Dear authors,
as the referees' comments were very positive I am happy to suggest publication subject to minor revisions.
Please submit a revised version of your manuscript following the recommendations of the referees.
Best regards,
Theresa Blume

Dear Authors,

given the positive comments of the referees and your revisions I would be happy to recommend this manuscript for publication.

However, I still see a few minor points that should be taken care of prior to publication.

1) more a question than a recommendation: what do you think about shortening the title to:
"Semi‐automated effective width extraction from time‐lapse RGB imagery of a remote, braided Greenlandic river"?
2) Figure 2: the dashed polygon in the second to last image does not match up with the classified image. It seems like the dashed polygon is both shifted and also covering a larger area than the classified image. Please correct the dashed polygon to avoid mismatch. I would also call it a "dashed polygon" instead of a "dotted polygon" in the text.
3) Figure 4: Please correct the time axis in both plots - the labels at the tickmarks are not equidistant, e.g. between 1st and 2nd tickmark is a difference of 8 days and between 2nd and third there is a difference of 4 days in the upper plot and for the lower plot for example between 1st and 2nd tickmark is a difference of 43 days and between 2nd and third there is a difference of 8 days. Please use a standard time axis here so that data gaps become apparent straight away. Depending on the frequency of the data gaps (which are difficult to assess with the way the axis is currently plotted) it would be helpful to connect the data points with a thin grey line, so that the temporal fluctuations are are more easily identified.
4) how much does the time of day of the images that made it to the stage of being used for the extraction of the effective width vary? I would imagine that there are quite strong diurnal fluctuations in river water levels, so morning vs afternoon images would show quite pronounced differences in effective width. How strong are these diurnal fluctuations and would it make sense to further limit the data set to images taken during a similar period/time of day within the diurnal fluctuation. That way we would avoid showing a morning effective width for day n and an afternoon effective width for day n+1, thus producing a difference between those days that is merely the result of the time of day of image capture.
5) page 4 line 2: delete "an", line 3: replace "can" with "to"
6) Section 3.1: "azimuth angles between degrees were found to produce shadows" - some information is missing here - at which degrees does this occur?

Dear Authors,

there seems to have been a misunderstanding with respect to my comment 2)

2) Figure 2: the dashed polygon in the second to last image does not match up with the classified image. It seems like the dashed polygon is both shifted and also covering a larger area than the classified image. Please correct the dashed polygon to avoid mismatch. I would also call it a "dashed polygon" instead of a "dotted polygon" in the text.

You answered:
"These polygons actually do align- the image is oblique enough for this area to appear much larger in the original version as opposed to the orthorectified version (and this is also why there is much distortion out of the region of interest). However, we do agree that the parellipiped in the upper figure does not satisfyingly map below. We have changed the dashed lines to contour around the protrusions in the fore and background to make this link more intuitive. We have changed ‘dotted’ to ‘dashed.’"

However, I am talking about the last two images in the sequence - please correct the dashed frame so that it matches the outline of the classified image. This way the readers can directly compare the photo with the classification result. Please also make sure that the classified image really is the result of the photo - in some instances it seems that areas that look like water are classified as gravel, so maybe the classified image is from a different point in time than the photo.

With respect to my comment 3) about the issue of diurnal variations in flow and the timing of the images, your response was:
"...the result of the similarity filtering is that images are collected at very similar sun angles, which in effect is at the same time of day each day. This sun angle definition of time of day (rather than actual time) is crucial in the arctic, where lengths of day and intensity of insolation change rapidly from week to week. Thus, this filtering process eliminates this concern."

However, in the corrected time series plots it looks like we do see diurnal variations in the data, ranging from 100 to 200 m in effective width even during the limited temporal window defined by the sun angle limitations. This point should be at least briefly discussed in the manuscript.

Furthermore you state that daily or better coverage is achieved - from Figure 4 it seems like there are quite a few data gaps, so it might be of interest to explain that the method provided data on x out y days of the melt season for both years.

I also just noticed that the caption of Figure 1 does not refer to Figure 1b - please also explain what is shown here.

Best regards,
Theresa Blume

Dear authors,

going over your revised manuscript I noticed that you must have uploaded a previous version that does not contain the changes you are referring to in your response. Please upload the latest version as soon as possible!

With respect to your response: I do not see the stronger diurnal variations at the end of the melt season 1 that you are referring to here "note the greater diurnal variation in melt season 1 during the end-of-season low flow, Figure 4".
Do you mean that they are greater than in the first half of the melt season? I do not see that. The fluctuations at the end of the season are about 100m or less, while in the first half they can also be more than 100m.

Please also mention the fact that there are some misclassified areas in the last image of Figure 2 either in the text or in the caption of the figure.

Best regards,
Theresa Blume

Dear Authors,
given your latest changes and corrections I will now recommend your manuscript for publication in HESS.
Best regards,
Theresa Blume