Seasonal drought prediction for semiarid northeast Brazil: what  is the added value of a process-based hydrological model?

Pilz, Tobias; Delgado, José Miguel; Voss, Sebastian; Vormoor, Klaus; Francke, Till; Costa, Alexandre Cunha; Martins, Eduardo; Bronstert, Axel

doi:https://doi.org/10.5194/hess-23-1951-2019

Articles | Volume 23, issue 4

https://doi.org/10.5194/hess-23-1951-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/hess-23-1951-2019

© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 23, issue 4

Research article

|

11 Apr 2019

Research article |

| 11 Apr 2019

Seasonal drought prediction for semiarid northeast Brazil: what is the added value of a process-based hydrological model?

Tobias Pilz, José Miguel Delgado, Sebastian Voss, Klaus Vormoor, Till Francke, Alexandre Cunha Costa, Eduardo Martins, and Axel Bronstert

Download

Final revised paper (published on 11 Apr 2019)
Preprint (discussion started on 20 Aug 2018)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'Review 1', Anonymous Referee #1, 15 Sep 2018
- AC1: 'Reply to Referee #1: Suggestion for an alternative title', Tobias Pilz, 25 Sep 2018
RC2: 'comments', Anonymous Referee #2, 16 Sep 2018
- AC2: 'Reply to Referee #2: clarification on calibration period and discussion of further GCM employment', Tobias Pilz, 25 Sep 2018

Peer-review completion

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

ED: Reconsider after major revisions (further review by editor and referees) (01 Nov 2018) by Shraddhanand Shukla

AR by Tobias Pilz on behalf of the Authors (13 Dec 2018) Author's response Manuscript

ED: Referee Nomination & Report Request started (20 Dec 2018) by Shraddhanand Shukla

RR by Anonymous Referee #1 (03 Jan 2019)

Suggestions for revision or reasons for rejection

The authors have addressed most of the reviewer comments and have made a number of revisions which improved the manuscript. I recommend that the manuscript is accepted with a few minor clarifications detailed below:

Sect. 3.5 (Page numbers are based on the revised manuscript.)

P 9 line 25 “an regression approach”, an should be “a”

P 9 line 26 “They fitted” should be “they fit” but did you also do this? If so, focus on your application of the approach to make this clear.

P 9, line 27. The sentence starting with “As response variable…” is a little confusing. Maybe a comma after variable and remove “for model fit”. Also, they are other possible predictors, so clarify that these are possible predictors that were examined in this study.

P 9, line 30 – Would help to provide a reference for this a heuristic search algorithm or name the particular program that was used. Does the algorithm determine “best” based on a particular goodness of fit metric?

P 9 line 28. This description is a little confusing. Is it really the “effect of each predictor” or the correlation observed in the random forest?

Table 2. Adjusted R2 is a more fair metric to report because it accounts for the number of predictor variables in the regression equation. I realize that this table comes from the Delgado et al paper, but I would switch to adjusted R2 if possible. This is especially important given the large number of predictor variables in these equations. For example, see the discussion here: http://thestatsgeek.com/2013/10/28/r-squared-and-adjusted-r-squared/ Was any cross validation was done with these equations in Delgado et al? To ensure that these equations aren’t overfitting the data, this would be very important.

P 16 How are the predictors related to the performance metrics within the random forest approach?

P 20 line 4 check for missing word

Figure 8. Would be good to expand the caption here and explain how these “Predictor importances for each response variable” were determined and a definition for what they are important for.

Hide

RR by Anonymous Referee #2 (26 Jan 2019)

ED: Publish subject to minor revisions (review by editor) (20 Feb 2019) by Shraddhanand Shukla

AR by Tobias Pilz on behalf of the Authors (07 Mar 2019) Author's response Manuscript

ED: Publish subject to minor revisions (review by editor) (09 Mar 2019) by Shraddhanand Shukla

AR by Tobias Pilz on behalf of the Authors (19 Mar 2019) Author's response Manuscript

ED: Publish as is (27 Mar 2019) by Shraddhanand Shukla

AR by Tobias Pilz on behalf of the Authors (30 Mar 2019)

Short summary

This work investigates different model types for drought prediction in a dryland region. Consequently, the performances of seasonal reservoir volume forecasts derived by a process-based and a statistical hydrological model were evaluated. The process-based approach obtained lower accuracy while resolution and reliability of drought prediction were comparable. Initialisation of the process-based model is worthwhile for more in-depth analyses, provided adequate rainfall forecasts are available.