Dear Authors
Thank-you for your revised manuscript. As you know, both reviewers suggested revisions. As both were willing to see a revised manuscript, I expect to send it back to them. However, it seems to me that the revision rather ignores the reviewer 1 comments about the regression model: linearity; correlation between upstream discharge and level; how the tide is accounted for (implicitly in Zdown).

I think the final manuscript should be self-contained, incorporating a summary of your responses to reviewers so that (i) readers who have the same questions as the reviewers can also see an answer within the manuscript and not have to search among the comments and responses, (ii) the reviewers can see that you have improved the manuscript by incorporating their comments where appropriate. [Especially, Reviewer 1 wanted “major revision” but I do not think that your manuscript changes in view of Reviewer 1 comments amount to major revision.]

Therefore I am asking for you to revise your manuscript again, so that the manuscript itself takes proper account of the reviewers’ comments, including those mentioned above and below, before I send it back to the reviewers.

Yours sincerely
John Huthnance (editor)

Detailed comments.

Lines 119-120. “(partially influenced by the dynamics of river discharge debouched from upstream tributaries . .” You have responded to referee’s comment in your comments but this still needs explaining in the manuscript. Also your comment “To account for the influence of residual water level slope, in the previous manuscript we have explicitly introduced the Zup into the regression model” does not explain how Zup is related to slopes upstream of the inflow.

Equation (3) tends to identify “tidal” Pt with Zdown. Indeed Zdown will include tides, but it will also include effects of “wind, ocean-temperature and ocean-salinity” (Reviewer 1). The last paragraph of section 4 should discuss these other influences.

Section 4.1. The section heading suggests that you should discuss here the error bars in the coefficients (figure 3). The errors relate to the Reviewer 1 comment about correlation between upstream discharge and level reducing the determinacy of the regression coefficients. The manuscript lacks this link, so does not reflect your response to the comment.

Section 4.2. This still starts “. . interpolation . . by means of piecewise cubic Hermite interpolants”; I do not see any change in the revised manuscript relating to the Reviewer comment about consequent errors in estimated slopes, or relating to your response. Anyway, how does the Matlab “gradient.m” function estimate / interpolate slopes?

Data availability. Please see https://www.ocean-science.net/policies/data_policy.html and comply with the “Statement on the availability of underlying data” there.

Dear Authors
I now have both referees’ comments on the latest version of your manuscript. They are favourable overall but there are some “Referee comments” which I copy below, as well as some of my own, for you to address in a revised manuscript please.
Yours sincerely
John Huthnance (Editor)

Referee comments (prefaced by line numbers).
95 2.2 Datasets. In the study both discharge and water level at the upstream station is used, which implies that there is are continuous measurements of both the stage and flow velocity. Yet measurements methods can change over time, and even if they do not change, they require frequent recalibration to account for morphological changes at the gauging station. Therefore, it would be insightful to provide some information on how discharge at the Datong station is measured, if the method of measurement changed during the study period, and most importantly, if it was regularly updated to account for scouring of the bed, after the TGD had been constructed.
116-117 standard deviation function → standard deviation
120-121 ”daily averaged water levels observed at the DT hydrological station are not uniform for identical river discharge” → “There is no unique stage-discharge relation at the Datong hydrological station”
121-122 ”due to the influence of external forcing [...]”. A potentially important factor, the stage-discharge hysteresis, is not mentioned. Is it not relevant at Datong? I suggest to provide a rough estimate of the stage-discharge hysteresis.
139 variance function → variance
176 Note that Gezhouba is also a run-of-the-river dam, and therefore should not considerably influence the discharge regime.
212 increased → increasing
215 “The standard error [...] represents the standard deviation” → “The error-bars [...] represent the standard error”; The standard error and standard deviation are related but not identical (serr ∝ sd/√nsample).
217 is robust → is fitting well. [”Robust” in statistics implies that a method to suppress outliers was employed, which is not the case here.
215-217 ”[...] the standard error [...] suggests that the proposed triple linear regression model is [fitting well] with limited uncertainty”. Remove the qualifier ”with limited uncertainty”, as a good fit does not imply low uncertainty. In general, the goodness of fit to the measured values improves when more parameters are added, but the reliability of predicting values at moments for which no measurements are available decreases (overfit). (See also my recommendation in the previous revision to validate the model through bootstrapping.)
312 constant value of local mean sea level → constant mean sea level
310-313 ”[The] channel deepening [...] tend[s] to increase in the landward direction [..]. This phenomenon can be primarily attributed to the constant value of local mean sea level or the ultimate base level that the topography tends to approach due to erosion.”
I cannot follow this argument, as the constant sea level in combination with the seasonal discharge variation promotes, not prevents, scouring c.f. theoretical work by (Lamb et al., 2012) for the Mississippi and measured longitudinal river profiles of the Mahakam (Sassi et al., 2012) and Kapuas (Kästner et al., 2017). I propose two alternative hypotheses: First, reduced sediment supply initially just results in scouring downstream in the vicinity of the dam, after which the scour slowly propagates further downstream with time. Second, the reduction of seasonal discharge variation by the TGD reduces the overdeepening near the sea.
323-328: Since the TGD continues to deprive the Yangtze of sediment, it is reasonable to assume that the scouring will continue. Can the authors hypothesize how the water levels will evolve in future?
This also points to a potential methodological limitation of the study, as the mean conditions are treated as if they were stationary before and after the dam construction, while the geometric influence has likely gradually increased since construction of the dam due to ongoing scouring.
References
Kästner, K., A. J. F. Hoitink, B. Vermeulen, T. J. Geertsema, and N. S. Ningsih, Distributary channels in the fluvial to tidal transition zone, Journal of Geophysical Research: Earth Surface, 3 (122), 696–710, 2017.
Lamb, M. P., J. A. Nittrouer, D. Mohrig, and J. Shaw, Backwater and river plume controls on scour upstream of river mouths: Implications for fluvio-deltaic morphodynamics, Journal of Geophysical Research: Earth Surface (2003–2012), 117 (F1), 2012.
Sassi, M. G., A. J. F. Hoitink, B. Brye, and E. Deleersnijder, Downstream hydraulic geometry of a tidally influenced river delta, Journal of Geophysical Research: Earth Surface, 117 (F4), 2012.

