Review Reports - An Analysis of Fog in the Mainland Portuguese International Airports

Round 1

Reviewer 1 Report

This is an overall well written paper on fog classification and frequency at Lisbon and Porto International Airports. However it felt short in exploring ways how to usefully use this study results for developing methods for better forecast the fog occurrence, although it seems plausible that the connection made with weather types, large scale circulation patterns obtained by forecasting models should be the basis to use the work results, but how that could be done is not at all clear in the paper. So adding a couple of paragraphs on that would improve the paper usefulness, Faro airport was justifiable ignored in the main analysis as there are very few occurrences, however it would be interesting to know if those rare occurrences are randomly distributed among the weather type or occur essentially at a few specific weather types.

Author Response

We are very grateful for your kind and positive comments and suggestions.

We hope we were able to address all of them in a comprehensive way. We believe that the manuscript has been improved based on your suggestions.

Please see attachment.

Our best regards,

Pedro M.P. Guerreiro (on behalf of all authors)

Author Response File: Author Response.pdf

Reviewer 2 Report

Review of the manuscript atmosphere-953085

An analysis of fog in the mainland Portuguese International Airports

By Pedro M. P. Guerreiro * , Pedro M. M. Soares , Rita M. Cardoso , Alexandre M. Ramos

Summary: in the manuscript the authors use 10 years of METAR observations at Portugals three main airports in Faro, Porto and Lisbon to develop a climatology of the fog frequency and type according to the Tardiff and Rasmussen classification. These are then related to the weather conditions and the large scale flow pattern, mainly for Porto and Lisbon since the frequency of the fog at Faro is too small to develop statistics.

Despite I am confident in the results that are presented, and that the manuscript is well written, I find the results not surprising and not novel. In my view the analysis can be published elsewhere as a technical report or climate atlas, but not as a scientific paper since the paper does not offer new physical insight, neither new forecasting methods. The scientific community is producing so many manuscripts these days, while one hardly has time review and read it, and I find that journal papers should really have an added value in terms of new methods or physical insights or data analysis techniques, or surprising findings. More details are found below.

Recommendation: reject

Major points

The paper’s introduction and conclusion section both start with the usual story that fog is critical for the transportation sector, but that forecasting of fog is challenging. However, the paper does not offer any improvement in terms of improved forecasting by either better insights or better numerical or statistical modelling. Even with the observational data at hand one could have set up a statistical forecast like in Roman-Cascon et al (2016) (https://rmets.onlinelibrary.wiley.com/doi/full/10.1002/qj.2708) who built on the work of Menut et al (2014) (https://link.springer.com/article/10.1007/s10546-013-9875-1 ), and you could compare the applicability of the method for your sites compared to the ones in Paris, Cabauw and CIBA.
Ln 57: I am concerned about the word climatology since the climate is defined by 30 years of meteorology. Fog is one of these phenomena with high interannual variability, so I recommend not to call it a climatology. Also the paper does not defend why only these 10 years were chosen. The airports have a long history so more historical data should be present on the shelf, and at least the record could be continued to 2019. Or if the 10 years were selected for homogeneity in the observational methods or instruments, this should be mentioned.
Table 2: first and second row have the same message, row 2 is just row 1 divided by 10 which does not make much sense
Figure 1: the authors should discuss the sensitivity of their results to the selected classification. Alternative classifications like the GrossWetterlagen or Lamb Weather Types are also on the market. Would this result to different conclusions?
Ln 122 and further: the paper misses an opportunity to discuss the role of aerosol on fog formation. It is not only “weather” likely.
Ln 171-172: this is just a text that should be the figure caption. Can be removed from the main manuscript.
Figure 5: what is the unit of the color bars? What has been plotted in this color?
Figure 6 and further: For Lisbon most fog occurs for a ENE wind, which I can understand, but in the weather type analysis suddenly SW is advertised as one of the dominant wind directions. Is this a consistent result? What is the physical mechanism from an SW main wind to a cloud layer that lowers over Lisbon? Idem for ENE wind? The paper needs some more explanation about the physical mechanism.
Figure 9: is this only for the fog cases?
Ln 265: 14 Aug 2012 is outside your selected range of years! So more data are present.

