Validation of precipitation reanalysis products for rainfall-runoff modelling in Slovenia

Alexopoulos, Marcos Julien; Müller-Thomy, Hannes; Nistahl, Patrick; Šraj, Mojca; Bezak, Nejc

doi:https://doi.org/10.5194/egusphere-2022-1279

Preprints

https://doi.org/10.5194/egusphere-2022-1279

Preprints

06 Dec 2022

| 06 Dec 2022

Validation of precipitation reanalysis products for rainfall-runoff modelling in Slovenia

Marcos Julien Alexopoulos, Hannes Müller-Thomy, Patrick Nistahl, Mojca Šraj, and Nejc Bezak

Abstract. Observational data scarcity often limits the potential of rainfall-runoff modelling around the globe. In ungauged catchments, earth-observations or reanalysis products could be used to replace missing ground-based station data. However, performance of different datasets needs to be thoroughly tested, especially at finer temporal resolutions such as hourly time steps. This study evaluates the performance of ERA5-Land and COSMO-REA6 precipitation reanalysis products (PRPs) using 16 meso-scale catchments located in Slovenia, Europe. These two PRPs are firstly compared with a gridded precipitation dataset that was constructed based on ground observational data. Secondly, a comparison of the temperature data of these reanalysis products with station-based air temperature data is conducted. Thirdly, several data combinations are defined and used as input for the rainfall-runoff modelling using the GR4H model. A special focus is on the application of an additional snow module. Both tested PRPs underestimate, for at least 20 %, extreme rainfall events that are the driving force of natural hazards such as floods. In terms of air temperature both tested reanalysis products show similar deviations from the observational dataset that was catchment-specific. Additionally, air temperature deviations are smaller in winter compared to summer. In terms of rainfall-runoff modelling, the ERA5-Land yields slightly better performance than COSMO-REA6. If a re-calibration with PRP has been carried out, the performance is similar compared to the simulations where station-based data was used as input. Model recalibration proves to be essential in providing relatively sufficient rainfall-runoff modelling results. Hence, tested PRPs could be used as an alternative to the station-based based data in case that precipitation or air temperature data are lacking, but model calibration using discharge data would be needed to improve the performance.

Received: 16 Nov 2022 – Discussion started: 06 Dec 2022

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Preprint (PDF, 1693 KB)

Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
Preprint (1693 KB)

Download & links

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Journal article(s) based on this preprint

12 Jul 2023

Validation of precipitation reanalysis products for rainfall-runoff modelling in Slovenia

Marcos Julien Alexopoulos, Hannes Müller-Thomy, Patrick Nistahl, Mojca Šraj, and Nejc Bezak

Hydrol. Earth Syst. Sci., 27, 2559–2578, https://doi.org/10.5194/hess-27-2559-2023,https://doi.org/10.5194/hess-27-2559-2023, 2023

Short summary

Marcos Julien Alexopoulos, Hannes Müller-Thomy, Patrick Nistahl, Mojca Šraj, and Nejc Bezak

Interactive discussion

Status: closed

RC1:
'Comment on egusphere-2022-1279', Anonymous Referee #1, 28 Jan 2023

I have read the manuscript entitled " Validation of precipitation reanalysis products for rainfall-runoff modeling in Slovenia", which is a comparative analysis of two reanalysis precipitation datasets, ERA5-Land and COSMO-ERA6, to runoff simulation over different regions of Slovenia
the manuscript is relatively interesting and well written. I consider that it is suitable for publication in the EGUsphere journal, although, I would like to give some comments to the authors for the general improvement of the manuscript.
1- Please use more recent papers that are relevant to your topic. Below you can find some new studies:
https://link.springer.com/article/10.1007/s11269-022-03081-9
https://www.tandfonline.com/doi/full/10.1080/02626667.2019.1691217
2- Improve the quality of all figures. For example, in Fig1, 2, texts and numbers are blurred and need to improve.
3- What are your criteria for mesoscale, large-scale, and microscale categories?
4- Some of the studied catchments have an Area value lower than 100 km2, While ERA5-Land cells are higher than 100 km2. How did you deal with this issue? I think using this dataset for small catchments is not suitable. However, Still I don't understand your criteria for mesoscale. these catchments fall in the microscale category.
5- The values of numbers in Table 3 are not clear and I can't check the dataset combinations.
6- The authors used the GR4H model for runoff simulation. Based on the model structure (Fig 4), this model doesn't consider the snow component through the modeling. How do you explain this issue for snowy catchments of Slovenia?
7- For precipitation comparison, which approach is used in this study? Point-scale or Grid-scale approach? However, Based on fig1 and fig3 the density of your ground-gauge observations is not so good.
8- It is recommended to plot stream flow time series for a better understanding of model performance for simulating peak and low flows.
9- For the estimation of ET, which formulas and datasets are used for running the GR4H model? Please clarify this issue in the revised paper.
10- Why didn't the authors consider the uncertainty analysis for using the GR4H model?
Best

