Generation and Analysis of Gridded Visibility Data in the Arctic

Round  1

Reviewer 1 Report

The paper is very well written and presented. Inferring visibility from weather data is very useful since visibility is not measured in all weather stations.
The ANN approach is interesting and relevant for the type of model presented here. Estimation error is acceptable.
Figures 7 to 9 show averages of estimated visibility at several time scales. However, it is interesting to add some examples of visibility maps produced for single measurements, without averaging. This will allow to see the spatial variability of this parameter.

Reviewer 2 Report

General Comments:

    The manuscript is about training a neural network model based on past data to project recent visibility data (2016) in a data-sparse region like the arctic. Such a model could be valuable in the arctic for the future. The modeling is done well and the references are appropriate. The figures are neat and readable. The work is original and as such should be published. There are some issues that should be addressed regarding the presentation and explanation that will strengthen the paper. 

Major comments:

The sentences in the introduction (23-26): These need to be re-written. These sentences overstate the case a bit. This reviewer has reviewed the IPCC AR 5 report extensively. In these projections, the arctic will be nearly ice free in the later part of the 21st century in the more extreme CMIP 3 and 5 model scenarios (RCP 8.5). See page 1087 in AR 5 Ch 12 section 12.4.6.1.

Table 4: There's little discussion of Table 4 and the results are mainly good, but what about months 6-8 where three of the values in test 2 are quite high? Is there a reason? This should be addressed/explained. 

Minor:

Line 97: Should the Rumelhart and McClelland (1986) just be abbreviated as [8]?

Table 2: the table is very difficult to read. It's difficult to determine what row 1 1981-2015 represents. Please make teh text smaller so things appear on one line.

3. Table 3, Fig. 4, and Fig 5 and discussion. The relative error is below 0.22 and 0.20 respectively. is there an acceptable value for relative error using visibility data? Also, it is encouraging that more than 3/4 of the data show relative error less than 0.15 or some value like that. This strengthens the presentation.  

4. Line 297-298: Terminology "grade" is suddenly used. Is this "Visibility level" from Table 1? Make it clear that these are the same.

5. References:  the journal name can be shown as: for example in reference 1: "J. Appl. Meteorol. Clim."  (in italics of course). 

Specific Comments:

Line 7 suggest changing "world" to "arctic".

Line 29: suggest "military risks but also are associated"

Table 3

Reviewer 3 Report

General:

Overall it is an important topic and needs some attention from the scientific community. Before goes to publication, needs some improvements are summarized below. Most important issue is that no Discussions are given, this is needed. Also, conclusions need to be improved, see below. Introduction also needs some improvements, stating that some work are done for arctic vis and findings……

Abstract; last few sentences are not scientifically written, needs some attention.

LN27; Gultepe et al (2019, Pure and App. Geo.) stated that Vis is important parameter over Arctic regions and in general it affect aviation industry significantly. Gultepe et al also stated that broken ice surfaces in Artcic will results more fog events, reducing Vis (Gultepe et al 2003, Atmosphere and Ocean, 41, 15-34)

Page 3; 94-99; needs some references here.

INTRO: provide some Vis analysis based on models (refs…..), see Gultepe et al 2018: Marine fog: Observations and models.

LN151; provide a ref of Gultepe et al (2018) Marine fog.

LN157-160; Gultepe et al 2019 stated that Vis is most important parameter for aviation applications and RHw is very sensitive for Vis calculations, and needs better than 10% accuracy.

You need a DISCUSSION section for quality of work and issues.

Gultepe et al 2006 (J App. Met); 2014(Atmos res) suggested that Vis and microphysical parameters are nonlinearly related to each other and RHw and Nd, LWC are most important parameters for fog Vis calculation.

LN 270-277; provide results as …. North of 80 Lat or East of 180 Lon

Again do a separate section on Discussion.

Conclusions:

LN 306-307; Vis, as one…….the Arctic, is receiving  and increasing ……(Gultepe et al 2019, PAAG Aviation meteorology).

LN321; what are these features???

LN324-325;  grade 8-10 to level 1-10, and most severe changes happens…….

Please provide your findings properly in an itemized way.

Comments for author File: Comments.pdf

Author Response

Reviewer 1 Report

Reviewer: 1

Comments and Suggestions:

The paper is very well written and presented. Inferring visibility from weather data is very useful since visibility is not measured in all weather stations. The ANN approach is interesting and relevant for the type of model presented here. Estimation error is acceptable. Figures 7 to 9 show averages of estimated visibility at several time scales. However, it is interesting to add some examples of visibility maps produced for single measurements, without averaging. This will allow to see the spatial variability of this parameter.               