Author Response

We are very grateful for your kind and positive comments and suggestions.

We hope we were able to address all of them in a comprehensive way. We believe that the manuscript has been improved based on your suggestions.

Please see attachment.

Our best regards,

Pedro M.P. Guerreiro (on behalf of all authors)

Response to Reviewer 2 Comments

Summary: in the manuscript the authors use 10 years of METAR observations at Portugal’s three main airports in Faro, Porto, and Lisbon to develop a climatology of the fog frequency and type according to the Tardif and Rasmussen classification. These are then related to the weather conditions and the large-scale flow pattern, mainly for Porto and Lisbon since the frequency of the fog at Faro is too small to develop statistics.

Response Summary: In Portugal, fog studies have been focused exclusively on the international airport of Lisbon. 1D high resolution simulations were preformed to characterize dynamical processes of the fog formation (Miranda and Teixeira, 2001), and performance assessment of Numerical Weather Prediction Models was carried out aiming forecasting tools improvement (Belo-Pereira, 2016). The present manuscript intends to expand the study of fog to the Portuguese international airports of Porto (at North) and Faro (at South) through a different approach. The connection of large-scale circulation patterns to the fog occurrence was preformed using weather types; the assessment of the local conditions prior to the fog occurrences using observational data, and the classification of fog types according to primary mechanisms closely related to the fog formation; the persistence and the intensity of all fog types, offers, on one hand, a new operational insight on fog forecasting as preconditioning diagnosis, and, on the other hand, enhancement of the numerical weather models performance.

Major points

The paper’s introduction and conclusion section both start with the usual story that fog is critical for the transportation sector, but that forecasting of fog is challenging. However, the paper does not offer any improvement in terms of improved forecasting by either better insights or better numerical or statistical modelling. Even with the observational data at hand one could have set up a statistical forecast like in Roman-Cascon et al (2016) (https://rmets.onlinelibrary.wiley.com/doi/full/10.1002/qj.2708) who built on the work of Menut et al (2014) (https://link.springer.com/article/10.1007/s10546-013-9875-1 ), and you could compare the applicability of the method for your sites compared to the ones in Paris, Cabauw and CIBA.

Response 1: The probability of the fog formation computed from a pdf as preconditioning diagnosis is indeed a very interesting approach. Actually, it was tested with METAR data, yet the half-hour METAR sampling was insufficient to produce reliable diagnosis.

Ln 57: I am concerned about the word climatology since the climate is defined by 30 years of meteorology. Fog is one of these phenomena with high interannual variability, so I recommend not to call it a climatology. Also the paper does not defend why only these 10 years were chosen. The airports have a long history so more historical data should be present on the shelf, and at least the record could be continued to 2019. Or if the 10 years were selected for homogeneity in the observational methods or instruments, this should be mentioned.

Response 2: We are very much in agreement in this point. Despite the aeronautical climatological period of five years or more, foreseen in the Technical Regulation Manual WMO nº.49, a larger dataset undoubtedly better supports the results and the study robustness. The observational data (METAR) from the three international airports from the period between 2002 and 2011, provided by the Portuguese weather service (IPMA), was tested in the beginning of this study, and its consistency enable to use it as the definitive dataset. However, the last 8 years METAR were recently available, but was not possible to include it in time in this study and expand the dataset up to 18 years.

Table 2: first and second row have the same message, row 2 is just row 1 divided by 10 which does not make much sense

Response 3: Despite being another format of the same information we intended to enclose it to the 10-year period.

Figure 1: the authors should discuss the sensitivity of their results to the selected classification. Alternative classifications like the GrossWetterlagen or Lamb Weather Types are also on the market. Would this result to different conclusions?