Citation: https://doi.org/10.5194/egusphere-2022-1279-RC1
- AC1: 'Reply on RC1', Hannes Müller-Thomy, 14 Mar 2023
  
  We thank RC1 for the constructive review. Please find our reply as supplement.
  
  Citation: https://doi.org/10.5194/egusphere-2022-1279-AC1
RC2:
'Comment on egusphere-2022-1279', Anonymous Referee #2, 03 Mar 2023

The authors present a study evaluating two reanalysis products against a gridded precipitation dataset that was constructed based on ground observational data from errors in input sources and errors propagated into hydrologic simulation across 16 catchments in Slovenia. The paper is within the scope of EGUsphere, however, major revisions must be made before considering publication. Below lists my major concerns regarding the scope of this study, methodology, and results presentation.

Major comments:

1. The scope or motivation of this study is not well articulated. Apparently, reanalysis datasets are not used to solve precipitation, but rather other meteorological variables. Precipitation variable is a by-product. Most of the reanalysis datasets do not directly assimilate precipitation into model chain (Although ERA5 and MERRA2 did assimilate precipitation). For hydrologic utilities - like floods the authors mentioned, a reanalysis dataset is not qualified and should not be a supplement for real observations unless no observations available. In summary, the authors should provide a strong argument prior to their analysis - why use reanalysis for hydrologic modelling since observational datasets exist.
2. In the introduction, I am not convinced by the facts why the authors chose to evaluate reanalysis data. In line 61-63, the authors stated that " mostly data gathered by means of remote sensing technology. ". First, I would like the authors to state why they are not considering other satellite derived products which have been proven to perform better than reanalysis. I am concerned the scope of this study. See my first comment.
3. The authors site a range of literature on evaluation of reanalysis in Lines 52- 93. However, I don’t see a summary of what the reviewed studies did not do, that is compensated by the current study and why reanalysis for Slovenian catchments. The authors should make this clear in the introduction.
4. Why is the ARSO-d data considered as the benchmark? There are other observational precipitation datasets available such as IMERG, MSWEP and GSMaP. The authors should provide more information to claim that ARSO-d data is more accurate than reanalysis products. More details of the on this data should be provided, such as data quality, and the data source, validation etc.
5. How is calibration applied to different forcing?
6. Include the location of the catchments in Table 1 (Lat and Lon), catchment characteristics relevant in the formation and dominant hydrological processes
In line 250 – 254, the authors state “The simulation period is split-sampled into…..”but do not justify the use of Split Sample method in model calibration. Refer to Arsenault et. al., 2018. The hazards of split-sample validation in hydrological model calibration.
7. Line 268 – 273 Please check the font and line spacing

Citation: https://doi.org/10.5194/egusphere-2022-1279-RC2
- AC1: 'Reply on RC1', Hannes Müller-Thomy, 14 Mar 2023
  
  We thank RC1 for the constructive review. Please find our reply as supplement.
  
  Citation: https://doi.org/10.5194/egusphere-2022-1279-AC1
- AC2: 'Reply on RC2', Hannes Müller-Thomy, 14 Mar 2023
  
  We thank RC2 for the constructive review. Please find our reply in the supplement.
  
