Lake thermal structure drives interannual variability in summer anoxia dynamics in a eutrophic lake over 37 years

Ladwig, Robert; Hanson, Paul C.; Dugan, Hilary A.; Carey, Cayelan C.; Zhang, Yu; Shu, Lele; Duffy, Christopher J.; Cobourn, Kelly M.

doi:https://doi.org/10.5194/hess-25-1009-2021

Articles | Volume 25, issue 2

https://doi.org/10.5194/hess-25-1009-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/hess-25-1009-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 25, issue 2

Research article

|

25 Feb 2021

Research article |

| 25 Feb 2021

Lake thermal structure drives interannual variability in summer anoxia dynamics in a eutrophic lake over 37 years

Robert Ladwig, Paul C. Hanson, Hilary A. Dugan, Cayelan C. Carey, Yu Zhang, Lele Shu, Christopher J. Duffy, and Kelly M. Cobourn

Download

Final revised paper (published on 25 Feb 2021)
Supplement to the final revised paper
Preprint (discussion started on 05 Aug 2020)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Comments on Ladwig et al. HESS 2020-349', Anonymous Referee #1, 24 Aug 2020
- AC1: 'Reply to referee 1', Robert Ladwig, 27 Oct 2020
RC2: 'Comments to hess-2020-349', Chenxi Mi, 29 Aug 2020
- AC2: 'Reply to referee 2', Robert Ladwig, 27 Oct 2020

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

ED: Publish subject to revisions (further review by editor and referees) (30 Oct 2020) by Anas Ghadouani

AR by Robert Ladwig on behalf of the Authors (30 Oct 2020) Author's response Manuscript

ED: Referee Nomination & Report Request started (18 Nov 2020) by Anas Ghadouani

RR by Anonymous Referee #1 (24 Nov 2020)

Suggestions for revision or reasons for rejection

Thank you very much for the opportunity to relook at this manuscript describing the internal (physical and biological) and external factors leading to interannual variability in the extent anoxia in Lake Mendota. I applaud the authors in using this multi-model approach to address this issue. I think the authors have addressed many of my original issues. However, I think there are two main points that should be addressed and several minor things that would help the reader (see my specific comments).

I think two of the models still need to be examined a little further.
Regression model. I think that using the regression model to evaluate the importance of various factors in the dynamic model is interesting, but I think the authors should also try to use regression as an independent approach to further validate the results of the dynamic model. Normally I would prefer that the regression model ONLY use non-modeled information to make any conclusions; however, I realize that given the frequency of data collection this may be very difficult. Therefore, given that GLM-AED model simulates the physics quite well, I think the authors should run one more regression. That regression would use actual summer average Chl-a instead of GPP in the regression. This would show whether the lake actually behaves like GLM-AED says it does.
My other concern is in the presentation of the results of the deductive model. I think the authors need to tighten this up a bit. First, describe how J(z) is actually computed, describe what we are seeing in Figure 3, use one year for example, and then describe what the volumetric part of this model really means. My first impression was that this model was giving very different results than the other two models. But I can see the volumetric part of the model may also represent variability in productivity and changes in stratification (although this is not described in the Discussion). I think my major confusion here is with the description of the model and Figure 3. Personally, I don’t like any approach where I am interpreting the slope and intercept of noisy data. As data get noisy the slope goes closer to 0 and thus changes the overall interpretation.

