This manuscript conducted a systematic projection on the runoff and cryospheric elements including glacier, snow and frozen soil in a typical mountainous catchment. Overall, the manuscript is well structured and written and easy to follow. It is suitable for publication in HESS, especially for this special issue. However, I would like to point out two major concerns regarding the uncertainty and reliability of the results.

1. The model validation is poorly conducted. Although the authors claimed that the parameters are adopted from a previous study in this catchment, some results related to model performance should be presented to show the confidence of model. If I understand correctly, the model in this study is the combination of the model in Gao et al. (2022) and the Δh-parameterization. Isn’t there new parameter brought by this module compared to the previous version? Could the -Cryo model simulate something that cannot be simulated by -FS model, and if so, how does the model perform on simulating this additional objective? It is rather easy to simulate the relative change of glacier thickness, but simulating the absolute thickness of glacier is difficult, which significantly influences the conclusions such as the time glacier will disappear. So, again, please present some results to show reliability of glacier simulation. Even though all the simulations are the same with -FS model, some results need to be provided to show the confidence of model.

2. The uncertainty issue is addressed inadequately, although the authors mention it in the limitation section. I understand that this study aims to perform a systematic projection on the mountain cryosphere and hydrology, thus does not discuss much about the uncertainties of model parameter and GCM bias correction. However, I think the authors should at least report the uncertainties from different GCMs, given that eight GCMs are adopted for climate projection. The uncertainty range should be provided for the values in the main text (e.g., L304~314) and Figures (e.g., Figure 4).

Other specific issues:

1. Please adjust the paragraph format to justified.
2. Please provide a table to list the meaning of all the variables in Table 3 and Figure 3, to make them easier to find.
3. There are many GCMs in CMIP6. Why are these eight GCMs selected?
4. I suggest the authors to reconstruct the Methodology section to make it more readable. It would be better to introduce the model first, and then introduce the spatial discretization of the catchment (the first paragraph of the current Methodology section), because the elevation band and HRU is the simulation unit of the model. Besides, more details of Δh-parameterization method need to be provided in the 3.1.1 section. The current description is too general, which is difficult to understand for a reader not familiar with this method.
5. Why a single value is provided for some parameters in Table 2, but a range is provided for others?

Climate change and cryosphere variation pose huge threats to local and downstream water resource security, and economic, social, and ecological sustainable development. However, there is still lack of integrated modeling tools to systematic project future changes of cryosphere and its impacts on hydrological regime in diverse climate change scenarios. Thus, this model projection study for the Hulu catchment in the Upper Heihe river has clear novelty. This topic fits well with the scope of this special issue on “Hydrological response to climatic and cryospheric changes in high-mountain regions”. However, I agree with anonymous Reviewer 1 that the authors need to validate the proposed model and add uncertainty envelope in their results and figures, which are very important for the reliability of model and results. In addition, I have some minor concerns before the manuscript can be considered for publication.

• In the abstract, two climate scenarios (SSP2-4.5 and SSP5-8.5) are not mentioned.
• Table 2. What are the optimized parameter values?
• Some important references are missing in the text. For example, there is lack of reference about the study site; no reference about CMIP6 dataset. Line 216~219, References are needed for the Δh parametrization. Some new model developments for small cold region catchments are not well cited, including but not limited to https://hess.copernicus.org/articles/27/4409/2023/; https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2022WR033363;
• It is better to add a landscape classification map in Figure 1.
• Line 201~204, the equal weighted average method could be more clearly demonstrated by equations. Please give specific functions.
• Some figures are not mentioned in the main text, such as Figure 2, 3.
• Line 235. How did you calculate the snow cover days and snow water equivalent?
• Line 375. How did you calculate the lower limit of permafrost?
• The conclusion can be shortened.
• Some figures have small font size, e.g. Figure 1, 7, 8, and 11.

This work presents projection of future changes in glacier and runoff in the Upper-Heihe River. I have some major concerns about the methods, as follows, which the authors may consider addressing in the revision:

1. Model Evaluation: While the authors claim the model has been evaluated in previous works, the newly integrated (or refined) glacier module could impact simulated runoff. I suggest the authors conduct a comprehensive evaluation of the model in simulating discharge, glacier, snow and soil water.

2. Parameterization of the Δ-h Module: The Δ-h module, originally developed and applied in Alpine glaciers in Switzerland, adopted empirical parameter values from long-term observations of glacier area over Switzerland. Its practical application in China has not been well evaluated. The authors should provide more details on the verification of this module in their study area, as well as on the determination of parameter values.

3. Coarse Spatial Resolution of CMIP Model Products: This study was conducted in a small basin with 23km2, but the CMIP products were only downscaled to a resolution of 0.5 deg, much larger than the basin size. To reduce uncertainty, downscaled inputs at higher resolutions would be beneficial.

4. Uncertainty in the Analysis: The results inevitably imply significant uncertainty from model inputs, parameters, and assumptions. A full assessment of modeling uncertainty is highly recommended.

5. Sharp Decreases in Glacier Thickness after 2040:  In figure 5, both glaciers exhibit a sharp decrease in thickness but only small changes in volume after 2040. More explanation is needed, as such sharp changes in climate are not observed.

6. Change 'day' to DOY in the y-axis of Figures 6a-b.

7. Conclusion: the conclusion is wordy and could be more straightforward and informative.