Response:

Firstly very thanks for the recognition and suggestions of the reviewer. I have added 2 maps of visibility in the arctic at 00:00 on August 1, 2018 and 00:00 on September 1, 2018 (Fig. 11) without averaging and analyzed the spatial variability of visibility at the 2 time points. We find that visibility could vary widely across the Arctic at the same time point. Visibility distribution at 00:00 on August 1, 2018 in the Arctic is more complex than that at 00:00 on September 1, 2018. At the same time, nearshore visibility is better than offshore visibility generally and visibility in the Barents Sea and Greenland Sea is relatively lower compared with that in other sea areas. Relevant content has been added on line 361-371.

Author Response

Reviewer 2 Report

Comment 1:

The sentences in the introduction (23-26): These need to be re-written. These sentences overstate the case a bit. This reviewer has reviewed the IPCC AR 5 report extensively. In these projections, the arctic will be nearly ice free in the later part of the 21st century in the more extreme CMIP 3 and 5 model scenarios (RCP 8.5). See page 1087 in AR 5 Ch 12 section 12.4.6.1.          

Response:

Very thanks for the suggestions of the reviewer and I noticed that the original sentences of our paper did overstate the case. I have read relevant content in AR5 of IPCC and rewritten the sentences. Relevant content has been shown on line 22-26.    

Comment 2: 

There's little discussion of Table 4 and the results are mainly good, but what about months 6-8 where three of the values in test 2 are quite high? Is there a reason? This should be addressed/explained.         

Response:

Very thanks for the suggestions of the reviewer and I have noticed the issue. After careful analysis of the model and data, we find that relative errors in months 6-9 are quite high when training the model using data of months 2 and 3. Also, relative errors in months 7-9 are always high regardless of which month's data is used as training data (Table 3, Table 4). We think two reasons may lead to the result. One reason is the visibility data in months 2-3 are relatively single which cause the inadequate training of the parameters of the neural network. Another reason is the changes of visibility in months 6-9 in the Arctic are more complex compared with that in other months, especially in July and August (Fig. 8). Both of the two reasons lead to the trained parameters of the Neural Network could not infer visibility accurately. Relevant content has been added on line 247-254.

Comment 3: 

Line 97: Should the Rumelhart and McClelland (1986) just be abbreviated as [8]?      

Response:

Very thanks for the suggestions of the reviewer and I have added relevant content on line 109.

Comment 4: 

Table 2: the table is very difficult to read. It's difficult to determine what row 1 1981-2015 represents. Please make teh text smaller so things appear on one line.

Response:

Very thanks for the suggestions of the reviewer. Since that data from 1981 to 2015 is used to train the model and data in 2016 is used to test the accuracy of the inferred visibility based on the trained model, table 2 presents the amount of data for these two periods separately. I have added relevant content on line 191-193.

Comment 5: 

Table 3, Fig. 4, and Fig 5 and discussion. The relative error is below 0.22 and 0.20 respectively. Is there an acceptable value for relative error using visibility data? Also, it is encouraging that more than 3/4 of the data show relative error less than 0.15 or some value like that. This strengthens the presentation.

Response:

Very thanks for the suggestions of the reviewer. I could not find the acceptable value for relative error using visibility data but I learned that gridded visibility data used in study is mostly from numerical weather forecast (NWP) and the uncertainty in visibility calculations in NWP can reach 29% to 30% because of the uncertainty in visibility parameterization, so the visibility data generated in this paper is relatively more accurate. Also, I have added the content that more than 3/4 of the data show relative error less than 0.15 or some value like that to strengthen the presentation. Relevant content has been added on line 240-246. 

Comment 6: 

Line 297-298: Terminology "grade" is suddenly used. Is this "Visibility level" from Table 1? Make it clear that these are the same.   

Response:

Very thanks for the suggestions of the reviewer. Terminology "grade" is the same as terminology "level" from Table 1. I have changed all “grade” in this article to “level”.

Comment 7: 

References: the journal name can be shown as: for example in reference 1: "J. Appl. Meteorol. Clim." (In italics of course).

Response:

Very thanks for the suggestions of the reviewer. I have revised relevant content according to the reviewer’s suggestions.

Comment 8: 

Line 7 suggest changing "world" to "arctic".

Response:

Very thanks for the suggestions of the reviewer and I have revised relevant content according to the reviewer’s suggestions on line 6.

Comment 9: 

Line 29: suggest "military risks but also are associated".

Response:

Very thanks for the suggestions of the reviewer and I have revised relevant content according to the reviewer’s suggestions on line 28-29.

Author Response

Reviewer 3 Report

 Point 1:

Abstract; last few sentences are not scientifically written, needs some attention. 

Response 1:

Very thanks for the suggestions of the reviewer and I noticed the problem existing in the sentences. I have revised the writing and relevant content is shown on line 15-18.    