  Citation: https://doi.org/10.5194/egusphere-2022-1279-AC2

Interactive discussion

Status: closed

RC1:
'Comment on egusphere-2022-1279', Anonymous Referee #1, 28 Jan 2023

I have read the manuscript entitled " Validation of precipitation reanalysis products for rainfall-runoff modeling in Slovenia", which is a comparative analysis of two reanalysis precipitation datasets, ERA5-Land and COSMO-ERA6, to runoff simulation over different regions of Slovenia
the manuscript is relatively interesting and well written. I consider that it is suitable for publication in the EGUsphere journal, although, I would like to give some comments to the authors for the general improvement of the manuscript.
1- Please use more recent papers that are relevant to your topic. Below you can find some new studies:
https://link.springer.com/article/10.1007/s11269-022-03081-9
https://www.tandfonline.com/doi/full/10.1080/02626667.2019.1691217
2- Improve the quality of all figures. For example, in Fig1, 2, texts and numbers are blurred and need to improve.
3- What are your criteria for mesoscale, large-scale, and microscale categories?
4- Some of the studied catchments have an Area value lower than 100 km2, While ERA5-Land cells are higher than 100 km2. How did you deal with this issue? I think using this dataset for small catchments is not suitable. However, Still I don't understand your criteria for mesoscale. these catchments fall in the microscale category.
5- The values of numbers in Table 3 are not clear and I can't check the dataset combinations.
6- The authors used the GR4H model for runoff simulation. Based on the model structure (Fig 4), this model doesn't consider the snow component through the modeling. How do you explain this issue for snowy catchments of Slovenia?
7- For precipitation comparison, which approach is used in this study? Point-scale or Grid-scale approach? However, Based on fig1 and fig3 the density of your ground-gauge observations is not so good.
8- It is recommended to plot stream flow time series for a better understanding of model performance for simulating peak and low flows.
9- For the estimation of ET, which formulas and datasets are used for running the GR4H model? Please clarify this issue in the revised paper.
10- Why didn't the authors consider the uncertainty analysis for using the GR4H model?
Best

Citation: https://doi.org/10.5194/egusphere-2022-1279-RC1
- AC1: 'Reply on RC1', Hannes Müller-Thomy, 14 Mar 2023
  
  We thank RC1 for the constructive review. Please find our reply as supplement.
  
  Citation: https://doi.org/10.5194/egusphere-2022-1279-AC1
RC2:
'Comment on egusphere-2022-1279', Anonymous Referee #2, 03 Mar 2023

The authors present a study evaluating two reanalysis products against a gridded precipitation dataset that was constructed based on ground observational data from errors in input sources and errors propagated into hydrologic simulation across 16 catchments in Slovenia. The paper is within the scope of EGUsphere, however, major revisions must be made before considering publication. Below lists my major concerns regarding the scope of this study, methodology, and results presentation.

Major comments:

1. The scope or motivation of this study is not well articulated. Apparently, reanalysis datasets are not used to solve precipitation, but rather other meteorological variables. Precipitation variable is a by-product. Most of the reanalysis datasets do not directly assimilate precipitation into model chain (Although ERA5 and MERRA2 did assimilate precipitation). For hydrologic utilities - like floods the authors mentioned, a reanalysis dataset is not qualified and should not be a supplement for real observations unless no observations available. In summary, the authors should provide a strong argument prior to their analysis - why use reanalysis for hydrologic modelling since observational datasets exist.
2. In the introduction, I am not convinced by the facts why the authors chose to evaluate reanalysis data. In line 61-63, the authors stated that " mostly data gathered by means of remote sensing technology. ". First, I would like the authors to state why they are not considering other satellite derived products which have been proven to perform better than reanalysis. I am concerned the scope of this study. See my first comment.
3. The authors site a range of literature on evaluation of reanalysis in Lines 52- 93. However, I don’t see a summary of what the reviewed studies did not do, that is compensated by the current study and why reanalysis for Slovenian catchments. The authors should make this clear in the introduction.
4. Why is the ARSO-d data considered as the benchmark? There are other observational precipitation datasets available such as IMERG, MSWEP and GSMaP. The authors should provide more information to claim that ARSO-d data is more accurate than reanalysis products. More details of the on this data should be provided, such as data quality, and the data source, validation etc.
5. How is calibration applied to different forcing?
6. Include the location of the catchments in Table 1 (Lat and Lon), catchment characteristics relevant in the formation and dominant hydrological processes
In line 250 – 254, the authors state “The simulation period is split-sampled into…..”but do not justify the use of Split Sample method in model calibration. Refer to Arsenault et. al., 2018. The hazards of split-sample validation in hydrological model calibration.
7. Line 268 – 273 Please check the font and line spacing

Citation: https://doi.org/10.5194/egusphere-2022-1279-RC2
- AC1: 'Reply on RC1', Hannes Müller-Thomy, 14 Mar 2023
  
  We thank RC1 for the constructive review. Please find our reply as supplement.
  
