Investigation of the effects of the Greek extreme wildfires of August 2021 on air quality and spectral solar irradiance
Abstract. In August 2021, a historic heatwave was recorded in Greece which resulted in extreme wildfire events that strongly affected the air quality over the city of Athens. Saharan dust was also transferred over Greece in the same period due to the prevailing southern winds. The impact of these events on air quality and surface solar radiation are investigated in this study. Event characterization based on active and passive remote sensing instrumentation has been performed. The study shows that significantly increased levels of air pollution were recorded during the end of July/first week of August. The smoke led to unusually high AOD values (up to 3.6), high Ångström Exponent (AE) (up to 2.4) and a strong and negative dependence of single scattering albedo (SSA) on wavelength that was observed to decrease from 0.93 at 440 nm to 0.86 at 1020 nm signifying the presence of strong absorbing aerosols. While, the dust event led to high AOD (up to 1.4), low AE (up to 0.9) and positive dependence of SSA on wavelength that was observed to increase from 0.89 at 440 nm to 0.95 at 1020 nm indicating large forward scattering due to coarse particles. Furthermore, the analysis of the smoke aerosol optical properties during the transfer from the source to a distance of about 240 km revealed that the SSA and AE changed significantly during the transfer, which lasted approximately 9 h. The transport of the plume led to an impressive change in the spectral shape of SSA whose value significantly increased pointing to the aging of smoke and the dilution of plumes while the transport. The impact of dust and smoke on spectral solar irradiance reveals significant differences in the spectral shape of attenuation caused by the two different aerosol species. The attenuation of solar irradiance in UV-B irradiance was found to be least in case of dust and highest due to smoke (up to 60 % or more) and intermediate in the case of a mixture of smoke and dust. The attenuation was comparatively less in NIR region (mostly within 20 % but it even reached up to 40 % in the presence of smoke) and VIS region (but greater than NIR region). Also, the AOD variations from climatology led to decrease in UV Index up to 53 %, in vitamin-D up to 50 %, in photosynthetically active radiation up to 21 % and in GHI up to 17 %, with implications on health, agriculture and energy. This study highlights the wider impacts of wildfires that are part of the wider problem of the Mediterranean countries, whose frequency is predicted to increase in view of the projected increasing occurrence of summer heatwaves.
Akriti Masoom et al.
Status: open (until 24 Mar 2023)
- RC1: 'Comment on egusphere-2023-180', Anonymous Referee #1, 28 Feb 2023 reply
- RC2: 'Comment on egusphere-2023-180', Anonymous Referee #2, 13 Mar 2023 reply
Akriti Masoom et al.
Akriti Masoom et al.
Viewed (geographical distribution)
The manuscript of Masoom et al., “Investigation of the effects of the Greek extreme wildfires of August 2021 on air quality and spectral solar irradiance” presents an analysis of various air quality, aerosol, and radiation measurements during wildfire events in August 2021 in Greece. These wildfire episodes offer interesting cases to study, and indeed nice variety of measurements were utilized. However, there are problems in the current form of the manuscript, which made a thorough and proper evaluation difficult. It seemed that the analysis was not done carefully and thoroughly enough and then an unfinished version of the manuscript was submitted.
It was possible to me as a referee (and it will be possible to the readers of the final manuscript) to try to better understand the analysis regarding the AERONET measurements since the data are publicly available. Next, I will mainly clarify the problem points I see, concentrating on the use of AERONET measurements. However, I could not help but get the impression that there might be similar problems with the analysis of other measurements as well (but I cannot access them equally readily); the problem being that the analysis did not seem always careful enough and the justification of the obtained results always physically reasonable.
Based on that part of my evaluation, which concentrated on AERONET analysis (and MODIS too, as explained below), it seemed that in your analysis not enough attention was paid to the quality of the measurements and furthermore they were not included and analyzed in a fully consistent way. Moreover, the justifications for your findings were not always clear or convincing. With the related points, which I explain below, I will not try to make an exhaustive list; they are just a few examples. I think the authors need to do a major revision and carefully check and improve the entire manuscript.
The discussion in the page 20 was one of the most unclear and unconvincing ones. Did I follow correctly that you compared Antikythera and Athens Inversion data in the very same day and so that Antikythera measurements were mostly carried earlier in time than those in Athens, but you nevertheless explained that the comparison gives information about aerosol aging during the transport? Is this a correct interpretation? If I compare Level 1.5 of that day, I see that Antikythera and Athens measurements starts from 04:58 UTC, 14:20 UTC, respectively.