Response 4: The weather types used in this study were computed by an algorithm developed by Trigo and DaCamara (2000), from the Lamb weather type procedure. The classification produced 10 circulation types, which are very much representative of the different large-scale circulation patterns over the Iberian Peninsula. The original Lamb classification is centred over the British Islands, and the Western Europe, mainly the Iberian Peninsula, is not covered. On the other hand, the 30 types from the GrossWetterlagen classification would dissolve the circulation representativeness among the daily fog occurrences. Therefore, applying the GrossWetterlagen classification would result in inaccurate results.

Ln 122 and further: the paper misses an opportunity to discuss the role of aerosol on fog formation. It is not only “weather” likely.

Response 5: We completely agree. The nucleation mechanisms strongly depend on the aerosol as a condensation nuclei. The lack of aerosols observational data prevented to investigate the role of the aerosols in the classification of the fog types, for example. Likely in numerical models’ assessment, the local aerosols information would enhance the microphysical parameterization.

Ln 171-172: this is just a text that should be the figure caption. Can be removed from the main manuscript.

Response 6: Accepted.

Figure 5: what is the unit of the color bars? What has been plotted in this color?

Response 7: The unit of the colour bars is the number of the fog occurrences, and the colour gradient was chosen to better graphical reading.

Figure 6 and further: For Lisbon most fog occurs for a ENE wind, which I can understand, but in the weather type analysis suddenly SW is advertised as one of the dominant wind directions. Is this a consistent result? What is the physical mechanism from an SW main wind to a cloud layer that lowers over Lisbon? Idem for ENE wind? The paper needs some more explanation about the physical mechanism.

Response 8: Finding the connection between the large-scale circulation, the main water vapor sources and the fog formation, was a big challenge. This connection has revealed to be more obvious at Lisbon than at Porto and its explanation was included in the manuscript. The proximity from the Tagus estuary at ENE and the channelling effect of the advection from the sea, by the orography at SW, are the link between the fog formation and the SW flow, and the anticyclonic circulation. The SW flow is responsible for the moist air advection driven by a valley, from the sea towards to the airport plateau, as the anticyclone transports moisture from the nearby Tagus estuary.

Figure 9: is this only for the fog cases?

Response 9: Affirmative. The wind roses depict the wind direction and intensity observed during the fog events.

Ln 265: 14 Aug 2012 is outside your selected range of years! So more data are present

Response Ln265: Corrected to 2002.

Reviewer 3 Report

See attached file.

Comments for author File: Comments.pdf

Author Response

We are very grateful for your kind and positive comments and suggestions.

We hope we were able to address all of them in a comprehensive way. We believe that the manuscript has been improved based on your suggestions.

Please see attachment.

Our best regards,

Pedro M.P. Guerreiro (on behalf of all authors)

Response to Reviewer 3 Comments

Review of “An analysis of fog in mainland Portuguese International Airports”

The paper presents a climatological study of fog at Portuguese Airports; two in detail and one that is rarely impacted by fog and thus only briefly mentioned. The paper aims to examine the fog formation mechanism, the synoptic environment in which it forms and the local winds during fog events. A well-established fog type classification is adapted and applied to observations from the airports at Porto and Lisbon. Cloud base-lowering fog was found to be the dominant formation mechanism at both sites. Highlighted is the importance of local moisture advection in anticyclonic conditions. This type of climatological study can be an important tool to assist with the forecast of fog.

Response to general comment: In Portugal, fog studies have been focused exclusively on the international airport of Lisbon. 1D high resolution simulations were preformed to characterize dynamical processes of the fog formation (Miranda and Teixeira, 2001), and performance assessment of Numerical Weather Prediction Models was carried out aiming forecasting tools improvement (Belo-Pereira, 2016). The present manuscript intends to expand the study of fog to the Portuguese international airports of Porto (at North) and Faro (at South) through a different approach.

Broad comments

A broader literature review to include other climatological studies over Europe and other studies that use the Tardif and Rassmusen algorithm would be beneficial. Some suggestions are; Egli, S., Thies, B. and Bendix, J., 2019. A spatially explicit and temporally highly resolved analysis of variations in fog occurrence over Europe. Quarterly Journal of the Royal Meteorological Society, 145(721), pp.1721-1740.

