Dynamical Downscaling of ERA5 Data on the North-Western Mediterranean Sea: From Atmosphere to High-Resolution Coastal Wave Climate

Round 1

Reviewer 1 Report

This paper is a very sound study of the wind and waves in the Mediterranean Sea. A very thorough assessment of the skill of both the atmospheric model and the wave model is reported using an extensive data set. The performance is very good. I look forward to the next paper about the extreme values that I expect will be presented soon. This work will have important regional significance.

There seems to be a few errors in English in the abstract, but otherwise it is well prepared and ready for publication.

Author Response

Response to Reviewer 1 Comments

This paper is a very sound study of the wind and waves in the Mediterranean Sea. A very thorough assessment of the skill of both the atmospheric model and the wave model is reported using an extensive data set. The performance is very good. I look forward to the next paper about the extreme values that I expect will be presented soon. This work will have important regional significance.

There seems to be a few errors in English in the abstract, but otherwise it is well prepared and ready for publication.

Response: We thank the Reviewer for his interest in our work and in the one in progress about the extreme values. We made some changes in the abstract, especially in regard to the language and style (including fine/minor spell check).

The new abstract is:

A 29-year wind/wave hindcast is produced over the Mediterranean Sea for the period 1990 - 2018. The dataset is obtained by downscaling the ERA5 global atmospheric reanalyses, which provide the initial and boundary conditions for a numerical chain based on limited-area weather and wave models: the BOLAM, MOLOCH and WaveWatch III (WW3) models. In the WW3 computational domain an unstructured mesh is used. The variable resolution reaches up to 500 m along the coasts of the Ligurian and Tyrrhenian seas (Italy), the main object of the study. The wind/wave hindcast is validated using observations from coastal weather stations and buoys. The wind validation provides velocity correlations between 0.45 and 0.76, while significant wave heights correlations are much higher, between 0.89 and 0.96. The results are also compared to the original low-resolution ERA5 dataset, based on assimilated models. The comparison shows that the downscaling improves the hindcast reliability, particularly in the coastal regions, and especially with regards to wind and wave directions.

Author Response File: Author Response.pdf

Reviewer 2 Report

Review of the manuscript entitled “Dynamical downscaling of ERA5 data on the Mediterranean Sea: from atmosphere to high-resolution coastal wave climate” submitted to Journal of Marine Science and Engineering by V. Vannucchi et al.

This manuscript presents a dynamical downscaling of ERA5 data for the Mediterranean Sea with focus on the Ligurian and Tyrrhenian Seas. A validation with weather stations and buoys that the downscalled dataset generally outperforms ERA5 in the study area, particularly for wind and wave directions at coastal locations. I think the dataset, and therefore the manuscript, is of relevance for the marine science and engineering community. However, a number of aspects should be addressed before publication:

• Lines 153-156: The authors comment on how they achieve a smooth transition at the boundaries of the MOLOCH model’s domain. What about the temporal transition? How do they ensure that there is a smooth temporal transition between 24UTC at day-0 and 00UTC at day+1?
• Lines 192-199: Please be more specific about the range of output resolution, as it is not very clear in Figure 4 for the coastal locations. Also, I suggest adding NW to the title. The resulting (output) product provides high-resolution coastal wave climate only for the North-Western Mediterranean Sea.
• Line 194: what is the “prevailing wind direction”? please specify.
• Figure 6, 7 and 8: I suggest using the normalized Taylor diagram to compare the performance at different locations.
• Results/validation: In some instances, the ERA5 model performs slightly better for Hs and it is not properly commented in the text.
• Lines 494-496 repeat what was said just a few lines above (481-484).
• Lines 526-527: I would not describe the study area as a “large area of the North-Western Mediterranean Sea”. I think it is more accurate to say “ the Ligurian and Tyrrhenian Seas (North Western Mediterranean Sea).”
• Lines 590-595: “despite some limitations and shortcomings which are unavoidable when dealing with complex modelling chains” This is too general and ambiguous. Please be more precise. Also, I would not say “extremely robust” dataset.

Some minor formal aspects:

• Figure 11d: why are the axis changed? It would be better to keep consistency, ie same axis as Figure 11a-d.
• Line 463: I would not use WW3 to refer to developed hindcast. WW3, ie WAVEWATCH III, is not exclusive to this hindcast.

In addition, although the text reads reasonably well, I think it should be improved to meet the journal’s standards of quality. I recommend the authors to carefully edit the text (wherever necessary) to try be more concise, and correct existing English mistakes. Moreover, the manuscript is overly long and I therefore encourage the authors to shorten it. First, the entire manuscript will benefit from using shorter, and more concise, statements, and avoiding repetition. Second, in my opinion the calibration section could be substantially shorter and many of the corresponding tables/figures should be moved to SM. However, to improve readability it would be helpful to include a summary table with the settings that were used in the calibration process.

