Interactive comment on “Aerosol direct radiative forcing during Sahara dust intrusions in the central Mediterranean”

1. Page 22540, Line 11: Replace “to intialize radiative transfer simulations : : :” by “to perform radiative transfer simulations : : :”. Done 2. Page 22540, Line 13: The range of values of AOD “0.2-0.7” has an upper limit (0.7) which seems to be somewhat low for dust outbreaks in the Mediterranean according to the existing literature. This is likely due to the use of AERONET AOD data. During dust events, the ground sun photometers become saturated so that larger AOD values are missing from time series. On the contrary, corresponding satellite-based estimates (e.g. MODIS-based) report quite larger values. Thus, the reported here AOD, but also DRE values are affected and bounded. Even larger values are possible. As it is well know uncertainties of AERONET AODs are smaller than uncertainties associated to MODIS AODs. In addition, several studies have recently revealed that MODIS AODs are affected by cloud contamination (e.g. Gupta and Christoher, Atm. Phys. 8, 3311-3324, 2008; Schaap Atm.

surface net flux. Why not plot the net infrared flux separately? Given that net infrared flux at the surface is small, about 50 Wm-2 , the uncertainties in the infrared flux are masked by adding them to the larger solar component.
Please note that Fig. 10a shows the scatter plot of simulated net-fluxes values versus net fluxes measured by a net radiation trasducer that only provides net flux values (solar+IR). In fact, on page 22547, lines 17-21 of the manuscript it is written: ". . .A net radiation transducer (p056 RADNT, by SIAP+MICROS S.r.l., Italy) characterized by 1.5% accuracy, is routinely used at the ISAC-CNR Department of Lecce (www.basesperimentale.le.isac.cnr.it) to monitor broadband net-fluxes for the 0.3-30 µm spectral range, with two hemispheric (up and down) sensors. The net radiation transducer is located few hundred meters away from the AERONET sun/sky radiometer site. . .." On the contrary Fig. 10b shows the scatter plot of model net fluxes at solar wavelengths. Both data are based on the same aerosol properties but, have been retrieved from different radiative transfer models.
5. The computation of aerosol radiative effects, and also of the radiation fluxes themselves, is done using a two-stream radiative transfer model. Probably, this is not the most adequate tool for aerosol radiative effect studies.
We believe that the results plotted in Fig. 10a and 10b demonstrate the appropriateness of input and output data of the used radiative transfer model, even if more sophisticated model are available that probably we intend to use in the near future.
6. The applied methodology to identify dust events during the period 2003-2006 is not explained. Although it is not the primary subject of this study, it should be done. Moreover, the applied methodology may be problematic, since it results in aerosol Angstrom values as high as 1.5 (e.g. Fig. 5b) which are certainly not indicative of coarse dust aerosols. Probable problems with the methodology could affect the magnitude of com-C9209 puted aerosol AOD and DREs.
We believe that the contribution of anthropogenic particles of local origin and/or longrange transported are responsible for Angstrom values as high as 1.5 even during dust events. This mainly occurs when air masses from different source regions are advected over the monitoring site (e. g. Pavese et al., 2009). An Avantes radiometer was used in Pavese et al. (2009) to monitor a dust event over Lecce. In addition, as we have outlined in section 3 (page 22545, lines 22-29): "7-day analytical back trajectories by NASA GSFC (http:// www.aeronet.gfsc.nasa.gov/), true-color satellite images by the MODerate Resolution Imaging Spectroradiometer -MODIS (http:// modis.gsfc.nasa.gov/; King et al., 1992), AERONET products as the Angstrom coefficient (Å) and the fine mode fraction, and polarization-sensitive lidar measurements have all been used to detect Sahara dust intrusion events over the Mediterranean and in particular at the monitoring site (e.g. Tafuro et al., 2006;Tafuro et al., 2008). Hence, dusty days have been selected in accordance with different dust event markers, as it is outlined in section 3a." 7. Section 3 seems to be unreasonably placed before section 4. It refers to a unique dust event in 22 June 2006, and it is less important than sect. 4, which refers to the total number of identified dust events. The importance of sect. 3 probably consists in explaining the methodology applied on every event, but if this is the case, then sect. 3 should be renamed pointing to the Methodology. On the contrary, the use of term "methodology" in the name of sect. 4 should be avoided. highlight the methodology applied to infer dust events ." 8. The most interesting contribution of the paper is the separation between natural and anthropogenic aerosol optical depth (AOD), and mainly of aerosol DREs. Formulas are derived and given, which are claimed to be representative for the Mediterranean. However, the possibility of the more generalized application of these formulas to other sites as well, should be assessed and discussed.
The following sentence has been added at the end last paragraph: "Last but not least, we believe that the used methodology can generally be applied to all AERONET sites to evaluate the DRE contribution by fine and coarse mode particles." Other specific comments Abstract 1. Page 22540, Line 11: Replace "to intialize radiative transfer simulations : : :" by "to perform radiative transfer simulations : : :". Done 2. Page 22540, Line 13: The range of values of AOD "0.2-0.7" has an upper limit (0.7) which seems to be somewhat low for dust outbreaks in the Mediterranean according to the existing literature. This is likely due to the use of AERONET AOD data. During dust events, the ground sun photometers become saturated so that larger AOD values are missing from time series. On the contrary, corresponding satellite-based estimates (e.g. MODIS-based) report quite larger values. Thus, the reported here AOD, but also DRE values are affected and bounded. Even larger values are possible.