Vautard, R., Yiou, P. and Van Oldenborgh, G.J., 2009. Decline of fog, mist and haze in Europe over the past 30 years. Nature Geoscience, 2(2), pp.115-119.

Belorid, M., Lee, C.B., Kim, J.C. and Cheon, T.H., 2015. Distribution and long-term trends in various fog types over South Korea. Theoretical and Applied Climatology, 122(3-4), pp.699-710.

Gu, Y., Kusaka, H. and Tan, J., 2019. Impacts of urban expansion on fog types in Shanghai, China: Numerical experiments by WRF model. Atmospheric Research, 220, pp.57-74.

Akimoto, Y. and Kusaka, H., 2015. A climatological study of fog in Japan based on event data. Atmospheric Research, 151, pp.200-211.

Policarpo, C., Salgado, R. and Costa, M.J., 2017. Numerical simulations of fog events in southern Portugal. Advances in Meteorology, 2017.

Response 1:

Further clarification is needed in the methods section. The % of data available should be presented and whether any fog cases were unclassified by the algorithm due to missing data. In section 2.2 an explanation of the method used by Ramos et al. would be beneficial. In section 2.3 the application of the Tardif and Rasmussen (2007) should be explained in further detail. The Tardif and Rasmussen (2007) algorithm uses the positive “M-of-N” construct which states that 3 out of 5 consecutive hourly visibility observations must be below 1600 m and 1 below 1000 m. It is not stated if these are the same criteria used in this study. It is not clear how this is adapted for 30 minute observations and the addition of the 10 minute observations. Throughout the paper it is unclear when a fog event uses the Tardif and Rasmussen definition or an alternative. For example figure 5 shows fog “events” which are less than three hours long but, by the Tardif and Rasmussen definition of a fog event, fog events are only classified as events if they are at least three hours long.

Response 2: The fog is mainly a short event, and the application of Tardif and Rasmussen event construct algorithm would reduce substantially the events frequency. Therefore, the algorithm is adapted in a way to gather fog occurrences separated less than 2 hours and define short isolated occurrences as events also. No matter how the fog event last, the criteria is exclusively based on the separation of the fog occurrences.

Currently the conclusion section provides a good summary of the results but does not place the work in the context of previous work. For example the conceptual model of fog formation at Lisbon is described again but it is not clear how the work has contributed further evidence to this model. Can you quantify how frequently fog is formed by this mechanism? Similar, it is stated there is not a similar description for Porto but in the abstract it is stated that moisture advection from the ocean at Porto is the dominant mechanism. This is not clearly described in the conclusions. A short overview of how the work will be used by forecasters and how the work informs future investigations of fog should be further outlined.

Response 3: At Lisbon, the conceptual model (Teixeira and Miranda, 2001) indicates the slow advection as the mechanism favouring the formation of fog. The results have showed that 19% of the fog events were classified as advection type.

The connection of large-scale circulation patterns to the fog occurrence was preformed using weather types; the assessment of the local conditions prior to the fog occurrences using observational data, and the classification of fog types according to primary mechanisms closely related to the fog formation; the persistence and the intensity of all fog types, offers, on one hand, a new operational insight on fog forecasting as preconditioning diagnosis, and, on the other hand, enhancement of the numerical weather models performance.

To better understand the main physical mechanisms driving the diverse types of fog in Lisbon and Porto, it is of paramount relevance to perform high resolution regional earth system modelling simulations. These would allow to represent at km-scale the local to regional atmospheric flow and the highly relevant coupling processes between the atmosphere-ocean/lake and land that are important in the locations here studied.

Specific comments

Line 18 - “enables to classify 89.9%” - “is used to classify the fog formation mechanism for 89.9%” would be better.

Response: Accepted

Line 23 - “has showed” - “revealed” would be better

Response: Accepted

Line 24 - “ The horizontal visibility values vary between 200 and 300 meters.” Is this referring to the mean, median or minimum visibility?

