Implications of model selection: a comparison of publicly available, conterminous US-extent hydrologic component estimates

Saxe, Samuel; Farmer, William; Driscoll, Jessica; Hogue, Terri S.

doi:https://doi.org/10.5194/hess-25-1529-2021

Articles | Volume 25, issue 3

https://doi.org/10.5194/hess-25-1529-2021

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

https://doi.org/10.5194/hess-25-1529-2021

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

Articles | Volume 25, issue 3

Research article

|

26 Mar 2021

Research article |

| 26 Mar 2021

Implications of model selection: a comparison of publicly available, conterminous US-extent hydrologic component estimates

Samuel Saxe, William Farmer, Jessica Driscoll, and Terri S. Hogue

Download

Final revised paper (published on 26 Mar 2021)
Preprint (discussion started on 05 Jun 2020)
Supplement to the preprint

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'Review Saxe et al. (2020)', Janneke Remmers, 02 Jul 2020
- AC2: 'Response to RC1', Samuel Saxe, 14 Sep 2020
RC2: 'Review of Saxe et al. (HESSD, 2020)', Nans Addor, 19 Aug 2020
- AC1: 'Response to RC2', Samuel Saxe, 14 Sep 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) (23 Sep 2020) by Adriaan J. (Ryan) Teuling

AR by Samuel Saxe on behalf of the Authors (10 Nov 2020) Author's response Manuscript

ED: Referee Nomination & Report Request started (10 Nov 2020) by Adriaan J. (Ryan) Teuling

RR by Nans Addor (19 Nov 2020)

RR by Janneke Remmers (02 Dec 2020)

Suggestions for revision or reasons for rejection

The authors have worked on a revised version of the manuscript. Although, they have incorporated most of the feedback me and the other reviewer have given and the methods and results section have been improved, there is still a lot left to improve in terms of structure. As a reviewer who focuses on the scientific content in papers I have one more point:

1. Regarding section 2.3.4, section 3.2.5 and also Fig. 8, I made a remark concerning the use of imbalances during the last review. This was not addressed. I said the following:

“Concerning the imbalances, the terminology of ‘imbalances’ is proper, as opposed to using ‘residuals’. The implicit assumption here is that the residuals, i.e. model uncertainty, are considered constant for every case study. Clark et al. (2008) show that model uncertainty differs between climates. Especially, more arid climate have higher residuals. Thus, the analysis based on the relative imbalance can only be semi-reasonably compared within 1 ecoregion. I do think imbalances is an interesting concept to use.”

Clark, M. P., Slater, A. G., Rupp, D. E., Woods, R. A., Vrugt, J. A., Gupta, H. V., ... & Hay, L. E. (2008). Framework for Understanding Structural Errors (FUSE): A modular framework to diagnose differences between hydrological models. Water Resources Research, 44(12).

In the current version of the paper, this point is still valid, because this assumption is not mentioned in section 2.3.4. However, the authors still compare the different ecoregions in section 3.2.5. The results showed in the subplots of Fig. 8 are compared in section 3.2.5.

As a reader, I also have several suggestions that you can think about to improve the structure and readability of the paper:

1. In my opinion, the abstract is still too long. I noticed you have shortened it already, well done. At the moment the results are listed in detail. Perhaps, it is possible to summarise this more. It would be great if you can write the abstract in one paragraph.

2. The introduction is 4.5 pages. This is quite long. When I first read the introduction, my attention waned from line 70 onwards and I skipped ahead till the end of the introduction. For me, the introduction became interesting again from line 168. The section that I skipped contained a lot of background information regarding what is already done in literature. I understand that for a CSA study, it is important to also show what the current state of affairs is in literature. So, I agree that this should be placed in the article. I just wonder if this should be placed in the introduction. I would recommend to add another section after the introduction that details this information: “Current state of affairs in literature” or some other title. Since this new section would still be about 3 pages long, I would also recommend to divide this new section into several subsections to increase readability.

3. For Fig. 7, thank you for the explanation. I understand now why there is no plot for runoff (no satellite data available). I would still recommend moving SWE to the bottom and removing the empty plot. I think leaving this in would only raise questions. If there is an empty blank space in the bottom right, it will not disrupt the overall look of the figure as much.

4. Concerning Fig. 8, I agree with the reasons (the graphs are already information dense) the authors have given for leaving out 2 ecoregions. What I still do not understand though is the threshold for excluding these two, since the difference between the “Southern semi-arid highlands” ecoregion, which is excluded, and the “Temperate sierras” ecoregion, which is included, is only 0.29% of the whole CONUS land surface. Can this be clarified further?
The authors could consider the following. Since the main focus of section 3.2.5 is the four biggest ecoregions, Fig. 8 can just contain the results of the four biggest ecoregions and then the figures for the other 6 ecoregions are moved to the appendix. Because at the moment the results of the two smallest ecoregions are not included anywhere. I do assume the authors have those results too. As the authors want to provide an overview that is as complete as possible, I think it would be logical to include the graphs concerning the ecoregions that are (currently) excluded from Fig. 8 in the appendix.

5. Finally, in Fig. 2 panel b) Coefficient of variation has an ‘a’ in front of it in the graph. This should be changed to a ‘b’.

Kind regards,
Janneke Remmers
Wageningen University

Hide

ED: Publish subject to minor revisions (review by editor) (04 Dec 2020) by Adriaan J. (Ryan) Teuling

AR by Samuel Saxe on behalf of the Authors (16 Dec 2020) Author's response Author's tracked changes Manuscript

ED: Publish as is (11 Jan 2021) by Adriaan J. (Ryan) Teuling

Short summary

We compare simulated values from 47 models estimating surface water over the USA. Results show that model uncertainty is substantial over much of the conterminous USA and especially high in the west. Applying the studied models to a simple water accounting equation shows that model selection can significantly affect research results. This paper concludes that multimodel ensembles help to best represent uncertainty in conclusions and suggest targeted research efforts in arid regions.