Next Article in Journal
Local Climate Zones, Sky View Factor and Magnitude of Daytime/Nighttime Urban Heat Islands in Balneário Camboriú, SC, Brazil
Previous Article in Journal
Revealing a Tipping Point in the Climate System: Application of Symbolic Analysis to the World Precipitations and Temperatures
 
 
Article
Peer-Review Record

Effect of Model Structure and Calibration Algorithm on Discharge Simulation in the Acısu Basin, Turkey

Climate 2022, 10(12), 196; https://doi.org/10.3390/cli10120196
by Harun Alp 1,*, Mehmet Cüneyd Demirel 2 and Ömer Levend Aşıkoğlu 1
Reviewer 1: Anonymous
Reviewer 2:
Climate 2022, 10(12), 196; https://doi.org/10.3390/cli10120196
Submission received: 23 October 2022 / Revised: 29 November 2022 / Accepted: 6 December 2022 / Published: 8 December 2022

Round 1

Reviewer 1 Report

I enjoyed reading this elaborated work on the hydrologic modeling of a catchment with a combination of different model structures and calibration algorithms. There are several issues to be addressed in the revised version.

Firstly the language in the abstract should be polished.

For example, the authors used the verb “show” two times in one sentence.

Line 18: “Results have 18 shown that mHM and CMA-ES combination showed the best discharge simulation performance 19 according to NSE values (calibration: 0.67, validation: 0.60).”

Line 20: “Although statistically the model results 20 were classified as acceptable,…” unnecessarily long sentence.

L99-103: first define research gap then list the objectives. Here it is opposite, naming the “filling the gap in the literature” is a vague statement. Which gap?

This can be an alternative way:

“SUFI2 algorithm has 0.67 and 0.62 NSE 96 values for calibration and validation periods respectively, SCE algorithm has shown better performance with 0.73 (calibration) and 0.79 (validation) NSE values. There have been many studies on more performance comparison of similar and simple models; however, the effect of sophisticated models and global search algorithms on discharge performances in a headwater  catchment has not been studied yet. Selection of model and calibration algorithm is key for discharge simulations in catchment hydrology.

In this study, we integrated three different model structures with three calibration algorithms for Acısu Basin. For that we used PEST toolbox and ERA5 model inputs.”

L99 and 102: avoid repetition of “model type and algorithm”

L118: “upstream region of the Gediz” is that a headwater of Gediz?

Table 1: did you compare era5 and mgm data?

L245: 20 parameters of SWAT?

Figure 1 caption: study domain not location map. You should add representative land cover map and DEM as they are used by distributed models.

Results should include distributed discharge or runoff maps from mHM to show the skills of the model. Currently it is dry and map outputs of swat and mHM can be fruitful for model comparison.

L331: fix the reference

Discussion section: highlight the limitations of the gr4j and swat as compared to the mHM. You did opposite underpinning mHM’s model structure effect (L337-349). This section is meant to list the limitations of the study as well. What could be done but what you did? You could calibrate the mHM with remote sensing data, you could also benefit RS temporal data for gr4j and swat as they need time series and not raster data. You could validate mHM runoff with ERA5 Land runoff outputs of HTESSEL land surface model.

Conclusion L374-384: move to discussion section.

L364-369: remove unnecessary wordings.

L386-389: repetitions of results.

Except last paragraph all conclusion section must be re-written. Bullets can be useful to list main conclusions drawn based on results. No numbers should be shown from results.

 

 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

1

Similar studies have been published, which compared the SWAT and GR4J , Refer to

Open Water Journal > Vol. 5 (2018) > Iss. 1

https://scholarsarchive.byu.edu/cgi/viewcontent.cgi?article=1087&context=openwater

Robustness and performance of semi-distributed (SWAT) and global (GR4J) hydrological models throughout an observed climatic shift over contrasted French watersheds Etienne Brulebois1*, Marjorie Ubertosi2 , Thierry Castel1 , Yves Richard1 , Sabine Sauvage3 , José-Miguel Sanchez Perez3 , Nicolas Le Moine4 , Philippe Amiotte-Suchet1 1 UMR n°6282, CNRS University of Bourgog

 

2 

The difference between this manuscript and existing studies lies in the addition mHM model, which is the comparison among the GR4J, SWAT and mHM. However,  the manuscript has not provide the reference about mHM.  Mike SHE  is a 2D runoff routing model,  calculation, which is recommended to be compared with the above 3 current models(GR4J, SWAt, mHM).

