the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Oct 2023
On the production and validation of satellite based UV index maps
Abstract. This paper presents a method to calculate surface UV index maps from SEVIRI satellite images using a very simplified approach to account for clouds. We compare the resulting maps, which are produced near-real time operationally, to ground measurements from the Austrian UV network and affiliated stations. While the data agrees well for low altitude stations and in clear sky conditions, discrepancy at mountain stations and in certain cloud conditions between the map pixel and ground measurement UV indices gets higher. We discuss the sources of uncertainty in both values in detail, and highlight why a direct comparison of absolute UV index values for validating purposes is inadequate.
- Preprint
(23288 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on amt-2023-188', Anonymous Referee #1, 05 Nov 2023
The current manuscript presents a method to calculate surface UV index from SEVIRI satellite images at 600 nm using a simple linear method to account for cloud effects. The second part of the manuscript discussed the uncertainties in the comparison between modelled and measured UV index which are very useful to understand the difficulties of validation of satellite derived UV index. The manuscript is well written. However, I have some concerns about the proposed method. I would suggest major modification before the manuscript is able to be published.
Major points:
L28-30 & L235-236: One major point is why proposing such a simplified method? What is the advantage/disadvantage of this method comparing to the two lookup tables approach? Dose the gain of calculation time or other factors outweighs the lost of accuracy? More justifications or defined objectives are needed here.
L69-72: For the results presented in Figure 2, the influences of surface albedo and liquid water content are examined. But many other factors are not clear here, e.g. the variation of cloud altitude and thickness, cloud ice water content for high clouds, solar zenith angle as well as aerosol properties. I believe that there should be some investigations about those factors to support the use of this linear method or to be clearer about the deficiency of using the method.
Minor points:
End of the section 1: Maybe add a conventional paragraph to describe the structure of this manuscript?
L97-98: Any possible explanations for this overestimation? Is this caused by the errors of aerosol amount or ozone?
L99: Change “the ground” to “the ground UV” at the end of the phrase?
L168-169: I'm not sure about this argument. The goal of having satellite based UV index is to provide estimations of what will be observe at ground level (although they include errors too). If satellite UV index is different from the ground observation, then it should be that the satellite UV index has deficiencies, either due to the modeling methods or the limits of satellite instrument itself, e.g. SEVIRI's capability to capture small scale cumulus clouds.
L198-200: The proposed method itself has significant deficiency w.r.t. albedo based on Fig. 2. This should be also mentioned.
Figure 4: I suggest to add legend of point colors into the figure for easier reading.
Figure 6,7: maybe it’s better to add grids in the figure.
Citation: https://doi.org/10.5194/amt-2023-188-RC1 -
AC2: 'Reply on RC1', Verena Schenzinger, 13 Dec 2023
The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2023-188/amt-2023-188-AC2-supplement.pdf
-
AC2: 'Reply on RC1', Verena Schenzinger, 13 Dec 2023
-
RC2: 'Comment on amt-2023-188', Anonymous Referee #2, 07 Nov 2023
Review of the manuscript titled "On the production and validation of satellite-based UV index maps" of Schenzinger et al.
The manuscript presents a method for generating high-resolution UV index maps of Europe at 15-minute intervals. The method is tuned to enhance computational efficiency. The work is important as such timely UV index maps disseminate important information about the actual UV level to the general public. However, before publication, the manuscript should be improved taking into account both the general and specific comments.
General comments:
The manuscript's title should be revised to better encapsulate the essence of the research.
Moreover, the abstract should be rewritten to include the most important results in a more precise way. The plots are good and comprehensive, but please check that all of them are appropriately referenced within the text. Additionally, the methodology should be explained more clearly, please see the specific comments. I think the suggestion of using CMF for validation of satellite UV should already be included in the abstract.Furthermore, it is essential to elucidate the practical applications of these UV index maps and their significance, particularly with regard to the general public.
Specific comments:Abstract:
Please be more specific: ..." data agrees well" -> show numbers
..." gets higher" -> how much higher?