  Citation: https://doi.org/10.5194/egusphere-2022-1279-AC1
- AC2: 'Reply on RC2', Hannes Müller-Thomy, 14 Mar 2023
  
  We thank RC2 for the constructive review. Please find our reply in the supplement.
  
  Citation: https://doi.org/10.5194/egusphere-2022-1279-AC2

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

ED: Reconsider after major revisions (further review by editor and referees) (15 Mar 2023) by Daniel Viviroli

AR by Hannes Müller-Thomy on behalf of the Authors (24 Mar 2023) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (24 Mar 2023) by Daniel Viviroli

RR by Asghar Azizian (29 Mar 2023)

RR by Anonymous Referee #3 (09 May 2023)

Suggestions for revision or reasons for rejection

I have read the revised manuscript as well as the authors’ responses to the two referee reports. Overall, the manuscript is suitable for publication in Hydrology and Earth System Sciences, and the major concerns mentioned by the two other referees seem to be cleared out.

However, the following smaller issues and questions should be addressed before publication:

- In your response to point 4 by referee #2 (“Why is the ARSO-d data considered as the benchmark?”), you mention that you also wanted to implement a discussion in the revised manuscript to more clearly address questions in this direction. I may have missed it, but can you indicate where that has been added, or complete if it hasn’t been done yet?

- Conclusions: Can you make a statement on the extent to which your findings can be applied to other regions, i.e., to what extent can they be generalized, and where are limits to that?

- L53: Amend incomplete sentence "… as shown for e.g. by e.g. Gebremichael …”

- L61: Do you refer to a commercial context for a specific reason? Otherwise I would rephrase more generally towards applications.

- L229: Clarify or replace the term “catchment-intern analyses", it is used only here.

- L228: Clarify the “11 years (Jan-Aug) and 10 years (Sep-Dec)” in parentheses: Why these months?

- L273: You write that "both calibration and validation periods represent similar climate, and hence catchment conditions. " I would not put it that way, as there are other catchment conditions beyond climate.

- L275: Please be more specific in this sentence, it sounds rather generic: "In addition, different approaches are available as calibration and model evaluation strategies, and some studies support the split-sampling approach (e.g., Perrin et al., 2003)."

-L296: I would mention here in addition that a KGE value of more than -0.41 (i.e., 1-sqrt(2)) indicates model efficiency better than the mean flow benchmark according to Knoben et al. (2019).

- L426: The figure legend seems to be mixed with the main text.

- L440ff: Clarify what you mean by "ad-hoc calibration".

- L454: "No significant trends are observable in relation to latitude." Do you use “significant” in a statistical sense? If yes indicate the test done and the p-value used, otherwise change the wording.

- L476: Complete “distributed” at the end of sentence "PRP is of little added, value compared to a spatially distr."

- L500: Can you give some indications why there are overestimations > 1000 m a.s.l., and why deviations are smaller for catchments below this elevation?

- L503: "the *following* conclusions"?

- L510: "show*s*”

- L521: You conclude that bias-correction should be implemented, but on L112ff you mention that an advantage of the present research is the examination of raw data. Can you clarify?

- Equation 1: Should it be "RC[PRP, n]" and “RC[Obs, n]” instead of “RC[PRP]” and “RC[Obs]” (both in numerator and denominator)?

- Table 1: Use "discharge regime" instead of "water regime" for consistency?

- Figure 2: 1) The rivers do not seem to be in alphabetical order, but rather the maxima seem to increase from left to right. Please clarify and amend. 2) I recommend mentioning in the legend that the values are displayed on a logarithmically scaled y-axis. 3) Using different symbols for min/mean/max would improve readability of the figure.

- Figure 3: Is there a reason for the larger paragraph spacing between Mislinja, Dravinja and Radovna?