Some examples regarding typos/grammar/spelling mistakes included below (not meant to be exhaustive):

• Line 27: I would remove “nowadays”.
• Lines 40-41: “A reliable…” I suggest to rephrase this sentence which does not read well.
• Line 94-95: “realizing” is wrong here. A possibility could be “performing”
• Line 220: change “wind mean speeds” to “mean wind speeds”
• Figure 6 caption: “The plus symbols … respectively” sentence is repeated.
• Line 587: I suggest not to use “anyway”

Author Response

Response to Reviewer 2 Comments

We would like to thank the Reviewer for careful and thorough reading of this manuscript and for the constructive suggestions, which help to improve the quality of this manuscript.

Point 1: Lines 153-156: The authors comment on how they achieve a smooth transition at the boundaries of the MOLOCH model’s domain. What about the temporal transition? How do they ensure that there is a smooth temporal transition between 24UTC at day-0 and 00UTC at day+1?

Response 1: The 24UTC at day-0 and the 00UTC at day+1 are averaged.

At line 160 we add this sentence: “Data at the 24UTC at day-0 and the 00UTC at day+1 were averaged to obtain a smooth temporal resolution” 

Point 2: Lines 192-199: Please be more specific about the range of output resolution, as it is not very clear in Figure 4 for the coastal locations. Also, I suggest adding NW to the title. The resulting (output) product provides high-resolution coastal wave climate only for the North-Western Mediterranean Sea.

Response 2: Thank you for your helpful suggestions.

We have modified the title in “Dynamical downscaling of ERA5 data on the North-Western Mediterranean Sea: from atmosphere to high-resolution coastal wave climate”.

We have added more information about the range of the output resolution. At line 199 we added this part: “The output of the wave model was recorded hourly at all grid points for the integrated quantities: significant wave height (Hs), mean wave period (Tm), peak wave period (Tp), mean wave direction (Dirm) and peak wave direction (Dirp). The output was also recorded hourly in 2048 points for both mean and spectral wave parameters (Figure 4). Part of these points matched with the buoys location, whereas the others were distributed in the Mediterranean Sea with a resolution of about 111 km (∼1°). In the North-Western part the resolution of point distribution was increased to about 55 km (∼0.5°) and additional points were extracted along the coastline at a distance of 1 km along the Tuscany coast, about 4 km along the Sardinia and Corsica Islands, variable between 3.5 km and 9.0 km along the Liguria coastline, and between 9 km and 17 km along the French coastline up to Marseille. The mean wave parameters at the points corresponding to the buoys location were used for model calibration and validation.”

Point 3: Line 194: what is the “prevailing wind direction”? please specify.

Response 3: Thank you. It’s a mistake, we changed in mean wind direction

Point 4: Figure 6, 7 and 8: I suggest using the normalized Taylor diagram to compare the performance at different locations.

Response 4: Thanks for the suggestion. In order to reduce the number of overall figures we had chosen to represent all stations together. We had tried to normalize those Taylor diagrams, but they were unclear, so we opted for a not-normalized diagram without the cRMSE visualisation.

Point 5: Results/validation: In some instances, the ERA5 model performs slightly better for Hs and it is not properly commented in the text.

Response 5: We added a sentence in the wave validation section at the line 494, that highlights the ERA5 dataset slight better results in the case of La Revellata Taylor diagram.

“Only La Revellata buoy Taylor diagram (see Figure S6m in Supplementary Materials) shows slightly better results for the ERA5 dataset: the Hs correlation is around 2% higher (Table 10) and the Hs cRMSE is around 20% lower (Table 10).”

Point 6: Lines 494-496 repeat what was said just a few lines above (481-484).

Response 6: Thanks for your observation. We removed the lines 494 – 496 and modified the lines 481-484 with this:

“We underline the fact that the comparison with ERA5 data was only possible for the five buoys (no. 2, 3, 5, 11, 14 in Table 3) that were able to record the mean wave direction; the other buoys measured only the peak wave direction, excluding the Cap Corse buoy (no. 10), which did not record any wave direction.”

Point 7: Lines 526-527: I would not describe the study area as a “large area of the North-Western Mediterranean Sea”. I think it is more accurate to say “ the Ligurian and Tyrrhenian Seas (North Western Mediterranean Sea).”

Response 7: We appreciate your suggestion and correct the phrase. Now the sentence is:

“In this work, we presented the results of a dynamical downscaling of ERA5 reanalysis dataset [19], focused on the wind and wave climate of the Ligurian and Tyrrhenian seas (North-Western Mediterranean Sea).”

Point 8: Lines 590-595: “despite some limitations and shortcomings which are unavoidable when dealing with complex modelling chains” This is too general and ambiguous. Please be more precise. Also, I would not say “extremely robust” dataset.

Response 8: We changed this part. Now the final sentences are:

The work we presented demonstrated a good reliability of the downscaling process aimed at the improvement of high quality reanalysis products such as ERA5 data. The hindcast will be further expanded to reach 40 years, in order to provide a dataset for different scopes like climatological studies and extreme value analysis.

Some minor formal aspects:

Point 9: Figure 11d: why are the axis changed? It would be better to keep consistency, ie same axis as Figure 11a-d.