Response: Minimum visibility

Figure 1 – the mean sea-level pressure values are difficult to read. They are currently quite small and sometimes overlap.

Response: Corrected

Line 164 – 167. Figure 4 – Its not clear here the difference between daily fog occurrence and whole period fog occurrence. Additionally, it is unclear how this shows that fog is more daily recurrent at Lisbon.

Response: Concerning the frequency in the most common period, the frequency fog daily occurrences at Porto (79.4%) is lower than the frequency of the fog occurrences (including recurrent events in the same day) at Lisbon, which is 85.3%.

Line 177 – How many events are longer than 12 hours and when do these happen?

Response: 4 at Porto and 6 at Lisbon.

Figure 5 – A sequential colormap would improve this figure. Currently the reader is drawn to the times a single event occurs and not to the most frequent.

Response: Accepted

Line 183 – Its not clear at this stage of the paper that local advection is important.

Response: Accepted, however to wind roses show that the wind is important for moisture transportation.

Line 196-197 - “ The classification algorithm applied to Porto and Lisbon datasets enables to

classify 89.9% of the fog events at Porto and 92.4% at Lisbon.” This is grammatically incorrect–“The classification algorithm applied to the Porto and Lisbon datasets classifies 89.9% of the fog events at Porto and 92.4% at Lisbon.”

Response: Corrected

Line 199 – should be “at both airports” not “in”

Response: Corrected

Line 207 – 209 – Is this for both airports? Is there a difference between the number of cloud-base lowering events associated with the different wind speed thresholds at the two airports?

Response: Accepted

Line 217 - “slightly” – slight

Response: Corrected

Figure 7, 8 and 10 – It would be better to use the colors assigned to the different types in figure 3 for consistency.

Response: Accepted/Corrected

Line 226 – the weather types are presented in section 2.2.

Response: Corrected

Line 234 - “After analysing the fog large-scale favourable setting” - should be “After analysing the large-scale setting favourable for fog”.

Response: Corrected

Line 239 - “occur” should be “occurs”

Response: Corrected

Figure 9 – the brackets in the legend should be removed.

Response: Corrected

Line 258 - “ One type fog” should be “One fog type”.

Response: Corrected

Line 259 – Further justification for the choice of cases is needed. Do these cases fall into the most common time of year, wind speed, direction, weather type, onset and dissipation time for each type?

Response: Not for all the types of fog. At Porto, the precipitation type is a winter event, and the other types occur in the most common time of year. At Lisbon all types occur in the most common time of the year.

Figure 11 and 12 – the wind speed should be plotted from 0 to 360 not 350. Also can the lines that appear between some of the panels be removed?

Response: Corrected

Line 293 - “takes longer” should be “lasts longer”.

Response: Corrected

Line 310 - “ moist air as will exist over an estuary” should be “ moist air is also present over the estuary or coastal region”

Response: Corrected

Line 313 - “forecast” - “forecasts”

Response: Corrected

Line 327 – Changing this sentence to “The synoptic setting favourable for fog at the two locations is unsurprising.” would improve readability.

Response: Accepted

Line 358-359 – Remove this sentence. This is true but is just a consequence of the definitions used to define the fog types. Radiation fog by definition is related to radiative cooling.

Response: Accepted

Round 2

Reviewer 2 Report

Please see the attached file. Not much was changed to the manuscript, so there was little progress.

Response to Reviewer 2 Comments

Reviewer Response: I thank the authors for their further elaboration, however it does not address my concern! My main concern remains, i.e. what does this paper add to earlier works in terms of observational or modelling methods, and in terms of forecasting, or physical insights. The paper should really provide more new insights before it can merit publication in a scientific journal. Overall I have the impression the original manuscript hardly changed. Also some of the (valid) responses to my concerns have not been entrained in the paper, so the reader does not benefit from it.