3

 lines 203-204 

“mesoscale Hydrologic Model (mHM) is a fully-distributed, physically based and continuous hydrologic model published by a team from UFZ (Helmholtz Centre for Environmental Research).”,  The manuscript does not provide references, so it is difficult for readers to understand mHM, so it is difficult to understand the causes of errors in the models.

4

In Figure 3, more than 10 symbols and parameters appear, and the manuscript does not explain the meaning of the symbols. 

5

 In Figure 3,  What is the reason for dividing pr into 0.9 and 0.1 ?

6

In Figure 4. mHM cell structure,  the manuscript needs to explain the meaning of symbols

7

Lines 200-201, lines 215-216,

  SWAT and mHM use the Muskingum formulations to determine the runoff routing. Because Muskingum formulations are a 1D equations, which consist of 1D continuous equations and 1D momentum equation. The 1D momentum equation is based on the assumption that the river discharge Q is proportional to the river storage V, V=kQ. The k value has influence on the runoff routing.  How to determine this K value should be given.

 

8

   SWAT is a semi distributed hydrological model and mHM is a distributed hydrological model. However, both of them use the 1D runoff routing formulation, which will cause errors in the calculation. For the reasons of the error, please refer to the:

CREST-iMAP v1.0: A fully coupled hydrologic-hydraulic modeling framework dedicated to flood inundation mapping and prediction, Environmental Modelling and Software 141 (2021) 105051

9

Compare SWAT and MHM models. The cell division of mHM is relatively fine, which can reflect the characteristics of the spatial variability of soil permeability of the watershed.

For SWAT model, please explain the principle of Hydrological Response Unit (HRU) division, to ensure the accuracy of the 1D runoff routing calculation. For the mHM, please explain the principle of the cell division, to ensure the accuracy of 1D runoff routing

 

10

The values in the table need to be marked with unit, such as Table 1

11

The symbols appearing in Table 3 need to explanation their meaning.

The symbols appearing in Fig.6 and Fig.7 need to explanation their meaning

12

The existing models are compared and used to calculate the runoff of the watershed. The manuscript lacks the innovation in the hydrological model. The author needs to understand the cause of the error of the existing models, and then improve the accuracy of the calculation through the calibration. The research of this manuscript needs to be improved.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors have addressed all my comments and I support publishing the manuscript.

Author Response

We appreciate reviewer#1 for his/her contributions and wise suggestions.

Reviewer 2 Report

1

Only part of the suggestions were considered in the revised manuscript(climate-2017182-peer-review-v2.pdf), the author needs to answer the all suggestion one by one. For examples, the numerical error caused by 1D routing formulation based on the Muskingum scheme,  and the author needs to comparison the mHM with Mike SHE model.

2  Lines  219-223

“mesoscale Hydrologic Model (mHM) is a fully-distributed, physically based and con tinuous hydrologic model published by a team from UFZ (Helmholtz Centre for Environ[1]mental Research) [25]. Fundamental numerical approaches about hydrologic processes of the mHM are tested by using well-known and acknowledged lumped models such as HBV [26] and VIC [27] “

Regarding mHM, this revised manuscript  provides a reference [25].  Ref. [25] only introduces part of the formula of mHM in the appendix, it does not explain the meaning of symbols. 

Lumped hydrological  model is a low-precision model,  and Distributed hydrological mode is a high-precision model, and using Lumped model to verify Distributed model is inherently problematic

 

Ref. [25]  Samaniego, L.; Kumar, R.; Attinger, S. Multiscale Parameter Regionalization of a Grid-Based Hydrologic Model at the esoscale. Water Resour. Res. 2010, 46

 

 

3

What is the fully distributed model (mHM) simulated? Hydrological models can be divided into the Lumped model, semi-distributed model and distributed model. How does this fully distributed model differ from the distributed model?

 

4  Liness 215-216  

“ Considering the information given in [24], we used SCS-CN (Soil Conservation Service – Curve Number) method in land phase, and Muskingum method in 217 routing phase”,

Muskingum scheme is a 1D routing formula, and it is assumed that the Channel volume V is  linear with discharge Q, V=KQ,  where k is a coefficient. In the actual calculation, the K value may be not reasonable, which will bring a large error.

