|2nd review of the paper of the paper “Extreme dust storm over the eastern Mediterranean in September 2015: Satellite, lidar, and surface observations in the Cyprus region” by Mamouri et al., submitted to ACP.|
In general the authors replied to the majority of my comments adequately. However, I find that the discussion about the vertical distribution and the AOD on the 8th of September (the day with the heaviest dust load) is not convincing enough (see major comments for details). Although, I believe that the authors describe indeed an extreme event as indicated by the very low visibility and PM10 concetration during the 8th September, but I am not persuaded based on the observations that the AOD is larger than 5, while it can be 3 (of course the event is still extreme).That said I suggest again publication to ACP once the authors have taken into account the following comments.
5) Page 2, Lines 14-15: Although, there are not observations from the CUT- TEPAK AERONET sun-photometer during the event, there are observations (at least Level 1.5) from other sun-photometers in the region like AgiaMarina_Xyliatou and SEDE_BOKER. It is important to compare MODIS AOD retrievals with AERONET values in order to establish how good there are in the case of large AODs (>2). This is important, especially, from the moment that you are using MODIS AOD over Cyprus in Section 3.1.
We try to compare the MODIS observations with our own lidar observations (section 3.2). We do not include other stations. The dust outbreak was so inhomogeneous, what does it help to include other stations? The uncertainty of MODIS values is clear, about 0.15 times AOT.
REPLY: Indeed, this is the expected uncertainty based on statistical comparisons with AERONET. However, here we have a specific event and as such MODIS AOD retrievals can be from very good to very bad. The comparison of MODIS with the lidar confirms the good performace of the retrieval when the AOD is about 1 or less. This is a useful addition to the paper. On the other hand the 8th September, there are not lidar observations to compare with. However, this appears to be the most important day of the event (at least close to the surface, where observations are presented) with high AOD based on the MODIS images (as the surface is not visible). As I said in my first review it is important to validate the MODIS retrievals for high AOD as the validation against AERONET (e.g. Levy et al., 2013) is mainly for AODs less than 1 and simply we do not know the performance of MODIS AOD for the extreme cases with AOD greater than 1 or 2. Finally, the estimation of the AOD using the surface observations and extrapolated vertically based on the radiosonde is very uncertain regarding the thickness of the layer. Just looking at both the potential temperature and the relative humidity I would say that the thickness is 0.8 km and not 1.5 km used in text, while there is an inversion at ~0.3 km, which could also be the top of the dust layer. Note that the lowest value is in agreement with the lidar observations of the 9th September. On the other hand the value of 0.8 km multiplied with 6000 Mm-1 gives an AOD of 4.8, which is closer to the values reported by MODIS to Figure 6. Briefly the MODIS AOD image of the 8th September should be given and the retrievals should be validated against the AERONET observations for this day. Ideally, this should be done also for the other days of the event 7-11 September (but it is not necessary given the comparison with the lidar, although there is significant time difference between the lidar observations and MODIS AOD, while for AERONET the time difference can be less than 30 min).
12) Page 3, Line 30: “A two-layer structure ... on 8 September,” this is pure speculation. Either present observations or delete it.
As already given above, we now include radiosonde information of 8 September. These data perfectly corroborate our ‘speculation’. So, there was definitely a 1.5 km thick and well-mixed layer (with base at surface on 8 Sep, noon). And by using the visibility-related extinction coefficient of 6000 Mm-1 we end up with an optical depth of 9. All in all, lidar and radiosonde data are in perfect harmony for the entire period from 7 to 11 September. This motivated us to use the radiosonde profiles alone to estimate the layering on 8 September. We find, this is fully justified. So, we do now use the 8 September temperature and humidity profiles to explain the dust layering and that the dust was well mixed from the ground up to 1500 m height (as inidcated by the humidity profiles on 8 Sep) and caused and optical depth of 6-9. All this is now discussed in Section 3.2
REPLY: As I explained in my previous reply to comment 5 I disagree that the lower layer is well mixed and up to 1500 m. This is a crucial point for the AOD estimation during this day.
13) Page 4, Lines 5-7: This is just a speculation, as you do not have any information about the vertical distribution of the dust. The authors should underline more this fact using a stronger word than assumption. The fact that the Troodos Mountains were not covered by dust means that possibly the dust layer was below 2 km, but it could be just 500 m (or lower) like the 9th September (Figure 5). On the other hand, Figure 6 shows through the backscatter coefficient that the layer was not homogeneous with the higher values at surface during the 9th September, so this could be the case for the 8th. At this point MODIS maps can give a hint about the area with the high AOD (even saturated at 5), while also the AOD observations from MODIS/Terra could be useful to check the temporal variability. Certainly, this result although plausible can not be one of the main conclusions of the paper as it is based on assumptions which can not be verified. Instead the authors could use the MODIS AOD.
As mentioned, we now include radiosonde information of 8 September….. These data indicate a well-mixed 1.5 km thick lower dust layer. And with surface extinction coefficients of 4000-6000 Mm-1 we end up with AOT of 6-9. And this also in agreement with MODIS. MODIS stops when the AOT is largere than 5. And there are many MODIS pixels (areas) with higher AOT than 5, as discussed in section 3.2. MODIS does not provide data higher than 5. We use MODIS TERRA and AQUA data and do not see strong temporal variability…
REPLY: Similarly to my reply to comments 5 and 12, I disagree about the thickness of the lower layer and the fact that it is well mixed. In any case I do not think that it is difficult for you to add one more figure to the paper with the MODIS AOD of the 8th September, especially from the moment that you have the data already. It is just 10 min work to create it.
1)Page 1, Lines 20-22 and Page 2, Line 1: These are your results, so I do not think that it appropriate to start the paper with them, so please delete them. If it's not the case provide the references.
2)Page 2, Lines 6-7: “However … decade.” Where this statement comes from?
3)Page 4, Line 5: “On 8 … exceeded.” I think it is better to move this sentence to Section 3.
4)Page 4, Line 12: “+/-50”. I think the % is missing.
5)Page 4, Line 31 and Page 5, Lines 1-3: “During … neglected.” These are results and are better suited for the Section 3.
6)Page 5, Line 20: Clearly, I do not see what the HYSPLIT model adds on the paper, especially from the moment that MODIS images show the origin and evolution of the dust plume and the starting point of the backwards trajectory is not Limassol.
7)Page 6, Line 1-2: Why you do not present radiosondes also for the other dates of the event 7-11 September? This will further corroborate the connection of the layering seen by the lidar and the radiosondes.
8)Page 6, Line 8: Replace “exceeded already” by “reach”.
9)Page 6, Line 18: AOT of what? The lidar? You have also the time series of MODIS in Figure 6.
10)Page 6, Line 23: How the values of 8-10 km are estimated?
11)Page 7, Lines 2-9: In the model deficiences you are referring to cloud convenction. However, in MODIS images there are not clouds in the source region.
12)Page 7, Line 25: Why there is not comparison of the lidar and MODIS the 7th September?
13)Page 7, Lines 29-30: Once again, why you do not show the radiosondes?
14)Page 7, Line 32: How the values of AOD for anthropogenic and marine aerosols are estimated?
15)Page 11, Lines 17-18: “The highlight … observations.” Sorry to disappoint you, but the most important day of the event the 8th September, there are not lidar observations. So, this sentence is not justified, please delete it.
16)Page 11, Line 18: According to your reply to my comment 15, I do not think appropriate to call your lidar state-of-the-art. Please delete this characterization.
17)Page 11, Line 22: “constellations” you mean “conditions”?