Major points

The paper’s introduction and conclusion section both start with the usual story that fog is critical for the transportation sector, but that forecasting of fog is challenging. However, the paper does not offer any improvement in terms of improved forecasting by either better insights or better numerical or statistical modelling. Even with the observational data at hand one could have set up a statistical forecast like in Roman-Cascon et al (2016) (https://rmets.onlinelibrary.wiley.com/doi/full/10.1002/qj.2708) who built on the work of Menut et al (2014) (https://link.springer.com/article/10.1007/s10546-013-9875-1 ), and you could compare the applicability of the method for your sites compared to the ones in Paris, Cabauw and CIBA.

Reviewer response: I think the fact that you tested this method for your site and found (according to the authors) data resolution is too limited is already more novel than the rest of the current content of the paper. However, I am not sure why a higher time resolution than 30 min is needed? Earlier studies successfully applied this method using hourly data.

Ln 57: I am concerned about the word climatology since the climate is defined by 30 years of meteorology. Fog is one of these phenomena with high interannual variability, so I recommend not to call it a climatology. Also the paper does not defend why only these 10 years were chosen. The airports have a long history so more historical data should be present on the shelf, and at least the record could be continued to 2019. Or if the 10 years were selected for homogeneity in the observational methods or instruments, this should be mentioned.

Reviewer response: This is a very formal response in which the authors try to hide somehow behind the burocracy of the WMO. However, this manuscript is intended for a scientific journal that aims to bring the basic knowledge further. So I think a short sentence that reflects the caveat that the dataset is relatively short to consider climate scales is enough.

However, I find the last 8 years of which the data appear to be present MUST be included. Lack of time is not a scientific argument to improve the quality of the paper and the analysis.

Table 2: first and second row have the same message, row 2 is just row 1 divided by 10 which does not make much sense

Response 3: Despite being another format of the same information we intended to enclose it to the 10-year period.

Reviewer response: Yes I do understand that, but it has no meaning. The redundant line MUST be removed.

Figure 1: the authors should discuss the sensitivity of their results to the selected classification. Alternative classifications like the GrossWetterlagen or Lamb Weather Types are also on the market. Would this result to different conclusions?

Reviewer response: I do not understand the reasoning behind “On the other hand, the 30 types from the GrossWetterlagen classification would dissolve the circulation representativeness among the daily fog occurrences. Therefore, applying the GrossWetterlagen classification would result in inaccurate results.”. Although I may agree GrossWetterlagen are not the best for Portugal, it could well be that all your fog events end up in a small set of GrossWetterlagen. So the argument is still valid. At least add that the Trigo and DaCamara (2000) classification is based on the Lamb weather type strategy. This MUST be added

Ln 122 and further: the paper misses an opportunity to discuss the role of aerosol on fog formation. It is not only “weather” likely.

Ln 171-172: this is just a text that should be the figure caption. Can be removed from the main manuscript.

Response 6: Accepted.

Figure 5: what is the unit of the color bars? What has been plotted in this color?

Response 7: The unit of the colour bars is the number of the fog occurrences, and the colour gradient was chosen to better graphical reading.

Reviewer response: I understand this, but this extra explanation still does not appear in the caption. MUST be added!

Figure 6 and further: For Lisbon most fog occurs for a ENE wind, which I can understand, but in the weather type analysis suddenly SW is advertised as one of the dominant wind directions. Is this a consistent result? What is the physical mechanism from an SW main wind to a cloud layer that lowers over Lisbon? Idem for ENE wind? The paper needs some more explanation about the physical mechanism.

Reviewer response: it would be good to add this piece of discussion to the paper since it enhances the insight for the reader.

Figure 9: is this only for the fog cases?

Response 9: Affirmative. The wind roses depict the wind direction and intensity observed during the fog events.

Ln 265: 14 Aug 2012 is outside your selected range of years! So more data are present

Response Ln265: Corrected to 2002.

Author Response

Thank you very much for your comments.

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Although improvements to the manuscript have been made not all the comments have been sufficiently addressed. Please see the attached file.