After the line #435, the explanations are then simply not possible to follow. SSA is an extrinsic property, so some more explanations would be needed for your discussion regarding dilution during the transport. And even if the dilution would have been in principle a physically reasonable argument, AOD levels in Antikythera on that day (August 7th) are much higher than in Athens? Did you consider it and what was your interpretation, that during the transport AOD increased substantially? I think there was entirely other source of wildfire affecting Antikythera (as I explain below). The unclear reasoning continues in that paragraph: what density would have to do with SSA spectral shape, etc?
I would argue that the whole comparison between the two stations on that day is not meaningful, if the “end station” measurements are made before the “starting station”. However, I would argue that there is even a much stronger point, wildfire event that strongly affected Antikythera and not Athens, which you did not discuss. And this is the reason daily mean AOD_500 in Antikythera and Athens (on August 7th) were 0.53 and 0.38, respectively. This fire was in Mania peninsula and can be seen in the Aqua of the day before (hopefully the link becomes ok, but if not, please do see the Aqua MODIS corrected reflectance from August 6th, for instance from NASA WorldView). https://go.nasa.gov/3SEK9XK
This seems to have been a quite strong and active fire and estimating the prevailing wind directions, “interpolating” from your trajectories, this fire has very likely clearly affected Antikythera measurements in the morning of August 7th. Do you disagree?
Regarding AERONET measurements, you decided to use Level 1.0 Direct Sun product and Level 1.5 Inversion product. One can justifiably say that this should not be done. But at the very least, then the measurements should have been studied more closely (what to include in the analysis and what not). This did not seem to have happened.
There is a set of criteria used to decide whether a measurement in Inversion product will rise from Level 1.5 to Level 2.0 (limit in sky-error, SZA, in AOD at 440nm etc). I would argue that those should have been considered measurement-by-measurement even in the case if Level 1.5 measurements are used (to include some quality assurance). However, since this was not done, some results are reported that might not have been that reasonable to emphasize at all. For instance, in line #366 (“SSA reaches very low values (even below 0.7)”), this particular SSA measurement, you refer to, is from August 12th when during the measurement AOD at 440nm was 0.15 and the sky-error 8.5%. I do not think it is meaningful to report and discuss AERONET SSA (highly uncertain) measurement during this kind of low-AOD conditions?
Line #354: “maximum AOD values between 0.78 and 0.39 at C5”. This was a relatively clean day, while only the two last measurements of the day (in Level 1.0) show a drastic increase, and this is exactly from where you draw your conclusions of maximum AOD values? Do you really trust in these measurements? If you plot AE, you see that it drops very quickly at the very same time (and the coarse mode AOD suddenly rises accordingly). Maybe this was not a very clear smoke day at all (quite low AOD during entire day and then two cloud contaminated AOD measurements in the end of the day)? At least I would suspect that these two measurements are clearly cloud contaminated ones (sudden AOD_coarse (AE) increase (decrease)), this is why AOD increases rapidly. Overall, the mean daily AOD_500 ~ 0.11 would suggest that his was a quite clean day. I was left with an impression that you did not analyze AERONET measurements very carefully.
Also related to this comparison of two stations on August 7th, did you really use consistently Level 1.5 inversion product? It seems to me that your Athens data was based on Level 2.0 and Antikythera on Level 1.5 (if both were from Level 1.5 the spectral difference seems much smaller) on August 7th – is this correct interpretation. But if you did not use the AERONET measurements consistently, why was that?
From time to time the manuscript started “suddenly” with a discussion that was difficult to really link with the discussed just before. Just one example: how the discussion starting at the line #367 is explaining the results introduced just earlier in the paragraph? It was not possible to follow.
Line #396, “High AOD and high AE in morning and high AOD and low AE in evening”. Was it really so? If I look at the diurnal pattern, is see that AOD is drastically lower and quite low in the evening than in the morning. Do you agree? This is just one example of sentence or statement that leaft the impression that the work was not done thoroughly and carefully enough.
Line #441, what is PM there?
Line #489: with the sentence, starting here, you explain why AE is larger in smoke than dust dominated case. However, it does not explain why AE would affect GHI change as you found, which the reader assumed to read about (and what would be interesting and relevant, compared to the pretty obvious explanation you gave, which was not relevant in that context).
Figure 10: perhaps not the best legend titles used in the upper plots (not all the simulations are at UV wavelengths).
Comment: perhaps interesting additional information would have been available if you studied the wavelength dependence in AE (so called spectral curvature effect) or AAE during these episodes.