Also, describe the source of uncertainty and how much they affectPage 1, line 11, I think you should add a sentence or two about the Arctic ozone loss which occurs almost every spring (not only in 2020). And please check the reason for the record low ozone e.g. in Benrhard et al., 2020, and references therein:
Bernhard, G. H., Fioletov, V. E., Grooß, J.-U., Ialongo, I., Johnsen, B., Lakkala, K., Manney, G., Müller, R., Svendby, T. : Record-breaking increases in Arctic solar ultraviolet radiation caused by exceptionally large ozone depletion in 2020. Geophysical Research Letters, 47, e2020GL090844. https://doi.org/10.1029/2020GL090844 , 2020
For Arctic springtime ozone loss in general, see e.g.,
Bernhard G., Fioletov V., Grooss J.-U., Ialongo I, Johnsen B, Lakkala K, Manney G., Müller R., Svendby, T., 2023: Ozone and UV radiation [in ”State of the Climate in 2022”], Bull. Amer. Meteor. Soc., 104 (9), S 308 -S 310 , https://doi.org/10.1175/2023BAMSStateoftheClimate.1.
Page 1, line 20: You could add that those proxy data can be used as input to radiative transfer models to produce satellite UV products, from which the maps can be plotted. And add references to TOMS/OMI/TROPOMI algorithm papers.
Page 1, line 22: "A lot of previous works distinguish cloud-free from cloudy situations by employing a radiative transfer model for clear sky calculations and a separate one to account for the cloud effects (Verdebout, 2000; Schallhart et al., 2008; Chubarova et al., 2012;
Lakkala et al., 2020)". This is not the case for the TROPOMI algorithm (Lakkala et al., 2020). For TROPOMI, there are two lookup tables, the first calculates cloud optical depth, and the second one directly all sky UV. Clear sky UV is not calculated separately. See e.g.Lindfors, A. V., Kujanpää, J., Kalakoski, N., Heikkilä, A., Lakkala, K., Mielonen, T., Sneep, M., Krotkov, N. A., Arola, A., and Tamminen, J.: The TROPOMI surface UV algorithm, Atmos. Meas. Tech., 11, 997–1008, https://doi.org/10.5194/amt-11-997-2018, 2018.
Page 2: line 41: ..." especially the ozone concentration, but also aerosol optical depth." -> + pollutants?
Page 1 and 2: It's difficult to follow what is new in your study compared to the method of Schallhart et al., 2008. Please reformulate the way you explain an already existing method (Schallhart et al., 2008) and your new method. E.g. in line 49 you write "We only do this calculation..." Do you mean in Schallhart et al., 2008 or is it something new that your study implements?
Page 2, line 55: Specify the inputs you use from CAMS: total ozone ....+ ...+
Do you get the beta parameter also from CAMS or from where?Page 3, lines 69-79: It took me a lot of time to understand what you have done in Figure 2. And still, it's not clear if the plot is only based on model calculation. Is the text related to ground measurements related to Fig. 2? I suppose not. Then I suggest that the place of the ground instrumentation is not in this Section 2.1. If you think they are in the right place, then open up more, about how you use them.
Figure 2: Why don't you make a plot of the CMFs? x-axis calculated with scaled 600nm radiance and y-axis with 300nm clear sky/all sky?
Figure 2: Please explain the nonlinearity due to higher ground albedo.
Page 3, line 72: "To be able to compare the results for different albedos.." -> Please open a little bit: Do you mean different cloudiness + ground albedo conditions, which are seen from the satellite as "one combined albedo"?
Page 4, line 92: Please include the uncertainty or error range for all r-values you show.
Page 4, line 97: "the clear sky model has a positive bias, i.e. the satellite map.." I wonder, if you say that the clear sky model is based on input from CAMS, shouldn't you take directly the CAMS clear sky UV index product? Anyway, the bias is mostly due to the bias in the input of the clear sky calculations.
Page 5, line 121: "as large differences" -> as large absolute differences (when you look at relative differences it's the other way round).
Page 5, Figure 5 is not referred.
Page 5. I don't really see the point of comparing a satellite retrieval with 30-minute average ground measurements. The ground measurements should be 1-5 min data or 1 scan or so.
Figure 8: Why the satellite-model didn't capture the rain at 14 UTC? From the all-sky camera, it seems to have been an overcast situation.