5

The present manuscript only compares 3 existing hydrological models, some model concepts are fuzzy, it is lack of innovation.

 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 3

Reviewer 2 Report

The reference given in the revised manuscript has a brief description of the MHM  model,  it is difficult to understand the model.

In the revised manuscript, the existing model are compared, and the Lumped hydrological models are used to verify the distributed model, such as mHM. There may be some problems in the accuracy of the mHM adopted in this manuscriptr. On November 23, it was suggested to use the Mike SHE for comparative verification. The authors answered that the Mike SHE is not opened and is not easy to verify the mHM with Mike SHE.

As for the runoff routing for the watershed, it has been discussed in detail in the literature [xx-1], as shown below. The 1D routing schemes, such as the Muskingum scheme had obvious errors in calculating the 2D routing i of the basin, so SWE ( 2D scheme) was used to compute the runoff routing in the watershed.

For this purpose, I consulted several articles, as shown below. It is suggested to use the following reference example for verification, namely the V-shaped valley example, which is commonly used to verify the rationality of the results of the hydrological model or hydrodynamic model. It is suggested to verify the accuracy of mHM.  After verification, the reviewer will recommend whether to accept the paper or not.

  [xx-1]  CREST-iMAP v1.0: A fully coupled hydrologic-hydraulic modeling framework dedicated to flood inundation mapping and prediction,Environmental Modelling and Software 141 (2021) 105051

 [xx-2]  Kim, J., et al., Coupled modeling of hydrologic and hydrodynamic processes including overland and channel flow. Advances in Water Resources, 2012. 37: p. 104-126.

[xx-3]  Sulis, M., et al., A comparison of two physics-based numerical models for simulating surface watergroundwater interactions. Advances in Water Resources, 2010. 33(4): p. 456-467.

[xx-4]  Di Giammarco, P., E. Todini and P. Lamberti, A conservative finite elements approach to overland flow: the control volume finite element formulation. Journal of Hydrology, 1996. 175(1-4): p. 267-291.

[xx-5]  Panday, S. and P.S. Huyakorn, A fully coupled physically-based spatially-distributed model for evaluating surface/subsurface flow. Advances in water Resources, 2004. 27(4): p. 361-382.

 

Author Response

Please see the attachment (pdf).

 

Reviewer:

Comment 1) The reference given in the revised manuscript has a brief description of the MHM  model,  it is difficult to understand the model.

Reply from authors: The authors thank the Reviewer for her/his constructive comments on the manuscript. We added a detailed presentation on mHM model description to our repository. Here is the direct link:

https://web.itu.edu.tr/demirelmc/mhm_basics_full.pdf

Moreover, the manual of the model is available in UFZ Leipzig gitlab page available directly at this link.

https://git.ufz.de/mhm/mhm/-/blob/5.10_fixed/doc/mhm_manual_v5.10.pdf

 

Multi-parameter regionalization framework of the mHM model is explained in detail in this paper:

Schweppe, R.; Thober, S.; Müller, S.; Kelbling, M.; Kumar, R.; Attinger, S.; Samaniego, L. MPR 1.0: a stand-alone multiscale parameter regionalization tool for improved parameter estimation of land surface models. Geosci. Model Dev. 2022, 15, 859–882, doi:10.5194/gmd-15-859-2022.

The new routing scheme of the mHM model is explained in detail in this paper:

Thober, S.; Cuntz, M.; Kelbling, M.; Kumar, R.; Mai, J.; Samaniego, L. The multiscale routing model mRM v1.0: simple river routing at resolutions from 1 to 50 km. Geosci. Model Dev. 2019, 12, 2501–2521, doi:10.5194/gmd-12-2501-2019.

We hope that we provided enough in-depth good quality sources from top journals and internal UFZ presentation file (pdf) only for his/her attention. We will be happy to provide more information in AGU2022 in Chicago (https://www.agu.org/Events/Meetings/Fall-Meeting-2022). S/he is more than welcome to the UFZ sessions and papers focusing on mHM applications.

 

 

Comment 2) In the revised manuscript, the existing model are compared, and the Lumped hydrological models are used to verify the distributed model, such as mHM.

Reply from authors: We don’t agree with the comment. We did not use lumped model to verify the mHM model. The reviewer is referred to this Science article to use appropriate scientific language for Earth system models. Reading only its abstract will reveal the importance of using appropriate jargon. We evaluated model results from different model structure and calibration algorithm. We compared the model-algorithm performances under fair comparison constraints.