As it is well know uncertainties of AERONET AODs are smaller than uncertainties associated to MODIS AODs. We believe that the results plotted in Fig. 10a and 10b demonstrate the appropriateness of input and output data of the used radiative transfer model, even if more sophisticated model are available.
2. Page 22543, Lines 5-6: Are the numbers given the centers of the spectral bands? It should be specified. Also, 8 solar and 20 infrared bands seem to be unbalanced in terms of aerosol optical properties and forcings. A larger number of bands in the solar C9212 is more suitable, since aerosol properties and effects are highly variable in the solar, and especially in the ultraviolet-visible wavelengths. The sentence: .." In particular, eight solar (0.35 µm, 0.45 µm, 0.55 µm, 0.65 µm, 1.00 µm, 1.6 µm, 2.2 µm, and 3.0 µm) and twelve infrared (4.25 µm, 5.35 µm, 6.25 µm, 7.35 µm, 8.75 µm, 10.30 µm, 11.75 µm, 13.90 µm, 17.20 µm, 24.30 µm, 37.00 µm, and 80.00 µm) subbands are considered to properly account for the spectral dependence of atmospheric particle properties: the optical properties (extinction, single-scattering albedo, and asymmetry factor) of the atmospheric particles remain constant in each of the 20 subbands." has been replaced in the revised manuscript with: .." In particular, eight solar and twelve infrared subbands are considered to properly account for the spectral dependence of atmospheric particle properties: the optical properties (extinction, single-scattering albedo, and asymmetry factor) of the atmospheric particles remain constant in each of the 20 subbands. We intend to use in the future a new model with a larger number of solar and IR spectral bands.
Please note that optical proprieties are set constant in each subband: i.e. does not vary within a subbands.
4. Page 22543, Lines 16-17: ": : : Sahara dust intrusion : : : to 2006 year": how these events have been identified over the study period? It should be specified. What are the C9213 criteria that have been applied in order to derive the 26 dust events listed in Table 1?
On page 22544, lines 22-29 it is written: .." 7-day analytical back trajectories by NASA GSFC (http:// www.aeronet.gfsc.nasa.gov/), true-color satellite images by the MODerate Resolution Imaging Spectroradiometer -MODIS (http:// modis.gsfc.nasa.gov/; King et al., 1992), AERONET products as the Angstrom coefficient (Å) and the fine mode fraction, and polarization-sensitive lidar measurements have all been used to detect Sahara dust intrusion events over the Mediterranean and in particular at the monitoring site (e.g. Tafuro et al., 2006;Tafuro et al., 2008). Hence, dusty days have been selected in accordance with different dust event markers, as it is better outlined in section 3a." 5. Page 22543, Lines 28-29: ": : : are averaged : : : spectral range": How are they averaged? Why to average over 0.3-0.7µm? It is reported above that there are 4 solar subbands around 0.35, 0.45, 0.55 and 0.65 µm. This is inconsistent with ths band (0.3-0.7) reported here. Do you assume constant surface albedo values over the entire solar range of wavelengths? 0.35 -0.65 µm represent centers of spectral subbands as it is specified in the revised manuscript. In addition, we have decided to assume constant surface albedo values from 0.3 to 0.7 µm.
The surface albedo is also constant from 0.7 to 5 µm. Done 2. Page 22545, sect. 3.1: The introduction of this section here seems unreasonable or at least not justified. The aim of the paper is to evaluate aerosol DREs at Lecce over the period [2003][2004][2005][2006]. Why focus on a specific dust event and study it separately?
The following sentence has been added at the beginning of sect. 3: The dust outbreak of June 22, 2006 is analyzed in detail in this section in order to highlight the methodology applied to infer dust events and to calculate DREs by all and anthropogenic particles. 4. Page 22547, Lines 1-2: "0.87 : : : over Lecce. : : :": The coarse mode in the bimodal distribution could be also attributed (at least to some extent) to maritime sea-salt aerosols (see e.g. Fotiadi et al., 2006). This is also supported by the back-trajectories (Fig. 1).
The suggested sentence has been added in the revised manuscript.
5. Page 22547, Lines 13-15: The results of Table 2 are discussed later on. They should be discussed here.

Done
Also, how does one explain that the surface DRE decreases in magnitude from 15:31 to 16:27, while it increases at TOA?

C9215
The following sentence has been added: It is also worth noting that the surface DRE decreases in magnitude from 15:31 to 16:27 UTC, while it increases at the TOA. The larger contribution of fine mode particles and the smaller k value found at 16:27 UTC contribute to this last result.
6. Page 22547, Lines 22-25: Comparison-validation for two points only is a problem.
The word "demonstrate" has been replaced with "support" 7. Page 22548, Line 6: It would be interesting to examine the role of the IR DRE during night.
Thanks for the suggestion, but we believe that it is not within the main paper's objectives and it will not be discussed to not lengthen the paper. Yes. Refractive index also depends on size, being the size a parameter that can be used to characterizes particles of different type: e.g. natural and anthropogenic.