Page 6, line 162: "In these particular meteorological conditions..." Please specify which kind of conditions
Page 6, line 193: "That leaves an erroneous input for aerosol as the source of error" -> What do you mean?Page 7, line 192/199: "Can't you force the albedo to be around 0.05 for snow-free surface?
Page 7, paragraph starting at line 214: You should specify that you are first talking of the ground albedo-> modeled clear sky UV index. And that in the next sentence you talk about the albedo the satellite sees (600nm). From the reflected radiation, the satellite can't know if it's reflected from clouds or from snow. For the satellite, clouds and snow look similar - a reflecting surface. At least this is the case for OMI/TROPOMI instruments, and they underestimate UVindex over snow as they can interpret snow erroneously as a cloud on a clear day.
e.g.Bernhard, G., Arola, A., Dahlback, A., Fioletov, V., Heikkilä, A., Johnsen, B., Koskela, T., Lakkala, K., Svendby, T., and Tamminen, J.: Comparison of OMI UV observations with ground-based measurements at high northern latitudes, Atmos. Chem. Phys., 15, 7391-7412, doi:10.5194/acp-15-7391-2015, 2015.
or
Lakkala et al. 2020If you mean something else, please rephrase.
Page 8, line 232: "This approach itself is widely employed...". Not in the TROPOMI algorithm (Lakkala et al.,2020).
Data availability: It's not clear if the presented method is already in operational use at https://uv-index.at/map/ ?
Citation: https://doi.org/10.5194/amt-2023-188-RC2 -
AC1: 'Reply on RC2', Verena Schenzinger, 13 Dec 2023
The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2023-188/amt-2023-188-AC1-supplement.pdf
-
AC3: 'Reply on RC2', Verena Schenzinger, 13 Dec 2023
The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2023-188/amt-2023-188-AC3-supplement.pdf
-
AC1: 'Reply on RC2', Verena Schenzinger, 13 Dec 2023
-
EC1: 'Editor comment on amt-2023-188', Marloes Penning de Vries, 14 Dec 2023
The submitted manuscript presents a method for the production of UV index maps from geo-stationary satellite data in a manner that is computationally more efficient than previous methods. Moreover, and in the scope of the discussion of validation results, the authors point out a number of factors complicating the comparison between ground-based and satellite-based data.
Although the manuscript is well written and the described methodology sound, I see a number of major shortcomings, some of which are mentioned by the reviewers as well:
1. The newly developed algorithm appears to be an update of an operational algorithm (which applies an approach based on two look-up-tables), but the "old" algorithm is not sufficiently described, hence, it is difficult to put the changes into perspective. More importantly, no comparison with the "old" algorithm is presented. In particular, since the authors rate the computational efficiency of the new approach a distinct improvement, there should have been a comparison with the "old" algorithm (in general, LUT approaches are not particularly heavy, although this depends on the size and dimensions of the tables).
2. The authors make an argument against the validation of satellite data using "ground data" as a reference in the case of variables like the UV index, noting that this is "basically all that is available". I would argue that, like for precipitation measured by a rain gauge, the ground-based observation of UV index is exactly the quantity one is interested in. That the quantity is difficult to obtain with similar accuracy from satellite data is, like for precipitation, a fact that needs to be taken into account, but cannot be avoided for obvious physical reasons. The discussion in sections 2.2.1 and 2.2.2 lists a number of valid arguments as to why satellite observations of variables that are defined at the ground (such as the UV index) are difficult to validate - but these arguments are not novel to the satellite retrieval community, which has been making numerous efforts in the past decades to validate aerosol load, precipitation rates - amongst many others. Similarly, the issue of data interpolation (for which, I would point out, more routines exist than the pyresample module has to offer) and the related issue of averaging has been visited by many remote sensing scientists in the past.
3. Lastly, I would like to ask the authors which users they are tailoring the UVI to. In the conclusions they mention: "public communication and for health purposes", implying a rather qualitative use. If this is the case, users are in all likelihood interested in if UVI>5 or UVI<2, for example, and not interested in the difference between 4.5 and 5 - it's just an index after all, with no clear physical definition. Then why do the authors go through so much trouble to make an accurate determination of the UVI and its uncertainties?
To summarize, I believe the described updated method lacks sufficient comparison with previous (and possibly other existing) methods; and apart from that, the manuscript offers only few novel insights for the AMT reader community.
