Significant uncertainty exists about how increasing temperatures and changing precipitation patterns will affect Arctic hydrological systems and, in turn, freshwater exports and associated biogeochemical fluxes to the oceans. Arctic hydrology is characterized by strong coupling between flow and thermal processes and is generally not well represented in Earth System Models. Rawlins and Karmalkar use the process-based model PWBM and two climate projections in a strong-warming scenario to assess changes in river flows across the Arctic. The study is well-designed and carefully executed, the manuscript is clear and well written, and the results will be of interest to the readership of TC. However, I have one concern/question that needs to be addressed.

Main question/concerns:

In Figure 1, which addresses model confirmation/evaluation for ALT, we can see good agreement in the mean between TPDC and PWBM forced by W5E5 (Figure 1d) but comparing 1a and 1b visually, it looks like the PWBM model is predicting significantly shallower active layer over the northernmost permafrost zone and deeper active layer in the southern parts of permafrost zone. In other words, TPDC and PWBM are producing very different trends in ALT with latitude. The fact that the two produce similar mean values is not an adequate criterion for judging the reliability of the model. An image showing the spatial distribution of the differences is needed here (e.g. like 1a and b, but for differences between TPDC and PWBM). In addition, a better metric would be the something like root mean-difference or similar metric that integrates differences across the permafrost zone. An explanation for the differences and the different trend is needed. If there is independent information available that could lend further support to the model result, that would help build confidence in PWBM’s ALT calculation.  

Other comments:

The manuscript has a good summary of the PWBM at about the right level of detail, but neglects one important piece of information: what is the spatial structure? I presume it’s not fully 3D, but a collection of independent columns with parameterized landscape runoff and routing through a river network? A brief description would help.

It would be useful to know what fraction of the contributing area for the major rivers comes from non-permafrost regions. This information would allow the reader to judge whether the results are coming mostly from trends in precipitation or from deepening of the ALT in a warming climate.

I’m not sure what is meant by “seasonally maximum ALT” in Figs 1 and 4 as ALT is already the annual maximum thaw depth. Isn’t this just ALT?

The manuscript correctly notes that subsidence, which is neglected in the model, may result in more discharge. It may also be worth noting that the cited modeling study by Painter et al. (2023) was specific to polygonal tundra so the effect on large river basins will depend on the fraction of those basins that contain polygonal tundra.

In this manuscript, the authors simulate frozen ground and streamflow across a pan-Arctic domain.  They simulate increased runoff, higher proportions of subsurface flow and a loss of permafrost.  A notable finding is a regional shift in runoff towards more northerly locations, which have higher amounts of soil carbon.  This leads to the conclusion that there could be enhanced carbon fluxes to the Arctic Ocean.  The paper is well written and presented.  I have several suggestions, which are embedded in the attached document, so I will only summarize in these comments.

The only major concern I have is the validation of the representation of frozen ground.  It is unclear how the authors represent discontinuous permafrost within a model grid.  Furthermore, I am not sure how the simulated ground thermal state is compared to observations.  I struggle with using permafrost class to validate simulated active layer thickness.  A better model validation approach is necessary.  

Could I suggest the authors present panels of differences (i.e., GLEAN subtract PWBM) in Figures 1 and 2 to show how different the model is from observations?  This would reveal where uncertainty is highest.  This is important because one of the main takeaways is the importance of regional differences in responses to warming and the impact this has on freshwater fluxes to the Arctic Ocean. 

Further to this point, the conclusion that the model performs well is based on an assessment of the model's performance over the entire domain.  This is why I am suggesting those extra panels in Figures 1 and 2 as they will help in assessment of uncertainty across sub-domains.

Finally, I like the paper and what the authors are trying to accomplish.  I am hoping that addressing these suggestions will make the conclusions in the final version of the paper a bit more defensible, and impactful.  On this note, I think the conclusions could be even punchier.  Please see my suggestions in the marked-up version.

Thanks for the chance to review the paper.