Editor comments (prefaced by line numbers)
125. “extern” → “external”
129-130. “In this study, the DT hydrological station was chosen as the upstream end, while the TSG gauging station was used as the downstream end.” I think this sentence should be moved to before the present line 120 where DT is referred to but the reader does not presently know that it gives Zup.
186-188. Not a sentence – no verb! Maybe (line 186) “. . (see Figure 2) for . .” although this makes a long sentence.
192. Omit “accounting for” which tends to suggest that the model only accounts for 4-13% of the standard deviations.
207-208. “(estimated using the Matlab ‘regress.m’ function with 95% confidence intervals)” is unnecessary, it repeats the figure 3 caption.
Equation (3) tends to identify “tidal” Pt with Zdown. Indeed Zdown will include tides, but it will also include effects of “wind, ocean-temperature and ocean-salinity” (Reviewer comment on earlier manuscript). The last paragraph of section 4 should discuss these other influences.
232-239. I think the Reviewer comment (on the earlier manuscript) about errors in estimated slopes is still relevant. See the reviewer comment above about lines 215-217; the figure 5 plots of slopes show signs of over-fitting. Anyway, how does the Matlab “gradient.m” function estimate / interpolate slopes?
315-316 and figure 9. The text and figure agree but Z1 as original and Z0 as new is confusing especially as they give ∆Z = - change in Z.

Dear Authors
Thank-you for your (re-)revised manuscript. I think there are still a few points that need clarification; please see specific comments below. Please also remember that on final publication all these comments and responses will be available to readers who will be able to see whether comments have been responded to appropriately. Accordingly I am asking for Minor Modifications.
Yours sincerely
John Huthnance (Editor)

Specific comments

Line 202. Please either omit “,” before “and an increase” or add “,” after “parameter” in line 203. [“after the construction of the TGD” in line 203 applies to all the parameters, not just γ.]

Lines 219-220. The text now reads “. . The error bars in Figure 3 represent the standard deviation of the estimated linear regression coefficients . .” Do you mean “standard deviation” or “standard error”? The reviewer suggested “standard error”. “standard error” seems more likely to me because I suppose that: you only have one estimated value for each regression coefficient; there is no standard deviation which would need many values; you are estimating the uncertainty (possible error) of the coefficient from the residual error in the regression.

Lines 268-270. “while it remained more or less constant during May to June (increasing slightly by 0.01 m), and it generally decreases during the rest of the year by approximately 0.54 m” does not correspond with the appearance of figure 6a. There is much variation during May and June (more than earlier in the year) and 0.54 m must be some sort of average of values that differ greatly with time and location. In line 310 you have 0.46 m not 0.54 m.

Lines 340-341. Please explain “due to the gradually increased geometric influence caused by the TGD” by reference to “ongoing scouring”.

In lines 135-139 you write “It should be noted that the imposed downstream water level Zdown also implicitly accounts for other nontidal factors, such as wind, ocean temperature and ocean salinity, which are assumed to be negligible in the regression model when compared with the tidally induced water level fluctuations featured by a typical spring-neap cycle (see Figure S2 in the Supplementary Material).” I think you should follow this up in the last paragraph of the Discussion section 4; are these other factors really negligible? [How much uncertainty do they introduce to the regression?]

Figure 5. From your present text I infer that the plotted slopes are quadratics deriving from the cubic splines used in the interpolation of elevation between gauging stations. Clearly the splines give continuous values of elevation and slope at each station; however, is there any basis for the variation of slope between stations? The plots of slopes look like over-fitting. Could you get continuous values of elevation and slope at each station using only quadratic splines for elevation?

Figure 6. The labels at the top and the caption suggest that the thick blue curve is ΔTOT at DT in (a), ΔBOU at DT in (b). I think you want “DT ΔQ” –> “ΔQ at DT” in the labels at the top.

Lines 321-322 and figure 9. I am happy with |ΔZ| etc. but Z1 as original and Z0 as new is still strange; usually 0 comes before 1.

Dear Authors
Thank-you for your (re-)revised manuscript. I am now asking for “technical corrections” according to the specific comments below. Please then enter your corrected manuscript into the Copernicus-OS editorial system; it will then be copy-edited and proofs will be created for you to check. Please ensure that your intended meaning is retained. Thank-you for publishing in Ocean Science.
Yours sincerely
John Huthnance (Editor)

Specific comments
Lines 246-248. Maybe replace by “However, caution should be taken with the interpretation of levels and slopes resulting from regression coefficients interpolated by splines: cubic splines might be regarded as over-fitting but smooth quadratic splines were found to introduce greater spatial undulations in levels.” ?
Lines 270-274. I think you mean “. . In January to March, the total alteration ∆TOT averaged approximately 0.28 m over five different gauging stations along the upper YRE, while its average was small (0.01 m) during May to June and negative (approximately -0.54 m) on average for the rest of the year (see Figure 6a and Table 2). . .”