Point 2: 

LN27; Gultepe et al (2019, Pure and App. Geo.) stated that Vis is important parameter over Arctic regions and in general it affect aviation industry significantly. Gultepe et al also stated that broken ice surfaces in Artcic will results more fog events, reducing Vis (Gultepe et al 2003, Atmosphere and Ocean,

41, 15-34)

Response 2:

Very thanks for the suggestions of the reviewer. I have cited the above two papers to demonstrate the importance of sea ice and visibility influencing the navigation risk of crossing Aps. Relevant content is shown on line 35.

Point 3: 

Page 3; 94-99; needs some references here.

Response 3:

Very thanks for the suggestions of the reviewer. I have cited three papers to demonstrate the importance of obtaining gridded data of visibility in the Arctic. Relevant content is shown on line 109.

Point 4: 

INTRO: provide some Vis analysis based on models (refs…..), see Gultepe et al 2018: Marine fog: Observations and models

Response 4:

Very thanks for the suggestions of the reviewer. I have read the paper you mentioned carefully and provided some visibility analysis based on models. At the same time, I add some content about the scheme and shortcomings of visibility microphysical parameterization in the paper to make readers know more about calculating visibility based on NWP. Relevant content is shown on line 50-67.

Point 5: 

LN151; provide a ref of Gultepe et al (2018) Marine fog

Response 5:

Very thanks for the suggestions of the reviewer and I have cited the paper to make our paper more rigorous. Relevant content is shown on line 154.

Point 6: 

LN157-160; Gultepe et al 2019 stated that Vis is most important parameter for aviation applications and RHw is very sensitive for Vis calculations, and needs better than 10% accuracy.

Response 6:

Very thanks for the suggestions of the reviewer and I have added these content in the paper to enrich our paper. Relevant content is shown on line 169-173.

Point 7: 

You need a DISCUSSION section for quality of work and issues.

Response 7:

Very thanks for the suggestions of the reviewer. I have added a DISCUSSION section for quality of work and issues in the paper. Relevant content is shown on line 399-408.

Point 8: 

Gultepe et al 2006 (J App. Met); 2014(Atmos res) suggested that Vis and microphysical parameters are nonlinearly related to each other and RHw and Nd, LWC are most important parameters for fog Vis calculation.

Response 8:

Very thanks for the suggestions of the reviewer. I have cited the paper you mentioned to demonstrate the nonlinear relationship between visibility and its influence factors. At the same time, I also added some content about introducing the influencing factors on visibility. Relevant content is shown on line 93-95 and line 171-173.

Point 9: 

LN 270-277; provide results as …. North of 80 Lat or East of 180 Lon

Response 9:

Very thanks for the suggestions of the reviewer. I have revised the sentence according to your suggestions to make the description more clear. Relevant content is shown on line 17, 331, 355 and 396.  

Point 10: 

LN 306-307; Vis, as one…….the Arctic, is receiving and increasing ……(Gultepe et al 2019, PAAG Aviation meteorology).

Response 10:

Very thanks for the suggestions of the reviewer. I have revised the sentence according to your suggestions. Relevant content is shown on line 375-376.

Point 11: 

LN321; what are these features???

Response 11:

Very thanks for the suggestions of the reviewer. By analyzing the result of table 3, table 4 and figure 5, we conclude that the visibility in the Arctic changes with a monthly feature and that using the multi-year data of one month as a training sample to infer the visibility of the month in which the training sample is located has a higher accuracy. By analyzing the temporal changes of visibility in the Arctic, we get that, in August, visibility changes slowly in the north of 80°N but quickly in the south and coastal areas, especially in the Kara Sea, Barents Sea and Norwegian Sea, which is consistent with the sailing experience in the Arctic Science Examination. However, visibility changes slowly around the Arctic in September. At the same time, visibility in the Arctic in July and August is lower than in other months of the year and the visibility in March and May is higher than in other months. So we get that visibility in the Arctic changes based on monthly features.

Point 12: 

LN324-325; grade 8-10 to level 1-10, and most severe changes happens…….

Response 12:

Very thanks for the suggestions of the reviewer. The conclusion is the result of the analysis of Figure 10 and we could clearly get that the yearly average of visibility in the Arctic mostly ranges from level 6 to level 8. I have revised the description of the finding to make it more clear. Relevant content is shown on line 397-398.

Point 13: 

Please provide your findings properly in an itemized way.

Response 13:

Very thanks for the suggestions of the reviewer. I have provided my findings properly in an itemized way. Relevant content is shown on line 378-398.

Round  2

Reviewer 2 Report

General Comments:

     The authors are to be commended for their work. They have addressed this reviewers concerns. This paper should be published. 

Author Response

Dear reviewer:

Thank you very much for your recognition and encouragement for me. 

Best regards!

First author of the paper

Reviewer 3 Report

I am satisfied with corrections. No more reviews from my side. Regards,