Kind regards,
Marloes Penning de Vries
Citation: https://doi.org/10.5194/amt-2023-188-EC1
Status: closed
-
RC1: 'Comment on amt-2023-188', Anonymous Referee #1, 05 Nov 2023
The current manuscript presents a method to calculate surface UV index from SEVIRI satellite images at 600 nm using a simple linear method to account for cloud effects. The second part of the manuscript discussed the uncertainties in the comparison between modelled and measured UV index which are very useful to understand the difficulties of validation of satellite derived UV index. The manuscript is well written. However, I have some concerns about the proposed method. I would suggest major modification before the manuscript is able to be published.
Major points:
L28-30 & L235-236: One major point is why proposing such a simplified method? What is the advantage/disadvantage of this method comparing to the two lookup tables approach? Dose the gain of calculation time or other factors outweighs the lost of accuracy? More justifications or defined objectives are needed here.
L69-72: For the results presented in Figure 2, the influences of surface albedo and liquid water content are examined. But many other factors are not clear here, e.g. the variation of cloud altitude and thickness, cloud ice water content for high clouds, solar zenith angle as well as aerosol properties. I believe that there should be some investigations about those factors to support the use of this linear method or to be clearer about the deficiency of using the method.
Minor points:
End of the section 1: Maybe add a conventional paragraph to describe the structure of this manuscript?
L97-98: Any possible explanations for this overestimation? Is this caused by the errors of aerosol amount or ozone?
L99: Change “the ground” to “the ground UV” at the end of the phrase?
L168-169: I'm not sure about this argument. The goal of having satellite based UV index is to provide estimations of what will be observe at ground level (although they include errors too). If satellite UV index is different from the ground observation, then it should be that the satellite UV index has deficiencies, either due to the modeling methods or the limits of satellite instrument itself, e.g. SEVIRI's capability to capture small scale cumulus clouds.
L198-200: The proposed method itself has significant deficiency w.r.t. albedo based on Fig. 2. This should be also mentioned.
Figure 4: I suggest to add legend of point colors into the figure for easier reading.
Figure 6,7: maybe it’s better to add grids in the figure.
Citation: https://doi.org/10.5194/amt-2023-188-RC1 -
AC2: 'Reply on RC1', Verena Schenzinger, 13 Dec 2023
The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2023-188/amt-2023-188-AC2-supplement.pdf
-
AC2: 'Reply on RC1', Verena Schenzinger, 13 Dec 2023
-
RC2: 'Comment on amt-2023-188', Anonymous Referee #2, 07 Nov 2023
Review of the manuscript titled "On the production and validation of satellite-based UV index maps" of Schenzinger et al.
The manuscript presents a method for generating high-resolution UV index maps of Europe at 15-minute intervals. The method is tuned to enhance computational efficiency. The work is important as such timely UV index maps disseminate important information about the actual UV level to the general public. However, before publication, the manuscript should be improved taking into account both the general and specific comments.
General comments:
The manuscript's title should be revised to better encapsulate the essence of the research.
Moreover, the abstract should be rewritten to include the most important results in a more precise way. The plots are good and comprehensive, but please check that all of them are appropriately referenced within the text. Additionally, the methodology should be explained more clearly, please see the specific comments. I think the suggestion of using CMF for validation of satellite UV should already be included in the abstract.Furthermore, it is essential to elucidate the practical applications of these UV index maps and their significance, particularly with regard to the general public.
Specific comments:Abstract:
Please be more specific: ..." data agrees well" -> show numbers
..." gets higher" -> how much higher?
Also, describe the source of uncertainty and how much they affectPage 1, line 11, I think you should add a sentence or two about the Arctic ozone loss which occurs almost every spring (not only in 2020). And please check the reason for the record low ozone e.g. in Benrhard et al., 2020, and references therein:
Bernhard, G. H., Fioletov, V. E., Grooß, J.-U., Ialongo, I., Johnsen, B., Lakkala, K., Manney, G., Müller, R., Svendby, T. : Record-breaking increases in Arctic solar ultraviolet radiation caused by exceptionally large ozone depletion in 2020. Geophysical Research Letters, 47, e2020GL090844. https://doi.org/10.1029/2020GL090844 , 2020
For Arctic springtime ozone loss in general, see e.g.,
Bernhard G., Fioletov V., Grooss J.-U., Ialongo I, Johnsen B, Lakkala K, Manney G., Müller R., Svendby, T., 2023: Ozone and UV radiation [in ”State of the Climate in 2022”], Bull. Amer. Meteor. Soc., 104 (9), S 308 -S 310 , https://doi.org/10.1175/2023BAMSStateoftheClimate.1.