C9216
The following sentence has been added in the revised manuscript: " In fact, analytical back trajectories indicate the origin area of the air masses advected at the monitoring site. True-color MODIS images provide some indication of the geographical extension of the dust event. AERONET products such as the Angstrom coefficient and the fine mode fraction can allow inferring the presence of dust and polarization-sensitive lidar measurements provide some indication of the vertical displacement of the dust plume." 4. Page 22550, Line 8: "black line in Fig. 5a : : :": the curves cannot be distinguished between themselves. Symbols or different colors could be used together with lines. Done 5. Page 22550, Line 14: ": : : However, along with : : : aerosol burden": To what extent? It can be computed.
The following sentence has been added in the revised manuscript: " In fact, the fine mode fraction η, i.e. the ratio between the fine-mode and the total optical depth at 0.55 ïĄ m spans the 0.3?-0.8? range on the selected dusty days." 6. Page 22550, Line 16: "plotted in Fig. 5b.": it would be useful to relate the points in Fig. 5b to the curves of Fig. 5a, at least, the points corresponding to fine aerosols in Fig. 5b (black circles and red rectangles). Done 7. Page 22550, Line 18: ": : : span the 1.5-0.23 range. : : :": dusty days with Angstrom values as high as those shown in Fig. 5b are hard to believe. Probably, this reflects the problem with the selection procedure of dusty days.
High Å values refer to dusty days with high contribution of fine mode particles. As we have clearly stated in the revised manuscript different dust event markers have been used to select each dusty days. C9217 8. Page 22551, Line 1: ": : : of n and k is: <n>=1.48ï C s0.01": values of <n> could be also given separately for fine and coarse aerosols.
No. The complex index of refraction is the same for particles of all sizes in accordance with AERONET inversion assumptions. 9. Page 22551, Line 11: "Instantaneous AODt, SSAt, and gt values : : :": how do these properties compare with those directly given by AERONET?
The sentence: " retrieve the vertical profile of the AOD fraction " has been replaced in the revised manuscript with " retrieve the aerosol vertical profile normalized to the AOD at 0.351 µm. Then, the vertical profile of the AOD fraction is used " 11. Page 22552, Line 12: replace "lidar on 24 June 2003 : : :" by "lidar on 24 July 2003 : : :". Done 12. Page 22553, Line 5: According to the two Figures, the differences in the net flux between model and measurements (Fig. 10a) are due to the IR fluxes. Differences as much as 50 W m-2 can be seen.
No. See the answer to comment 4 of "Specific Comments" 13. Page 22553, Line 12: replace "increase : : :" by "decrease in magnitude : : :". Attention should be made to avoid confusion, since the values are negative. Done C9218 14. Page 22553, Line 19: replace ": : : smaller than that : : :" by ": : : larger than that : : :". It is better to talk about magnitude. Done 15. Page 22553, Lines 20-27: "The solar : : : on monitoring time": It is better to remove Figure 12a (it does not add much to the information already given) and incorporate Fig.  12b into Fig. 11. Also, the IR and net DREs are not discussed at all.
16. Page 22554, Lines 1-10: The entire discussion of Fig. 13 (dependence on solar zenith angle, z) has a problem and merits further discussion. It s well known that AOD (and hence DRE) depends on the atmospheric optical mass, i.e. solar zenith angle. However, other factors like the suspended amount of particulate matter and the associated scattering and absorption properties are also responsible for AOD and DREs, so they have to be discussed in order to explain the features shown in Fig. 13. For example, note the changing dependence of AOD and DREs on z in some cases (e.g. solar DREt,TOA in 22 June 2006).
17. Page 22554, Lines 15-16: "The variability range : : : of large dust particles": Here it is claimed that n does not affect significantly AFE, opposite to what happens with DRE. An explanation for this should be given.
19. Page 22554, Lines 20-22: "It is worth noting : : : ToA and surface.": A reason for this should be given. What are the differences between the two studies? 20. Page 22554, Lines 23-29: "Figure 11b,e : : : dust intrusion events": this discussion should be moved to the previous page (22543). Concerning the percentage 47%, what are the cases (conditions) in which the IR aerosol DRE becomes comparable to the solar DRE? It might be worthy to examine and discuss this, before the conclusion in the following (last) sentence. Also, concerning the last sentence of the paragraph, it is C9219 not valid generally, r at least at the same significance level. It should be more specific, taking into account what was noted just before. No. 34-85% represent the contribution of fine-mode particles. On the contrary, 27-65% represent the contribution of anthropogenic particles 3. Page 22558, Lines 8-9: "To a first approximation : : : of aerosol present": This sentence should be rephrased. Done 4. Page 22558, Lines 25-27: "Nevertheless, we believe : : : to Bergamo et al. (2008a).": Why would they not be representative for other sites (affected by local pollution) as well?

C9220
The words "..not highly affected by local pollution" have been deleted.