Response to Reviewer 3 Comments

Review of “An analysis of fog in mainland Portuguese International Airports”

Response to Authors: The manuscript has been improved and several of the comments addressed. However, some of the comments have not been sufficiently address and further improvements are required. Please see the blue text blue highlighting further comments.

Broad comments

A broader literature review to include other climatological studies over Europe and other studies that use the Tardif and Rassmusen algorithm would be beneficial. Some suggestions are;

Egli, S., Thies, B. and Bendix, J., 2019. A spatially explicit and temporally highly resolved analysis of variations in fog occurrence over Europe. Quarterly Journal of the Royal Meteorological Society, 145(721), pp.1721-1740.

Vautard, R., Yiou, P. and Van Oldenborgh, G.J., 2009. Decline of fog, mist and haze in Europe over the past 30 years. Nature Geoscience, 2(2), pp.115-119.

Belorid, M., Lee, C.B., Kim, J.C. and Cheon, T.H., 2015. Distribution and long-term trends in various fog types over South Korea. Theoretical and Applied Climatology, 122(3-4), pp.699-710.

Gu, Y., Kusaka, H. and Tan, J., 2019. Impacts of urban expansion on fog types in Shanghai, China: Numerical experiments by WRF model. Atmospheric Research, 220, pp.57-74.

Akimoto, Y. and Kusaka, H., 2015. A climatological study of fog in Japan based on event data. Atmospheric Research, 151, pp.200-211.

Policarpo, C., Salgado, R. and Costa, M.J., 2017. Numerical simulations of fog events in southern Portugal. Advances in Meteorology, 2017.

Response 1:

Response to authors – please address this comment.

Further clarification is needed in the methods section. The % of data available should be presented and whether any fog cases were unclassified by the algorithm due to missing data. In section 2.2 an explanation of the method used by Ramos et al. would be beneficial. In section 2.3 the application of the Tardif and Rasmussen (2007) should be explained in further detail. The Tardif and Rasmussen (2007) algorithm uses the positive “M-of-N” construct which states that 3 out of 5 consecutive hourly visibility observations must be below 1600 m and 1 below 1000 m. It is not stated if these are the same criteria used in this study. It is not clear how this is adapted for 30 minute observations and the addition of the 10 minute observations. Throughout the paper it is unclear when a fog event uses the Tardif and Rasmussen definition or an alternative. For example figure 5 shows fog “events” which are less than three hours long but, by the Tardif and Rasmussen definition of a fog event, fog events are only classified as events if they are at least three hours long.

Response to authors – The application of the Tardif & Rasmussen (2007) algorithm has been clarified in the paper. However, the data availability has not been included and should be. Similarly, no further explanation of the method used to derive the weather types has been included nor a justification for the choice of method. This should be included.

Currently the conclusion section provides a good summary of the results but does not place the work in the context of previous work. For example the conceptual model of fog formation at Lisbon is described again but it is not clear how the work has contributed further evidence to this model. Can you quantify how frequently fog is formed by this mechanism? Similar, it is stated there is not a similar description for Porto but in the abstract it is stated that moisture advection from the ocean at Porto is the dominant mechanism. This is not clearly described in the conclusions. A short overview of how the work will be used by forecasters and how the work informs future investigations of fog should be further outlined.

Response to authors – This should be explicitly stated in the paper.

Response to authors– The conclusion section has been improved compared to the original.

Specific comments

Line 18 - “enables to classify 89.9%” - “is used to classify the fog formation mechanism for 89.9%” would be better.

Response: Accepted

Line 23 - “has showed” - “revealed” would be better

Response: Accepted

Line 24 - “ The horizontal visibility values vary between 200 and 300 meters.” Is this referring to the mean, median or minimum visibility?

Response: Minimum visibility

Figure 1 – the mean sea-level pressure values are difficult to read. They are currently quite small and sometimes overlap.

Response: Corrected

Response to authors – The text is clearer now.

Line 177 – How many events are longer than 12 hours and when do these happen?

Response: 4 at Porto and 6 at Lisbon.