Page 1, line 20: You could add that those proxy data can be used as input to radiative transfer models to produce satellite UV products, from which the maps can be plotted. And add references to TOMS/OMI/TROPOMI algorithm papers.
Page 1, line 22: "A lot of previous works distinguish cloud-free from cloudy situations by employing a radiative transfer model for clear sky calculations and a separate one to account for the cloud effects (Verdebout, 2000; Schallhart et al., 2008; Chubarova et al., 2012;
Lakkala et al., 2020)". This is not the case for the TROPOMI algorithm (Lakkala et al., 2020). For TROPOMI, there are two lookup tables, the first calculates cloud optical depth, and the second one directly all sky UV. Clear sky UV is not calculated separately. See e.g.Lindfors, A. V., Kujanpää, J., Kalakoski, N., Heikkilä, A., Lakkala, K., Mielonen, T., Sneep, M., Krotkov, N. A., Arola, A., and Tamminen, J.: The TROPOMI surface UV algorithm, Atmos. Meas. Tech., 11, 997–1008, https://doi.org/10.5194/amt-11-997-2018, 2018.
Page 2: line 41: ..." especially the ozone concentration, but also aerosol optical depth." -> + pollutants?
Page 1 and 2: It's difficult to follow what is new in your study compared to the method of Schallhart et al., 2008. Please reformulate the way you explain an already existing method (Schallhart et al., 2008) and your new method. E.g. in line 49 you write "We only do this calculation..." Do you mean in Schallhart et al., 2008 or is it something new that your study implements?
Page 2, line 55: Specify the inputs you use from CAMS: total ozone ....+ ...+
Do you get the beta parameter also from CAMS or from where?Page 3, lines 69-79: It took me a lot of time to understand what you have done in Figure 2. And still, it's not clear if the plot is only based on model calculation. Is the text related to ground measurements related to Fig. 2? I suppose not. Then I suggest that the place of the ground instrumentation is not in this Section 2.1. If you think they are in the right place, then open up more, about how you use them.
Figure 2: Why don't you make a plot of the CMFs? x-axis calculated with scaled 600nm radiance and y-axis with 300nm clear sky/all sky?
Figure 2: Please explain the nonlinearity due to higher ground albedo.
Page 3, line 72: "To be able to compare the results for different albedos.." -> Please open a little bit: Do you mean different cloudiness + ground albedo conditions, which are seen from the satellite as "one combined albedo"?
Page 4, line 92: Please include the uncertainty or error range for all r-values you show.
Page 4, line 97: "the clear sky model has a positive bias, i.e. the satellite map.." I wonder, if you say that the clear sky model is based on input from CAMS, shouldn't you take directly the CAMS clear sky UV index product? Anyway, the bias is mostly due to the bias in the input of the clear sky calculations.
Page 5, line 121: "as large differences" -> as large absolute differences (when you look at relative differences it's the other way round).
Page 5, Figure 5 is not referred.
Page 5. I don't really see the point of comparing a satellite retrieval with 30-minute average ground measurements. The ground measurements should be 1-5 min data or 1 scan or so.
Figure 8: Why the satellite-model didn't capture the rain at 14 UTC? From the all-sky camera, it seems to have been an overcast situation.
Page 6, line 162: "In these particular meteorological conditions..." Please specify which kind of conditions
Page 6, line 193: "That leaves an erroneous input for aerosol as the source of error" -> What do you mean?Page 7, line 192/199: "Can't you force the albedo to be around 0.05 for snow-free surface?