Oreskes, N.; Shrader-Frechette, K.; Belitz, K. Verification, Validation, and Confirmation of Numerical Models in the Earth Sciences. Science (80-. ). 1994, 263, 641 LP – 646, doi:10.1126/science.263.5147.641.

Comment 3) There may be some problems in the accuracy of the mHM adopted in this manuscriptr.

Reply from authors: We don’t agree with the comment. Please avoid using vague sentences. The reviewer is invited to use firm statements instead of using “may be” “can be” etc.

Model calibration is done to improve the model performance (accuracy is again a wrong term in modelling world). Data accuracy and model resolution are correct terms. Models are built based on assumptions, and they are approximations of real earth processes using mathematically and physically meaningful structures. We run the mHM model at 0.015625 degree (~2km) spatial resolution which is appropriate for the study area. We used ERA5 meteorological forcings at 0.25 degree. We also used high resolution soil data at 0.001953125000 degrees (~200m). We can provide all our setup for the reviewer for finding accuracy issues in the set-up and then using unprofessional language to implicitly blame our work and us (the researchers).

 

Comment 4) On November 23, it was suggested to use the Mike SHE for comparative verification. The authors answered that the Mike SHE is not opened and is not easy to verify the mHM with Mike SHE.

Reply from authors: Yes, adding Mike-she to the content is out of scope for this study. The reviewers should evaluate the current state of the manuscript by demanding too expensive extra work.

 

 

Please refer to this page to refresh the obligations for the referees.

https://www.hydrology-and-earth-system-sciences.net/policies/obligations_for_referees.html

 

  1. In as much as the reviewing of manuscripts is an essential step in the publication process, every scientist has an obligation to do their fair share of reviewing.
  2. A chosen referee who feels inadequately qualified or lacks the time to judge the research reported in a manuscript should return it promptly to the editor.
  3. A referee of a manuscript should judge objectively the quality of the manuscript and respect the intellectual independence of the authors. In no case is personal criticism appropriate.
  4. A referee should be sensitive even to the appearance of a conflict of interest when the manuscript under review is closely related to the referee's work in progress or published. If in doubt, the referee should return the manuscript promptly without review, advising the editor of the conflict of interest or bias.
  5. A referee should not evaluate a manuscript authored or co-authored by a person with whom the referee has a personal or professional connection if the relationship has the potential to bias judgement of the manuscript. Please see also our competing interests policy.
  6. A referee should treat a manuscript sent for review as a confidential document. It should be neither shown to nor discussed with others except, in special cases, persons from whom specific advice may be sought; in that event, the identities of those consulted should be disclosed to the editor.
  7. Referees should explain and support their judgements adequately so that editors and authors may understand the basis of their comments. Any statement that an observation, derivation, or argument had been previously reported should be accompanied by the relevant citation.
  8. A referee should be alert to failure of authors to cite relevant work by other scientists. A referee should call to the editor's attention any substantial similarity between the manuscript under consideration and any published paper or any manuscript submitted concurrently to another journal.
  9. Referees should not use or disclose unpublished information, arguments, or interpretations contained in a manuscript under consideration, except with the consent of the author.
  10. A referee should only suggest to the author the inclusion of additional literature if such suggestions are of substantial relevance for the scientific argumentation of the manuscript. Suggestions should be balanced regarding authors, journals, working groups, and institutions.
  11. If referees have any doubts about the conduct of certain individuals involved in the review process, the EGU ombudsperson for publications can always be contacted confidentially for help or advice.

 

 

 

Comment 5) As for the runoff routing for the watershed, it has been discussed in detail in the literature [xx-1], as shown below. The 1D routing schemes, such as the Muskingum scheme had obvious errors in calculating the 2D routing i of the basin, so SWE ( 2D scheme) was used to compute the runoff routing in the watershed.

Reply from authors: We didn’t use Muskingum in our mHM set-ıup. We used this method as explained in this paper:

Thober, S.; Cuntz, M.; Kelbling, M.; Kumar, R.; Mai, J.; Samaniego, L. The multiscale routing model mRM v1.0: simple river routing at resolutions from 1 to 50 km. Geosci. Model Dev. 2019, 12, 2501–2521, doi:10.5194/gmd-12-2501-2019.

