Hydrological modelling in complex-terrain areas is highly challenging due to uncertainties in model structure, parameters, and input data, further complicated by the issue of equifinality, which often obstructs accurate interpretation of hydrological processes. This study utilizes multi-source data to calibrate a semi-distributed hydrological model, thereby providing better constraints on key hydrological processes. The findings offer valuable insights and contribute to advancing the understanding of water cycle in complex terrains. Main concerns are listed as follows.

1) Regarding the hydrological and calibration strategy, more details are needed. Key information, such as the inputs and outputs of the model, model integration time step, temporal resolution of reference data used for model calibration, and the number of iterations for each calibration, needs to be clearly specified.

2) In terms of validation, the comparison of water balance components across different scenarios should be strengthened, particularly with respect to AET and glacier meltwater loss.

3) Given that runoff observations are rare in complex terrain areas, while satellite-observed evapotranspiration and glacier changes are more readily available, it is recommended to consider another scenario that explores the model performance when calibrated without using observed runoff (i.e., utilizing only AET and glacier meltwater loss), which may have broader applicability in complex-terrain areas.

Some minor points:

4) The title focuses on flood modelling, which is inconsistent with the paper's limited coverage of this topic. It is suggested that the title be revised to more accurately reflect the actual content of the paper.

5) Line 15, “the model … precipitation input”, observed streamflow was also used for model calibration.

6) Line 18, “Multi-variable calibration improved…”, Multi-variable calibration not always improved the simulation of hydrological fluxes, as evidenced by the poorer streamflow simulation in Scenario 1 compared to Scenario 4. However, multi-source data calibration can provide a more plausible representation of hydrological processes.

7) Line 184, the method for rainfall-snowfall partitioning should be provided.

8) Does DDF in Equation 3 represent the degree-day factor for snow, ice, or both?

9) Line 227-229, it would be appreciated to provide a figure or table related to these results.

10) Why are glacier mass losses not simulated in Scenarios 2-4?

11) Line 289, the basis for determining these weights requires clarification.

12) Line 423-424, it is suggested to provide the root mean square errors for calibration and validation periods, respectively.

13) Figure 5, it is recommended that the mean annual AET for both the calibration and validation periods be presented separately.

14) Figure 6, it is recommended to incorporate reference values for glacier mass loss into this figure.

15) It is recommended to compare the mean annual water balance across different scenarios, following the format of Figure 8.

General comments

In this work, the authors present an extended version of a semi-distributed hydrological model and evaluate its performance across multiple processes in the complex, monsoon-dominated regions of the Indian Himalayas. The manuscript is generally well written, and the results are discussed in depth. However, I have few methodological concerns that should be clarified prior to publication.

Novelty of the study: The authors should more clearly articulate the key scientific advances of this work. Extending and evaluating an existing hydrological model and exploring multiple calibration approaches are important objectives, but the manuscript does not sufficiently clarify what constitutes the main innovation compared with existing literature. This should be clearly stated, ideally in the introduction, so that the contribution of the paper is evident from the outset.

Model calibration procedures and uncertainty discussion: Although calibration is presented as a central component of the study, there is no explicit investigation of model parameter uncertainty. It remains unclear how the different calibration methods and their respective constraints influence the uncertainty associated with parameter identification. Presenting only a single optimal value for each parameter, is restrictive, particularly given that calibration is one of the main goals of the paper. A more comprehensive analysis of parameter uncertainty would significantly strengthen the study.

Model validation with observational data: The model validation against observed data appears limited. For example, the absence of independent river-flow validation at interior points of the basin is a limitation. Relying on a single outlet control point for such large catchments may mask the model’s ability, or inability, to reproduce flow dynamics in locations within the basin unseen during calibration. Additional validation would help demonstrate the robustness and transferability of the calibrated model. Are the calibrated parameters spatially varying across the basin or constant in space? How this would influence the simulated results?

Artificial influences of the hydrological regime: Regarding river flow, the authors should also consider discussing whether artificial reservoirs, hydropower plants, or other human interventions influence the hydrological cycle in the study area, and, if so, how these anthropogenic factors are represented or accounted for in the model.

Specific comments:

Line 118: How are these sub-basins defined? Are they delineated based on geomorphological attributes, land-use/land-cover, soil, geology, or another criterion?

Line 137: It appears that S1 has not been introduced yet.

Line 145: Why is GLEAM v4 not used? It offers higher spatial resolution and extended temporal coverage.

Lines 156–165: Could you provide more details on the methodology? When you refer to “summary data,” what exactly does this include? Are these spatially varying discharge time series for each sub-basin? At what temporal resolution?

Line 166: Could you clarify the elementary spatial unit used by the model for the different hydrological processes? Is it a 0.1° pixel?

Line 178: “Ice melting was simulated only…” On what basis are these grid points classified? Is this classification time-varying? Why is ice melting simulated only for these pixels and not for all pixels where snow is present?

Line 190: How are the initial conditions defined for standard grid points and for glacier grid points?

Line 225: Could you specify which snow and glacier model parameters are used and how many there are? Which parameters are fixed (and at what values), and which ones are calibrated?

Lines 245–250: Please define P and its units. Also, specify the parameter ranges and provide references for CF in methods adj1 and adj2, as well as for the parameters in Equation 5.

Line 253: The calibration period is three times longer than the validation period. Could you discuss or justify this choice? Reducing the calibration period could have allowed validation of the glacier module, which represents one of the novelties of this work.

Line 265: All calibration scenarios should produce the same output. In some cases these outputs result from uncalibrated parameters, while in others they come from calibrated ones. It appears that the reliability of the uncalibrated parameters is not assessed against reference data. Is there a reason for this?

Line 266: Could you specify how many parameters are involved in each scenario?

Line 268: Is glacier melt simulated in Scenario 2? If not, how do you justify this?

Line 312: You might consider beginning the results section with a figure or table from the main text, or alternatively moving S1 to the main text if it is essential.

Line 423: Regarding the term “lumped,” does this refer to the spatial basin average? Could you provide a spatially distributed quantitative metric (e.g., r, R², or bias) to assess how modelled and reference AET compare across the basin?