Specific Comments that I think would improve the paper:
1. Line- 21. Remove the word “evolutionary”.
2. Line 25 and later. I think the real strength in a regression model is to provide independent information that the dynamic model is simulating reality. See suggestion above.
3. Line 30. Make it read “a measured step upward”.
4. Line 48. There is a decadal shift in anoxia in Lake Mendota, and this should be brought into the final discussion a little better. This may be a major difference in what Snortheim described (line 60).
5. Line 83. Need to be careful here. Just because the model has high frequency output, it may not represent what is really happening in the lake. Empirically evaluating results of dynamic models may describe the mathematical equations in the model, but not how this particular lake actually works.
6. Line 92. You state that you are going to use data driven empirical models to evaluate observed data, that is really a good idea, and I think you need to do this more. Maybe by using Chl-a, you can get to this.
7. Line 96. I think you should add something about the decadal changes in Lake Mendota. Also state this in your Conclusions. Because the models don’t capture it, it suggests something outside of the physics and chemistry is driving it. This is a strength of the overall approach.
8. Line 117. I still have a problem with PIHM-Lake never really being presented in this paper or published elsewhere.
9. Line 134. You have all kinds of nutrient components. I think you need to describe the assumptions you made to not only go from TP and TN to all of them. Why would you double the only thing that you actually measured (Line 137)?
10. Line 139. Didn’t Lathrop present measured/actual loading to Lake Mendota in several papers? Seems funny that those estimates are still not mentioned.
11. Line 150. I think a reference is needed for the 1992-1994 data.
12. Line 180. Describe how J(z) is actually computed and how Jv and Ja are estimated from the slope and intercept of the relation between J(z) and alpha. So does each point in Figure 3 represent a different depth?
13. Line 225. Remove the word evolutionary.
14. Line 273. Is there any way to describe how you really combined simulated DO and measured DO to get the AF? Was the real data always used and then interpolated with simulated data?
15. Line 278 and 324 and 421. Can you make this into two regression models? One the way you did it and one with Chl-a?
16. Line 317. Something to consider for the future. In the regression model add a variable to represent the change in time: 0 for the first half and 1 for the second half. Then you can see if the change was significant.
17. Line 337. See comments above about explaining Figure 3. It would help to state that 0.16 is the average intercept and 0.04 is the average slope from all of the figures. Remove the word respiration, this is what gets confusing. By removing the word respiration, then physics is still in this part.
18. Line 343. Why would you add both pieces to get an estimate of SOD, shouldn’t you only use the 0.04?
19. Line 405. Should reference Table 2. I don’t think your RMSEs are similar to that referenced – they are bit higher.
20. Line 416. Rather than saying the AF has no significant differences, use the model not capturing things after 2010 as a strength and that there are decadal changes occurring in the lake.
21. Line 432. Rather than ignoring the results of the deductive model, add a line here about it representing all volumetric processes including the physics.
22. Line 444. Change to timing and strength of stratification.
23. Line 469. Hopefully Chl-a will show the same results.
24. Line 504. Add But this does show a decadal shift in the extent of AF.
25. Line 517. I really think the volume part of this model includes much of the physics associated with the volume of the hypolimnion and the length of stratification. This should be included. If you don’t it really looks like this model gives a completely different interpretation.
26. Line 533. I really think you are being too hard on GLM. If you calibrated it better you should not have a consistent hypolimnetic bias. It has been shown to work well on many lakes, so I would not criticize it so hard. I really think the biggest problem was not calibrating the phytoplankton, by not doing that it affected many things. I think that is the number one thing for future model development. And the second thing would be trying to simulate the change in phytoplankton that occurred in 2010.
27. Line 578. I don’t see any reason why earlier stratification would cause as shallower thermocline. However, a warmer epilimnion could cause a shallower thermocline.
28. Conclusions. Earlier you mention decadal shifts in the Abstract and Introduction. You found one using your models. This is a strength and should mention that by using GLM-AED you can say it was not driven by the physics, and it is probably driven by the changes in the biology.

Hide

ED: Publish subject to minor revisions (review by editor) (10 Dec 2020) by Anas Ghadouani

AR by Robert Ladwig on behalf of the Authors (18 Dec 2020) Author's response Author's tracked changes Manuscript

ED: Publish as is (20 Jan 2021) by Anas Ghadouani

AR by Robert Ladwig on behalf of the Authors (21 Jan 2021) Manuscript

Short summary

Using a modeling framework applied to 37 years of dissolved oxygen time series data from Lake Mendota, we identified the timing and intensity of thermal energy stored in the lake water column, the lake's resilience to mixing, and surface primary production as the most important drivers of interannual dynamics of low oxygen concentrations at the lake bottom. Due to climate change, we expect an increase in the spatial and temporal extent of low oxygen concentrations in Lake Mendota.