|1) The trends in Figure 2 and Figure 5 seem to be unrealistic, or at least inconsistent with the argument (Line 358-365) that extraction efficiency increases with increasing OC/TC (and thus, SSA). Specifically, in panel (b), how do the authors explain the sharp drop in babs,OA/babs,sol to values close to zero at high OC/TC and SSA? One would expect babs,OA/babs,sol to approach a constant value (due only to particle size effects) with increasing extraction efficiency, so one should expect an asymptotic behavior of babs,OA/babs,sol at large OC/TC and SSA values, opposite to what’s observed. It seems to me that the observed trends could be partially an artifact of the way babs,OA is calculated as the difference between the measured babs,tot and a calculated babs,BC (section 2.2.2). In any case, a discussion of the meaning of the trends observed at large SSA values should be added to section 3.1. |
2) Along the same lines, the data, especially the part of the space that has the sharp change (i.e. SSA > 0.9), should be further explored. For example, the data in Figure 2 could be plotted with the points color-coded by fuel type and with different scales on the y-axis to better zoom in the changes at SSA > 0.9 to see if the variability that is not explained by SSA (i.e. where the changes are very steep) could be explained by fuel type. The same applies to OC/TC.
3) Line 275 – 281: the choice of SSA = 0.7 and 0.825 as a cutoff between low uncertainty and high uncertainty ranges seems arbitrary. What constitutes high uncertainty? Also, in Figure 2, similar size error bars (especially in panel b) seems to exist on both sides of the cutoff lines.
4) The purpose and value of the Mie calculations in section 3.4 are not clear. It reads: previous studies have used scaling factors of 2. This study found a scaling factor of 2. In conclusion, a scaling factor of 2 should not be used because of reasons stated in the previous sections. Then why do the Mie calculations to begin with? The cautionary statement about the non-universality of the scaling factor of 2 should be inserted in section 3.1.
The presentation of the results could be reformulated in a more informative way. For the 3 experiments reported in section 3.2, the authors can compare babs,OA, babs,sol, and babs,Mie and use this comparison to determine the fractional contribution of particle size and extraction efficiency to the discrepancy.
5) Table 4: It would be more informative to present the data in Table 4 as scatter plot of AAE_OA vs AAE_sol, color coded with fuel type and with a 1:1 line.
6) Table 5: with geometric mean diameter of 397 nm, the size distribution is expected to extend beyond the SMPS measurement window. It is not clear how this issue was addressed.