Review of Godet et al.

Godet et al. provide comparison of methods to allocate river points to the most appropriate hydrological model grids. This task is important and becoming more important given the rise in number of gridded hydrological models being made available within hydrological research and operations. The paper compares three allocation methods: area-based, topology-based, and contour-based. The results indicate that contour-based methods, though computationally expensive, are more hydrologically relevant, with topology-based methods serving as a reasonable compromise. Area-based methods lead to numerous allocation errors, particularly for small catchments, and are recommended only for river points with large upstream drainage areas compared to the grid cell resolution.

I do have some questions for the authors about the transferability of their results outside the current test area in the Eastern Mediterranean region of France covering an area of 15,000 km² with the largest catchment size considered only 3000 km². They define “coarse-resolution” as a 1km hydrological modelling grid size. In the context of global hydrological modelling, 1 km is the benchmark to be deemed “hyperresolution” (Wood et al., 2011). While this paper is over a decade old, there remains relatively few hydrological models running at 1km scale, even at national scales. For a user of a model running at 5km or 10km or even coarser, are the conclusions in Godet et al. still valid? What about transferability to other regions? We know hydrology is heterogenous with complex river networks such as braided rivers; we know that high quality DEMS/vector river networks are not available in all regions, and that the quality of upstream catchment size metadata information can be missing or uncertain in some regions of the world. Very few of these uncertainties are considered or at least discussed in the paper.

There has been a limited amount of research comparing difference approaches to this important technical issue, therefore the paper by Godet et al. is a very useful reference to help guide others on selecting the most appropriate method/understanding the limitations of simpler methods. However, at a minimum I suggest more effort is needed to discuss uncertainties and transferability outside the limited test case used. I recommend this paper for publishing in HESS after such changes are made.

Main comments

• Pg3 L46-48: As per my summary above, from the perspective of gridded hydrological models that are not run at very local scales, then the definition of “coarse-resolution hydrological grid (1km×1km)” could arguably be considered “high resolution”. 1 km is the benchmark to be deemed “hyperresolution” (Wood et al., 2011) for global scale models. To what extent are these conclusions/methods transferable to coarser model resolutions that are often used (e.g. 5km, 10km or coarser)? Perhaps qualifying why 1km is deemed “coarse” and if so, does this limit the transferability of methods/conclusions?
• Pg4, L71: The parameter R (here R < 3) seems to be very dependent on the model resolution and catchment size. Why was R < 3 selected and was a sensitivity done on its selection? How would varying R to be larger or smaller impact the results? It will also be depended on catchment area of the station that you are trying to allocate. For example, for a catchment area of < 3000 km² (as is considered here) then R < 3 might be appropriate. However, if you are mapping river gauges in global gridded models and are considering stations in downstream sections of major world river basins (e.g. Amazon, Danube, etc.), then you would need an R much larger than 3 – this parameter needs to vary by both grid resolution and catchment size.
• Pg4, L74-79: How would method 2: 'topology-based method' work when the underlaying gridded hydrological model river network is different to the vector network. For example, if there are spatial mismatches where the vector for a river section does not overlap with the most appropriate model cell? Often the data source to derive a hydrological model river network grid is different from a vector river network.

Technical comments

• P6, L98-99: Can you please justify use of CSI = 0.4 and 0.6, and are these applicable to much larger catchment sizes?
• Pg 6, L99: “the search area is extended to the 49 closest grid cells”: why 49, please elaborate?

References

Wood, E. F., Roundy, J. K., Troy, T. J., Beek, L. P. H. van, Bierkens, M. F. P., Blyth, E., Roo, A. de, Döll, P., Ek, M., Famiglietti, J., Gochis, D., Giesen, N. van de, Houser, P., Jaffé, P. R., Kollet, S., Lehner, B., Lettenmaier, D. P., Peters‐Lidard, C., Sivapalan, M., Sheffield, J., Wade, A., and Whitehead, P.: Hyperresolution global land surface modeling: Meeting a grand challenge for monitoring Earth’s terrestrial water, Water Resources Research, 47, https://doi.org/10.1029/2010WR010090, 2011.