Response 9: Yes, you are absolutely right, we corrected the Figure in order to keep consistency. Now the Figure is:

Point 10: Line 463: I would not use WW3 to refer to developed hindcast. WW3, ie WAVEWATCH III, is not exclusive to this hindcast.

Response 10: Thanks. In the wave validation and in the discussions/conclusion the WW3 term was changed with “simulated” or “wave hindcast”.

Point 11: In addition, although the text reads reasonably well, I think it should be improved to meet the journal’s standards of quality. I recommend the authors to carefully edit the text (wherever necessary) to try be more concise, and correct existing English mistakes. Moreover, the manuscript is overly long and I therefore encourage the authors to shorten it. First, the entire manuscript will benefit from using shorter, and more concise, statements, and avoiding repetition. Second, in my opinion the calibration section could be substantially shorter and many of the corresponding tables/figures should be moved to SM. However, to improve readability it would be helpful to include a summary table with the settings that were used in the calibration process.

Response 11: Thank you so much. We made some changes to make our manuscript hopefully clearer and shorter. We also corrected some English mistakes.

The calibration section was shortened: we moved on the Supplementary Materials the visual comparison of the time series of Hs. The other tables and figures are useful to keep in the main text to make the reader understand all the steps that led us to the final settings choice.

We completely agree on a summary table with the settings that were used in the calibration process. At the line 361 we added the following part and the summary table.

“Table 8 reports the settings that evaluated in the calibration process and those employed for the hindcast: 60 s time step, explicit N-scheme, and ST4 source term parameterisation. During the last phase of the calibration process, a visual comparison of the time series of Hs was also performed (see section 1 in Supplementary Materials).”

                                            Table 8. Settings evaluated in the calibration process and final selection.

Point 12: Some examples regarding typos/grammar/spelling mistakes included below (not meant to be exhaustive):

Line 27: I would remove “nowadays”.

Lines 40-41: “A reliable…” I suggest to rephrase this sentence which does not read well.

Line 94-95: “realizing” is wrong here. A possibility could be “performing”

Line 220: change “wind mean speeds” to “mean wind speeds”

Figure 6 caption: “The plus symbols … respectively” sentence is repeated.

Line 587: I suggest not to use “anyway”

Response 12: Thanks to much. We corrected these mistakes and made general changes especially to the language and style (including fine/minor spell check), following also the suggestions of the others reviewers.

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors present a detailed wave study of the Mediterranean using an unstructured Wavewatch model forced by downscaled ERA5 winds. This is quite an interesting study, as use of Wavewatch in coastal regions is still novel. The paper is well written, the techniques are straight-forward and the descriptions are clear. I have no concerns with the paper and recommend publication in its present form. 

Author Response

Response to Reviewer 3 Comments

The authors present a detailed wave study of the Mediterranean using an unstructured Wavewatch model forced by downscaled ERA5 winds. This is quite an interesting study, as use of Wavewatch in coastal regions is still novel. The paper is well written, the techniques are straight-forward and the descriptions are clear. I have no concerns with the paper and recommend publication in its present form.

Response: We wish to thank the Reviewer for having appreciated our work.

We made some changes especially to the language and style (including fine/minor spell check), following also the suggestions of the others reviewers.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The writing of the manuscript has been improved and the comments have been addressed. There are however still a few typos and English language errors to correct. For example: 

Line 43: "The main contribution of a wave hindcasts is"

Line 203: "In the North-Western part the resolution of point..."

After another revision of the language, I recommend the publication of the article. I also suggest this additional change:

In order to better illustrate the settings that were tested in the calibration process, I suggest the authors to add a third column to the summary table (Table 8) to specify the parameters that were kept constant at each step of the calibration process, ie (1) N scheme, ST4; (2) 60 s, ST4; (3) 60 s, N scheme

Author Response

Point 1: The writing of the manuscript has been improved and the comments have been addressed. There are however still a few typos and English language errors to correct. For example:

Line 43: "The main contribution of a wave hindcasts is"

Line 203: "In the North-Western part the resolution of point..."

After another revision of the language, I recommend the publication of the article.

Response 1: We wish to thank the Reviewer for having appreciated our work. We corrected these mistakes and made general changes especially to the language and style (including fine/minor spell check).

Point 2: I also suggest this additional change:

In order to better illustrate the settings that were tested in the calibration process, I suggest the authors to add a third column to the summary table (Table 8) to specify the parameters that were kept constant at each step of the calibration process, ie (1) N scheme, ST4; (2) 60 s, ST4; (3) 60 s, N scheme

Response 2: Thank you for your helpful suggestion.

We modified the Table 8 and the previous phrase.

Table 8 reports the settings evaluated in the calibration process, those kept constant at each phase and those employed for the hindcast: 60 s time step, explicit N-scheme, and ST4 source term parameterisation.

Table 8. Settings evaluated and kept constant in the calibration process and final selection.

Author Response File: Author Response.pdf