Page 7, paragraph starting at line 214: You should specify that you are first talking of the ground albedo-> modeled clear sky UV index. And that in the next sentence you talk about the albedo the satellite sees (600nm). From the reflected radiation, the satellite can't know if it's reflected from clouds or from snow. For the satellite, clouds and snow look similar - a reflecting surface. At least this is the case for OMI/TROPOMI instruments, and they underestimate UVindex over snow as they can interpret snow erroneously as a cloud on a clear day.
e.g.Bernhard, G., Arola, A., Dahlback, A., Fioletov, V., Heikkilä, A., Johnsen, B., Koskela, T., Lakkala, K., Svendby, T., and Tamminen, J.: Comparison of OMI UV observations with ground-based measurements at high northern latitudes, Atmos. Chem. Phys., 15, 7391-7412, doi:10.5194/acp-15-7391-2015, 2015.
or
Lakkala et al. 2020If you mean something else, please rephrase.
Page 8, line 232: "This approach itself is widely employed...". Not in the TROPOMI algorithm (Lakkala et al.,2020).
Data availability: It's not clear if the presented method is already in operational use at https://uv-index.at/map/ ?
Citation: https://doi.org/10.5194/amt-2023-188-RC2 -
AC1: 'Reply on RC2', Verena Schenzinger, 13 Dec 2023
The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2023-188/amt-2023-188-AC1-supplement.pdf
-
AC3: 'Reply on RC2', Verena Schenzinger, 13 Dec 2023
The comment was uploaded in the form of a supplement: https://amt.copernicus.org/preprints/amt-2023-188/amt-2023-188-AC3-supplement.pdf
-
AC1: 'Reply on RC2', Verena Schenzinger, 13 Dec 2023
-
EC1: 'Editor comment on amt-2023-188', Marloes Penning de Vries, 14 Dec 2023
The submitted manuscript presents a method for the production of UV index maps from geo-stationary satellite data in a manner that is computationally more efficient than previous methods. Moreover, and in the scope of the discussion of validation results, the authors point out a number of factors complicating the comparison between ground-based and satellite-based data.
Although the manuscript is well written and the described methodology sound, I see a number of major shortcomings, some of which are mentioned by the reviewers as well:
1. The newly developed algorithm appears to be an update of an operational algorithm (which applies an approach based on two look-up-tables), but the "old" algorithm is not sufficiently described, hence, it is difficult to put the changes into perspective. More importantly, no comparison with the "old" algorithm is presented. In particular, since the authors rate the computational efficiency of the new approach a distinct improvement, there should have been a comparison with the "old" algorithm (in general, LUT approaches are not particularly heavy, although this depends on the size and dimensions of the tables).
2. The authors make an argument against the validation of satellite data using "ground data" as a reference in the case of variables like the UV index, noting that this is "basically all that is available". I would argue that, like for precipitation measured by a rain gauge, the ground-based observation of UV index is exactly the quantity one is interested in. That the quantity is difficult to obtain with similar accuracy from satellite data is, like for precipitation, a fact that needs to be taken into account, but cannot be avoided for obvious physical reasons. The discussion in sections 2.2.1 and 2.2.2 lists a number of valid arguments as to why satellite observations of variables that are defined at the ground (such as the UV index) are difficult to validate - but these arguments are not novel to the satellite retrieval community, which has been making numerous efforts in the past decades to validate aerosol load, precipitation rates - amongst many others. Similarly, the issue of data interpolation (for which, I would point out, more routines exist than the pyresample module has to offer) and the related issue of averaging has been visited by many remote sensing scientists in the past.
3. Lastly, I would like to ask the authors which users they are tailoring the UVI to. In the conclusions they mention: "public communication and for health purposes", implying a rather qualitative use. If this is the case, users are in all likelihood interested in if UVI>5 or UVI<2, for example, and not interested in the difference between 4.5 and 5 - it's just an index after all, with no clear physical definition. Then why do the authors go through so much trouble to make an accurate determination of the UVI and its uncertainties?
To summarize, I believe the described updated method lacks sufficient comparison with previous (and possibly other existing) methods; and apart from that, the manuscript offers only few novel insights for the AMT reader community.
Kind regards,
Marloes Penning de Vries
Citation: https://doi.org/10.5194/amt-2023-188-EC1
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 498 | 167 | 39 | 704 | 34 | 33 |
- HTML: 498
- PDF: 167
- XML: 39
- Total: 704
- BibTeX: 34
- EndNote: 33
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
