In this study, the authors developed a 5 arc-min spatial resolution dataset for channel routing at a global scale. This dataset is useful, however, I have some concerns referring to the potential application of this dataset. Since the river network data can not be directly resampled, the applications of this dataset will be very limited. I am not surprised that the simulation with higher resolution can have a better performance. It is not a convincing comparison.  

Thus, I can not recommend a publication of this manuscript for this prestigious data journal. I suggest a reject but recommend re-submission after a significant improvement. In particular, I’d like to see different spatial resolutions, including 1/2 degree, 1/4 degree, 1/8 degree, 1/16 degree.

Review

This manuscript describes about the new river network data for river routing models. The new data is developed using the latest river topography dataset MERIT Hydro, and its accuracy is assessed using various river-related datasets as well as model simulations using CTRIP. I think the manuscript contains adequate description as a data paper, and the estimated accuracy is promising. Given that the river network map is a widely used fundamental information in many hydrology and earth system science studies, I think the manuscript is worth publishing on ESSD, after minor corrections on a few ambiguous parts.

L68: some recent studies provide new upscaled river network based on MERIT-Hydro (see, e.g., Eilander et al. , 2021), they do not necessarily follow a D8 convention, and they do not provide model parameters consistent with the new river

70 network (such as sub-grid topography).

Please carefully review the paper Eilander et al. , 2021. Their IHU method also generated D8 format river network, and it also provide some sub-grid topography info. Thus, some descriptions in this sentence is not correct.

L191: The automatic algorithm of MERIT-Hydro chose the outlet that flows into the Nelson River basin.

Development of MERIT Hydro was done with extensive quality assessment, and some input data such as water mask and elevations are modified to ensure realistic river network. Thus, it is not proper to say “Nelson river is chosen as mainstem by “automatic algorithm”. Rather than that, the developer of MERIT Hydro decided that Nelson River to be the major outlet of the South Indian Lake, considering the existing diversion project.

L203 Fig. 8

Probably it is better to explain that how to treat Lake Hulun is the source of the difference. Lake Hulun is usually an inland lake without outlet, but it is connected to the Amur in flooding year. Whether to include Lake Hulun in Amur basin or not highly depends on the developer’s decision.

L204: This major difference can be neglected since it is within the arid region of the Arabian Peninsula.

This sentence is confusing, I assume the authors are discussing about the Amur basin, but why “Arabian Peninsula” is mentioned?

L328: we prefer here to focus on the routing part and capillary rise as well as floodplain evaporation deactivated.

It must be better to note that floodplain scheme affect river discharge and thus evaluation metrics is also affected. Therefore, there are some uncertainties in stating “increase in evaluation metrics meand river network quality is better”. This point must be discussed.

L406: impacted by the new parametrization

What the authors mean by “new parameterization”? Does this simply mean “new river network map” or does this mean “sub-grid topography parameters”? Please clarify.

L219: which impacts the generation of floodplains and aquifers sub-grid parametrization; 2. the use of observed-based river width for CTRIP-12D.

I assume floodplain scheme is deactivated for these simulations, so it is not reasonable to discuss its potential impacts on simulation here. Also, did ground water scheme considered in this test simulations? Please provide informations.

Review of ‘River network and hydro-geomorphology parametrization for global river routing modelling at 1/12° resolution’ by Munier and Decharme

General comment

In this work, Munier and Decharme reported a new global global-scale river network 1/12°, which was derived from the widely used MERIT-Hydro dataset. High spatial resolution river networks are increasingly important for current/future studies on water resources management, climate impact on hydrological processes (e.g., floods), etc. The updated river network represents a great advance in delineating global stream networks, although it is derived from previous datasets and models. In addition, the authors also derived a set of hydro-geomorphological parameters, which would facilitate future studies on network-based hydrological and geomorphological, or even biogeochemical studies (like greenhouse gas emissions from streams and rivers). This updated river network map with higher spatial resolution is thus quite important and will greatly contribute to the scientific community.  The manuscript was well organized, but some details were missing/lost, which should be addressed (see specific comments below).

My another concern is related to validation and data quality. The authors have tried to compare the new river network with previous network datasets or models. To help readers to follow, it may be clearer to present the comparison results (e.g., % in differences) in a table so that readers can easily find out the improvements or performance of this new river network product. This also applies to the derived hydro-geomorphological parameters, in particular for the groundwater and floodplain components, which warrant further data quality assessments.

Specific comments (with line number):

L117: what’s the difference between pixel and cell?

L137: if I understand correctly, 1000 pixels, if near the equator, is ~8.1 km2. It might be reasonable to assume a headwater stream develops within this area size in temperate regions. But for tropical regions, I’m afraid this threshold is too large (i.e., more than 1 headwater stream has developed in 8.1 km2) while for arid regions, the threshold is too small (i.e., a headwater stream may have not necessarily developed within 8.1 km2).

L150: what’s a D∞, please explain. 

Fig 3: the figure caption is repetitious. It is not necessary to repeat the text already shown in the text.

L160: have you assessed the error resulting from such diversion processing?

Fig 5: This global river network map is nice. But the delineated network results in some regions may be problematic, including Greenland, the Sahara desert, and perhaps the middle east (Saudi Arabia). The high river density in these regions is inconsistent with the real world. Also, in fig 6, why are there rivers in the Great Lakes in the USA/Canada?

L174: change ‘consists in’ to ‘consists of’, also in L353

L228: could river channel slope be estimated for each cell? With only one elevation for a cell, how could the slope be calculated? Please clarify.

L235: change ‘contrarily’ to ‘contrary’

L275: refs??  Also, add ‘a’ before ‘number of….’.

L320-322: references are missing.

L329: ‘nearest’

Review of the ESSD paper on global river network and hydro-geomorphology parameterization for global river routing modeling at 1/12 resolution

This paper uses the DRT method to upscale MERIT-Hydro hydrography datasets to 1/12 degree, and used it for CTRIP streamflow simulations and compare it with a coarse resolution CTRIP run. While this work is very interesting and involves lots of work, I found it lacking sufficient justification to be published in ESSD, as this journal focused more on “data” instead of “model simulation”. First, the 1/12 degree river network data and its hydro-geomorphology data seems to be specifically designed for CTRIP and I am wondering what is the wider use of this dataset for other models. Second, a larger portion of this study is on comparing two simulations of CTRIP runs, instead of focusing on the river network dataset. The authors seem to not have introduced new updates to DRT. So I cannot help asking what is their “data contribution”? It seems an existing method (DRT) was applied to an existing dataset (MERIT Hydro). To justify its publication in ESSD, I think authors will need to make more efforts to describe their contribution to data (instead of to model simulation). So I cannot recommend publish this paper unless these questions are sufficiently addressed.

Here are more comments:

FIG. 15: Can the authors mention it is daily or monthly evaluation? Can you add both daily and monthly evaluation here? Because routing models generally matter more for daily streamflow simulations than monthly. If it is monthly then L385 “clearly shows quite good performances”should be revised a little bit.

Fig. 16: I do not think this figure is separately needed. Because there is not much information in the main text (around L384), it can be added as a subplot to Fig. 15. Otherwise, authors should describe much more about Fig. 16 to justify the use of this figure.

L405 and Fig. 17: why only show stations with KGE > -1? Didn’t CTRIP-12D do better than HD for all KGEs? This is a bit confusing and needs more description.

Also, about the use of NIC, why not simply use KGE differences, and positive KGE means better performances?

Fig. 18: again, this figure has little new information than Fig. 17. L408 says “a closer look at” but I didn’t think this adds much information other than saying the same thing as Fig. 17.

Minor ones:

L93: “consist of”