Mitigating the Effects of Sea-Level Rise on Estuaries of the Mississippi Delta Plain Using River Diversions

Round 1

Reviewer 1 Report

This article describes the results of simulations that examined different Mississippi river management regimes under assumed future sea-level rise. The manuscript was clear and well-written, the results are relevant and well-explained. There minor errors that need to be corrected.

There are minor style errors throughout the document, for example:

Line 135. “future  condition  is available  Brown et al.,  2017”, should say  “future  condition is available in Brown et al., [citation number]”

A detailed review is needed of all the text prior to publication.

Additionally, more details are needed in the methods regarding the ICM. Although the model has been described in publications previously, at least a brief overview is needed. The first paragraph of the Supplementary material (ICM Overview) in addition general model inputs and outputs would help the reader understand the methods and results.

 

Author Response

Please see attached for response to reviewer comments.

Reviewer 2 Report

This paper is based on work conducted for the 2017 Louisiana Coastal Master Plan. There does not appear to be much new modeling here, but it is an application of the models already presented in the Master Plan. It is unclear whether the ICM model work has been formally peer-reviewed (it is not in the list of peer-reviewed sections in Appendix C, p. 10), yet it is cited in this manuscript in lieu of describing/validating the modeling in detail. The models themselves, which are described in Appendix C3 of the Master Plan are somewhat opaque and validation is extremely limited and without discussion of statistical probability that I can find.

Furthermore, the model runs in this manuscript depend on a series of assumptions inherent to the ICM which may propagate model error along the way, yet no attempt is made to identify an error term. Because there is no validation presented in this paper, and no error or uncertainty terms, it is impossible to assess the merit or significance of the outputs, or if the forecasted changes over time or between scenarios are beyond the margin of error. Unfortunately, only one sea level rise scenario is used, and so model sensitivity to this parameter is also unknown. Further complicating this, as the authors acknowledge, would be changes to precipitation/discharge and continued erosion of barrier islands and marshes which would further modify the hydrology of these systems. Additional changes or influences that are likely but not discussed are changes to riverine sediment concentration or feedbacks from increased nutrient loading.

According to RSME values presented in Appendix C3-23 (p 17), error within the various salinity categories is as high as 200% at 0.1 ppt for daily values, and 125% for 0.1 ppt at the annual level. These values are aggregated across all sites, indicating that some are in excess of these values. These RMSE values are large enough to push results into either adjacent salinity category, yet the results are presented as absolute values with no quantitative assessment of model accuracy. The anticipated changes due to sea level rise and diversion scenarios are probably smaller than these error values in at least some cases, but without error terms a reader cannot evaluate the merit or significance of this work.

Other comments:

Figure 7 is missing.

Figure 11 refers to the “receiving compartment” which is not shown or described, and so the sedimentation values, while large, are impossible to interpret and their significance at the coast-wide scale cannot be assessed.

The inclusion of fisheries effects in the discussion fails to mention any fisheries of commercial or recreational significance.

Author Response

Please see attached for responses to the review comments.

Author Response File: Author Response.docx

Reviewer 3 Report

I have only a few formal comments on the paper. It would be advisable to modify the abstract- to emphasize the importance of the topic.

In the methodology it would be good to add more about icm.

The results are very successful - just a few details: the quality of Figure 10 is very low.

In conclusion, it would be useful to add the effect of the changes, not just the general statements.

Author Response

Please see attached for response to review comments.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

While the authors have made several improvements to the manuscript in response to reviewer comments, there remain several unaddressed issues and I still cannot recommend this for publication. Some important details were added regarding salinity calibration, and the graphs in the supplemental material are helpful. They should be in the main manuscript. The table with mean bias and mean RMSE isn’t very useful though, as they could suggest to the reader that errors are smaller than they actually are. This table also appears to be filtered to a subset of sites with bias < 1ppt for some reason or the caption is unclear. The majority of results are presented in units of area which makes error even harder to interpret. If I am reading it correctly, only 76% of models met this goal (therefore 24% of models failed to meet this goal) and 20% of models have a bias greater than 1 ppt (how much greater is not discussed or difficult to find). Despite this addition to the methods, the results still do not contain error terms or discussion of error and the modeled changes in salinity regimes are potentially well within these error terms.

Regarding the issue of expressing results as a percentage or absolute ppt value, I would agree with the authors if the results were binned linearly. Since the authors chose to bin the results exponentially (supported by biologically meaningful thresholds), it seems reasonable to reflect the error as a proportion of bin size. Zizaniopsis, for instance, would be very sensitive to the 1 ppt error term, while Sporobolus (Spartina) would be largely unaffected by this bias. Therefore, the models perform worse as conditions get fresher—exactly what is being modeled. Furthermore, the appendix cited by the authors (C3-23) discusses error in terms of percent as well as ppt.

The discussion section on fisheries effects, while improved, is still weak and the manuscript would be stronger with this section removed. The authors are correct that it is likely complex. As is, the terms used to describe potential impacts are too vague to be of much use in restoration planning or interpreting the significance of the results. A species “shift” does not tell the reader much. Is the habitat better or worse for a given species? Is the population larger or smaller?

I am glad to read that the authors are preparing a manuscript that will address model performance/error. While the authors state that they believe land area modeling error is outweighed by sea level effects, they also state that they do not currently have the capability to assess this regarding salinity. If discussion of error and uncertainty is too lengthy for this manuscript, it should be published first, followed by the manuscript at hand.

I recommend that the authors conduct more thorough error assessment, and present results in the context of this error (i.e., 500 km² +/- 95% CI). As stated in my first round of comments, the results are somewhat meaningless without an error term. It is impossible to determine the significance of the modeled changes without an error term. If this needs to be a separate manuscript then it should be published first, followed by this manuscript.

The authors are presenting a case for how to spend 50 billion dollars on restoration projects, and how to protect communities and culture, yet decisionmakers are supposed to rely on models with large and/or uncertain error terms because it is “unable to be satisfactorily discussed in a mere paragraph”?