 

Moreover, regarding SWAT model, we replied the same question in the 2nd round review at comment 4 as below. We showed the sensitivity of SWAT to the routing parameters. The new Editor-in-Chief of Climate is invited to judge our reply and if necessary a new professional reviewer from hydrologic community must be invited to re-evaluate our manuscript since we face an unfair-unprofessional review processes by a biased reviewer.

Comment #4 : Liness 215-216  

“Considering the information given in [24], we used SCS-CN (Soil Conservation Service – Curve Number) method in land phase, and Muskingum method in 217 routing phases”,

Muskingum scheme is a 1D routing formula, and it is assumed that the Channel volume V is linear with discharge Q, V=KQ, where k is a coefficient. In the actual calculation, the K value may be not reasonable, which will bring a large error.

Reply #4:: We agree with the comment and run SWAT with different values of Muskingum parameters. SWAT+ has three governing parameters for Muskingum method namely “msk_co1”, msk_co2”, and “msk_x”. We analyzed the sensitivity of these values separately considering reviewer’s comment. As it was shown in the table below, making changes in these parameters has no effect on discharge performance. As a proof and to be transparent, we recorded our screen while using SWAT+ with different Muskingum values (https://youtu.be/VXWuvr_A8s4). The highly respected reviewer can watch the video and also try himself in a model set-up in his computer.

Table 1. Muskingum parameter effects on SWAT+

Parameter

Lower Limit

Upper Limit

Value

NSE

msk_co1

0

10

1

0.55

5

0.55

9

0.55

msk_co2

0

10

1

0.55

5

0.55

9

0.55

msk_x

0

0.3

0.05

0.55

0.15

0.55

0.25

0.55

Moreover, we tried to understand whether there is a bug in the model code or the reason of no-reaction to the Muskingum parameters by searching on internet. and found a statement of a model developer (Natalja) about the issue (Figure 1). Although, Ms. Natalja C. indicates that this is a reported issue, there is still no solution at the most up to date version we have used.

Figure 6. Model developer's response to related comment.

 

 

Comment 7) For this purpose, I consulted several articles, as shown below. It is suggested to use the following reference example for verification, namely the V-shaped valley example, which is commonly used to verify the rationality of the results of the hydrological model or hydrodynamic model. It is suggested to verify the accuracy of mHM.  After verification, the reviewer will recommend whether to accept the paper or not.

  [xx-1]  CREST-iMAP v1.0: A fully coupled hydrologic-hydraulic modeling framework dedicated to flood inundation mapping and prediction,Environmental Modelling and Software 141 (2021) 105051

 [xx-2]  Kim, J., et al., Coupled modeling of hydrologic and hydrodynamic processes including overland and channel flow. Advances in Water Resources, 2012. 37: p. 104-126.

[xx-3]  Sulis, M., et al., A comparison of two physics-based numerical models for simulating surface water–groundwater interactions. Advances in Water Resources, 2010. 33(4): p. 456-467.

[xx-4]  Di Giammarco, P., E. Todini and P. Lamberti, A conservative finite elements approach to overland flow: the control volume finite element formulation. Journal of Hydrology, 1996. 175(1-4): p. 267-291.

[xx-5]  Panday, S. and P.S. Huyakorn, A fully coupled physically-based spatially-distributed model for evaluating surface/subsurface flow. Advances in water Resources, 2004. 27(4): p. 361-382.

 

Reply from authors: The authors thank the Reviewer for her/his constructive comments on the manuscript. We read all the literature recommended and checked our routing scheme again. We found no difference in our discharge results. Please note that this is not a hydrodynamic study focusing on inundation areas/maps. This hydrologic modelling study only considers discharge hydrographs for calculating model performances. 2D routing schemes based on Saint Venant equations or 3D ground water movement scheme based on Richard’s equations are out of scope of this paper. If the reviewer thinks opposite, we should call the editor-in-chief and 5 more reviewers to evaluate our statements here. This comment is not relevant for this study.

We again invite the new Editor-in-Chief of Climate to judge our reply and if necessary new professional reviewers from hydrologic community must be invited to re-evaluate our manuscript since we face an unfair-unprofessional review processes by a biased reviewer as a second time. We have to ask how can a review decision from minor to major can happen while opposite direction is logical. A clear biased-mental look insisting irrelevant extra work is happening here.

 

 

Author Response File: Author Response.pdf

Back to TopTop