtained from the FAO Global Map of Irrigation Areas dataset, which integrates national and subnational irrigation statistics with geospatial information on irrigation locations and extents. The spatial distribution of irrigated land was further combined with the global equipped irrigated area dataset and the Spatial Production Allocation Model (SPAM) dataset to represent irrigated crop distribution across the basin during the period 1980–2020. Spatial distributions of irrigated crop types were derived from the SPAM Global dataset, which includes information for 46 crop types and has a spatial resolution of 10 × 10 km (IFPRI, 2024; available at https://www.mapspam.info/data/)

7. Clarified that runoff was first simulated at a daily time step and then aggregated to monthly runoff for each sub-basin (1980–2020). Irrigation water withdrawal was calculated at a monthly scale.

8. We have clarified the scenarios in the text and figure: (1) with reservoir operation and (2) without reservoir operation. The operation scheme follows Zhang et al. (2026) (https://doi.org/10.5194/hess-30-671-2026). 

9. Added as suggested.

10.We removed the national boundary lines and added the river network to highlight natural hydrological features and improve interpretability.

11. The figure now explicitly presents the uncertainty ranges (GCM spread) alongside the ensemble mean.

12. We added standard deviation to represent the inter-model spread across the five GCMs.

Dual Bars in d/f: These bars represent different SSPs (Climate Scenarios). We have updated the legend and caption.

13. We have updated the line colors in subplots (b) and (c).

Thank you very much for your thoughtful and constructive comments on our manuscript. We have carefully considered each point and will revise the manuscript accordingly. Our detailed responses are provided below.

1. Temporal Scale and Data Availability

Regarding the calibration period, we clarify that the observed streamflow data for the main stem were indeed available for the full period (1980–2020) and were used for model calibration and validation. However, data for several tributary stations only cover 2000–2020. In the original manuscript, we reported a uniform period (2000–2020) to maintain consistency in the description, which we now realize was misleading. In the revised manuscript, we will explicitly distinguish the calibration periods: 1980–2020 for the main stem and 2000–2020 for the tributaries.

In this study, hydro-meteorological data for 2021–2025 were incorporated using the ERA5-Land reanalysis dataset, which represents historical atmospheric and land-surface conditions rather than future scenario projections. However, because the future analysis in this study was conducted at a decadal scale (e.g., 2021–2030 and 2031–2040) to ensure consistency with the scenario-based projections, the period 2021–2025 was grouped within the 2021–2030 interval and analyzed together as part of the first future decade. We acknowledge that this was not clearly explained in the original manuscript and may have caused confusion. Following the reviewer’s suggestion, we will revise the manuscript to clarify it.

2. Clarification of Basin-scale Constraints

Following your suggestion, we will expand the Introduction to explicitly discuss the historical constraints hindering full-basin, multi-pressure analyses. We will focus on two primary obstacles: (i) data scarcity and asymmetry in transboundary contexts, and (ii) the computational complexity of simulating fine-scale human-water interactions within a large-scale hydrological framework.

3. Lateral Interdependency and National Boundaries

We appreciate this insight regarding the "left-right bank" relationships. Our THREW model utilizes the Representative Elementary Watershed (REW) as the fundamental unit, which is delineated based on hydro-geomorphological features rather than political borders. Irrigation withdrawals are represented as nodes at specific river reach locations. Water extracted at these nodes directly alters the streamflow, which is then propagated through the river routing process across connected REWs.

This structure allows the model to capture "lateral" interactions: even if riparian countries are situated on opposite banks, their respective withdrawals impact the shared river reach's water balance within the same or hydrologically connected REWs. We will revise the manuscript to include a more detailed explanation of the REW-based routing and node-link structure to clarify how these transboundary interdependencies are captured.

4. Reservoir Operation Schemes

The reservoir operations in our model are represented by a rule-based module that balances hydropower generation with downstream water requirements. This operational framework follows the established schemes detailed in our previous work (Zhang et al., 2026, https://doi.org/10.5194/hess-30-671-2026), which has been validated for the Lancang-Mekong cascade.

Specifically, the simulation incorporates:

Operating Rules: The module optimizes storage and release based on seasonal water availability and hydropower demand, ensuring that reservoir storage levels are managed within safety and operational limits.

Downstream Constraints: To ensure ecological and agricultural security, a minimum environmental flow requirement is strictly enforced as a baseline constraint for all release processes.

Model Structure: We individually simulate major reservoirs that have direct downstream irrigation dependencies, while smaller or less critical ones are aggregated into complexes to maintain computational efficiency.

In the revised manuscript, we will include a description of these operation rules, while also citing the relevant literature to provide further technical details.

5. Reconciliation of Upstream Withdrawal and Reservoir Regulation

We agree that post-2009 reservoir operations have significantly augmented dry-season discharge compared to historical natural flow. As shown in our results (Figure 3), our model captures this regulatory effect: after 2010, the "reservoir scenario" shows a moderation of dry-season irrigation pressure compared to the "no-reservoir scenario" (e.g., a reduction from 36% to 34%). This indicates the model is sensitive to the reservoirs' compensatory role.

However, the reason the overall water pressure continues to rise is the result of a "dual squeeze" from both supply and demand sides:

Supply Side (Natural Decline): Our analysis shows a sustained decline in natural runoff across the basin since 2000, likely driven by shifting climatic patterns. This reduction in the "total available pool" means that even with reservoir redistribution, the baseline water volume has shrunk.

Demand Side (Irrigation Expansion): Simultaneously, the consumptive use of water for irrigation has grown at a rate that outpaces the supplementary capacity of reservoir regulation.

In summary, while the reservoirs effectively redistribute water to the dry season, the net gain from regulation is partially offset by the overarching decline in natural inflow and further strained by intensifying irrigation withdrawals. In the revised manuscript, we will add a discussion to clarify these competing mechanisms.

6. Minor Issues

Figure 8: We will adjust the layout and font sizes to eliminate label overlap and ensure readability.

Figure 6: We will reconsider the visualization style to better present the specific data type and improve clarity.