Dear Ruby Leung,

I am please to say the reviewers were very happy with the revised version of the manuscript and response to their comments. I would like to move towards publication however after reading the revised manuscript I have a few minor comment that might require some clarification in the manuscript. Most importantly, as the main aim of the journal is to present geo-scientific models some more details on the numerical method are needed to improve the fit between the article and the readership.

I have added some specific comments below. Please rebut or make changes in light of all of these comments, after which I believe it will be possible to publish the final version.

Specific comments:

Abstract

The abstract is very long and this section below does not say anything except that you compared the results to 13 gauges and inundation from GIEMS.
“The streamflow hydrographs were reproduced fairly well for the majority of 13 major stream gauges. The river-stage hydrographs were modeled reasonably well for the 11 subbasins containing or close to 11 of the 13 stream gauges. The inundation estimates were comparable to the GIEMS observations.”
I would reduce the abstract length

Introduction

The introduction needs a thorough proof read. I spotted numerous minor errors.

Methods

The description of the inundation solver is very brief for a model description and it is not at all clear from the text how this actually works in conjunction with Figure 1. Rather it appears to simply list three equations (continuity, momentum and Manning’s). Furthermore, the Courant condition is not sufficient to guarantee stability of a diffusive wave model (Hunter, N. M., M. S. Horritt, P. D. Bates, M. D. Wilson, and M. G. F. Werner (2005), An adaptive time step solution for raster-based storage cell modelling of floodplain inundation, Adv. Water Resour., 28, 975– 991, doi:10.1016/j.advwatres.2005.03.007.)
GMD will publish model descriptions actually without the need for substantially new scientific insight, however the description needs to be more comprehensive.
How was the stable time step computed? Usually this is a combination of the smallest computational unit length in the model and the maximum simulated depth. For a diffusive wave model you might need to additionally consider the friction and lowest slope simulated (e.g. hunter et al). The time step seems very long for a diffusive wave model where very shallow gradients are being simulated. The model clearly works so there is perhaps no reason to make a big issue of this, however the time step appears more a pragmatic compromise based on computational cost and the scheme will have problems at higher resolution that should be noted.
In section 2.2 do you consider floodplain flows and if so how. Is a compound channel used or do you restrict flow to the channel only and use the floodplain for storage? If floodplain flow is allowed how was Manning’s n assigned. I think my confusion here propagates from the model description.
The first paragraph of section 2.4 makes no sense to me. Specifically what is the conditioned DEM? Is it the HydroSHEDS DEM before or after you have applied the vegetation correction or something else? Are these non-negligible errors in the final data used for the modelling?
For equation 7 how is the slope found?

Discussion

Page 30 “backwater effects on streamflow” section. I would not claim that the effect of backwatering on streamflow timing hasn’t been discussed in previous studies. For me this is a process that happens in all subcritical flow systems (most lowland rivers) and would be expected once you decide that the backwater effect is important.
In the discussion point 5 do you mean increase inundation extent rather than advance inundation which would suggest an earlier arrival when including backwater. Please check for clarity.
I’m not sure what is added by the paragraph starting “Building on previous studies (Baugh…” It seems to just say some of our conclusions agree and some do not with previous studies. I would either be specific here or just remove the paragraph. There are a few places where you provide quite general impressions of the results that would be stronger if supported by the statistics you calculated on model performance.
Page 35: Use the word more accurate rather than “better”

Dear Ruby Leung,

Thank you for the revised version of the manuscript. All of my comments and those of the reviewers have been addressed and I am delighted to accept the article for publication in GMD.

Kind Regards,
Jeff Neal