Figure 5 – A sequential colormap would improve this figure. Currently the reader is drawn to the times a single event occurs and not to the most frequent.

Response: Accepted

Response to authors – The figure is still the same as the original. Please outline why this has not been changed.

Line 183 – Its not clear at this stage of the paper that local advection is important.

Response: Accepted, however to wind roses show that the wind is important for moisture transportation.

Line 196-197 - “ The classification algorithm applied to Porto and Lisbon datasets enables to

Response: Corrected

Line 199 – should be “at both airports” not “in”

Response: Corrected

Line 207 – 209 – Is this for both airports? Is there a difference between the number of cloud-base lowering events associated with the different wind speed thresholds at the two airports?

Response: Accepted

Response to authors – There is now repetition here. The number at Porto is stated twice. Also line 212 should say cloud top cooling not “topclouds cooling”.

Line 217 - “slightly” – slight

Response: Corrected

Figure 7, 8 and 10 – It would be better to use the colors assigned to the different types in figure 3 for consistency.

Response: Accepted/Corrected

Line 226 – the weather types are presented in section 2.2.

Response: Corrected

Line 234 - “After analysing the fog large-scale favourable setting” - should be “After analysing the large-scale setting favourable for fog”.

Response: Corrected

Line 239 - “occur” should be “occurs”

Response: Corrected

Figure 9 – the brackets in the legend should be removed.

Response: Corrected

Line 258 - “ One type fog” should be “One fog type”.

Response: Corrected

Response to authors – Further justification is still needed in the manuscript. What about wind speed, direction, weather type, onset and dissipation time for each type?

Figure 11 and 12 – the wind speed should be plotted from 0 to 360 not 350. Also can the lines that appear between some of the panels be removed?

Response: Corrected

Line 293 - “takes longer” should be “lasts longer”.

Response: Corrected

Line 310 - “ moist air as will exist over an estuary” should be “ moist air is also present over the estuary or coastal region”

Response: Corrected

Line 313 - “forecast” - “forecasts”

Response: Corrected

Line 327 – Changing this sentence to “The synoptic setting favourable for fog at the two locations is unsurprising.” would improve readability.

Response: Accepted

Line 358-359 – Remove this sentence. This is true but is just a consequence of the definitions used to define the fog types. Radiation fog by definition is related to radiative cooling.

Response: Accepted

Author Response

Thank you very much for your comments.

Please see the attachment.

Author Response File: Author Response.pdf

Round 3

Reviewer 2 Report

Review of 3^rd version of atmosphere-953085

I am grateful the authors have extended the analysis with more data that are available, which had lead to a change of some conclusions about the main driving processes behind the fog. I recommend to publish this paper after some minor editorial corrections:

Remarks:

Ln 11, 12: process-> phenomenon

Ln 56: stablished -> established (typo)

Ln 60: wind intensity, temperature, and dew point-> 10-m wind speed, 2 m air temperature and 2-m dewpoint temperature

P8: layout: at the bottom of this page the names of two airports are listed but I think they below to the figure on top of page 9. The same happens at page 12/13.

Ln 290: choice criteria -> selected criteria

Caption figure 11: please add that you show 10-m wind speed, 2-m air temperature and 2-m dew point temperature.

Ln 367: In average -> On average

Ln 375: local -> locally

Author Response

Dear Reviewer,

I am very grateful for your valuable and generous support. The manuscript has been greatly improved from your review.

The errors pointed out in the last remarks were corrected in the manuscript.

Most recent changes are highlighted in blue.

On behalf of the other authors, thank you very much.

Kindly regards

Pedro Guerreiro

Reviewer 3 Report

Overall the paper has improved. There is one minor points that I believe would improve the paper further. Please see the attached file.

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

I am very grateful for your valuable and generous support. The manuscript has been greatly improved from your review.

The suggested literarure has enhanced the introductory insight of the manuscript.

Most recent changes are highlighted in blue.

On behalf of the other authors, thank you very much.

Kindly regards

Pedro Guerreiro