Please see my comments in the attached pdf.

General comments:

This work is highly valuable for the land surface modeling community because 1) improving the representation of subsurface hydrology remains a critical challenge for LSMs, and 2) computational efficiency must be carefully considered especially when applying to larger scale applications. This multiscale approach addresses both needs by enhancing subsurface hydrology representation while maintaining computational feasibility, which is commendable. The manuscript is well written with sufficient details in methodology and experimental results, but I do have a few comments, which I hope the authors could address to provide a clearer picture of how the proposed approach may benefit the community.

Specific comments:

1. L117: Is a 40-year spin-up sufficient for the groundwater system to reach equilibrium? Please include a justification for this chosen spin-up duration.
2. L166-168: Since Noah-MP can simulates recharge, water table change and baseflow vertically, it is unclear why these processes are only highlighted when combined with HydroBlocks. Please clarify.
3. L222-225: Is the choice of the parameter k sensitive to the size of the study domain and/or the size of the macroscale polygon? For global-scale application, what considerations should be made in selecting k and polygon sizes? Additionally, how would these choices affect computational cost?
4. L342-343: Why do the authors not include the validation for groundwater levels? Given that the lateral subsurface flow scheme likely influences groundwater recharge and can be remarkable for areas with high elevation difference and with intensive human water management, this variable is critical – particularly for applications in regions with intensive groundwater pumping for irrigation.
5. I think it would be beneficial to include a comparison to Noah-MP without the tiling framework. This would provide a clearer picture of how the proposed modeling framework compared to native Noah-MP simulation without lateral flow, in addition to its comparison with the finer HydroBlocks benchmark. Such comparison would inform regional or global modelers considering lateral flow integration.
6. Figure 12c: Could the authors clarify why the spatial pattern of subsurface flow divergence does not follow the west-east precipitation gradient as described in Figure 1? Although Figure 1 shows more like a climatology representation while this comparison only focuses on a single year, it might be helpful to display the precipitation distribution for the specific validation year alongside this subplot to illustrate how meteorological forcing affects subsurface flow divergence.
7. Figure 14: It is unclear why the impact of spatially averaged seasonality on soil moisture in the multiscale approach appears much greater than that on runoff.  I would expect there is large difference on runoff as well,especially under the assumption of baseline-controlled meteorology. Can the authors elaborate on this result?
8. L518-519: Are there in-situ observations available that could be used to evaluate whether the multiscale approach yields improved representation of streams and river networks compared to the benchmark?
9. Since one of the goals for the development of this approach is to preserve computational efficiency, it would be helpful to provide detailed information on the computational cost of the multiscale approach relative to the HydroBlock benchmark, as well as the Noah-MP simulation. Perhaps section 4.3 could be expanded to discuss performance metrics. Moreover, offering suggestions/comments on how this multiscale approach could be applied in regional (e.g., CONUS) or global simulations – including appropriate resolutions and parameter recommendations – would enhance its practical relevance. Currently, the discussion lacks clarity on how the multiscale approach compares computationally with either HydroBlocks or traditional 1-D LSMs like Noah-MP.
10. L566-568: It remains unclear to me whether the multiscale approach shows advantage in simulating subsurface dynamics and groundwater system as, 1) there is no evaluation against observations and 2) there is no investigation on the simulation for groundwater storage or water table depth. The authors may consider providing additional insights or evidence to earn this point.