- Figure 6: The figure legend isn’t clear to me: "Deviations of areal rainfall extreme values (5- and 50-year return periods) between ERA5-Land and COSMO-REA6 over all 16 catchments." ? But isn't ARSO-d the reference?

- Figure 9: It doesn’t seem correct to show negative values in white. KGE has a value range from inf to 1, so values below 0 do not fall outside of that range. Values <0 should rather have an own class in the color bar, and use a color that shows it’s a continuum, e.g., dark red.

- Figure 11: I recommend adding a 1:1 line and making the figure square for better interpretation.

- Figure 12: This is for the recalibrated model, isn't it? Mention this in legend for clarity.

Reference:
W. J. M. Knoben, J. E. Freer, and R. A. Woods, 2019: Technical note: Inherent benchmark or not? Comparing Nash-Sutcliffe and Kling-Gupta efficiency scores. Hydrol. Earth Syst. Sci. 23, 4323–4331. doi:10.5194/hess-23-4323-2019.

Hide

ED: Publish subject to minor revisions (review by editor) (10 May 2023) by Daniel Viviroli

AR by Hannes Müller-Thomy on behalf of the Authors (19 May 2023) Author's response Author's tracked changes Manuscript

ED: Publish as is (24 May 2023) by Daniel Viviroli

AR by Hannes Müller-Thomy on behalf of the Authors (30 May 2023)

Journal article(s) based on this preprint

12 Jul 2023

Validation of precipitation reanalysis products for rainfall-runoff modelling in Slovenia

Marcos Julien Alexopoulos, Hannes Müller-Thomy, Patrick Nistahl, Mojca Šraj, and Nejc Bezak

Hydrol. Earth Syst. Sci., 27, 2559–2578, https://doi.org/10.5194/hess-27-2559-2023,https://doi.org/10.5194/hess-27-2559-2023, 2023

Short summary

Marcos Julien Alexopoulos, Hannes Müller-Thomy, Patrick Nistahl, Mojca Šraj, and Nejc Bezak

Viewed

Total article views: 507 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
358	130	19	507	8	7

HTML: 358
PDF: 130
XML: 19
Total: 507
BibTeX: 8
EndNote: 7

Views and downloads (calculated since 06 Dec 2022)

Month	HTML	PDF	XML	Total
Dec 2022	105	38	7	150
Jan 2023	77	20	2	99
Feb 2023	37	15	1	53
Mar 2023	64	24	7	95
Apr 2023	23	13	0	36
May 2023	22	6	0	28
Jun 2023	22	12	2	36
Jul 2023	8	2	0	10
Aug 2023	0
Sep 2023	0
Oct 2023	0
Nov 2023	0
Dec 2023	0
Jan 2024	0
Feb 2024	0
Mar 2024	0
Apr 2024	0
May 2024	0
Jun 2024	0
Jul 2024	0
Aug 2024	0
Sep 2024	0

Cumulative views and downloads (calculated since 06 Dec 2022)

Month	HTML	PDF	XML	Total
Dec 2022	105	38	7	150
Jan 2023	77	20	2	99
Feb 2023	37	15	1	53
Mar 2023	64	24	7	95
Apr 2023	23	13	0	36
May 2023	22	6	0	28
Jun 2023	22	12	2	36
Jul 2023	8	2	0	10
Aug 2023	0
Sep 2023	0
Oct 2023	0
Nov 2023	0
Dec 2023	0
Jan 2024	0
Feb 2024	0
Mar 2024	0
Apr 2024	0
May 2024	0
Jun 2024	0
Jul 2024	0
Aug 2024	0
Sep 2024	0

Viewed (geographical distribution)

Total article views: 505 (including HTML, PDF, and XML) Thereof 505 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 04 Sep 2024

Download

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Preprint (1693 KB)
Metadata XML

Short summary

For the rainfall-runoff simulation from a certain area hydrological models are used, which require precipitation and temperature data as input. Since these are often not available as observations, we tested simulation results from atmospheric models. Two products were tested (ERA5-Land & COSMO-REA6) for Slovenian catchments, both lead to good simulations results. Their usage enables the of rainfall-runoff simulation in unobserved catchments as a requisite for e.g. flood protection measures.


Total:	0
HTML:	0
PDF:	0
XML:	0