|This work compiles already published results from 16 sites located in six European cities. Within this huge task, the effort is to identify relationships between meteo variables, gas phase composition and aerosol organic content with key NPF variables such as NPF frequency (the authors name it NPF probability), growth rate and formation rate. Several findings within this work justify publication; two most striking is the nonlinear relation of temperature with NPF probability and the fact that increased solar irradiance reduces the probability for NPF. |
Some issues require attention though.
Starting from most important to least severe
The authors have chosen to use only Ia events and as a result the probability shown in Table 2 is several factors smaller than those reported in literature for the same sites. However, from a brief search, the difference can be up to a factor of six. It is well understood why the authors made such a choice as formation and growth rates can be calculated reliably from these type of events only. How does this choice reflect to the results shown? I strongly support that for one site (e.g. Hyytiälä) an intercomparison is carried out to indicate to the reader the tentative differences. My recommendation is to do so only for NPF probability. This is critical as most studies in the end classify NPF as events, undefined and non-events lumping Ia, Ib and II classes into one. The authors should add to the caption of Table 1 the fact that only Ia events are considered.
How coarse is the time resolution of OC and sulfate measurements examined in this study? The first impression for the former is that they are derived from the thermal optical method. For the latter AMS is mentioned. However, AMS is typically used for short-term campaigns and this is a multi-year study. Do the authors mean ACSM? If the resolution of these measurements is coarser than 1 h, are they reliable to be used in NPF studies? The authors should clarify the time resolution of these measurements, both OC and SO4, in the manuscript and discuss any complications. If the authors indeed use AMS measurements what fraction of the time period discussed do they cover? Also, it would be worthwhile to mention the publications that refer to these measurements.
In each pair of variables (e.g. NPF probability and RH; growth rate and temperature) presented in this work, there seems to be a norm and one (or more) site that is an exception. Since the authors cannot fully explain why (and this is perfectly understood), it is worthwhile to mention that in the abstract or the conclusions or in both.
There is little relationship between RH and CS at most sites. Is this because CS was based on dried measurements and was not corrected for hygroscopic growth? This would be understood since chemical composition was lacking on most sites. Please discuss if CS was corrected for hygroscopic growth and how that affects the results presented.
ANOVA is only valid for normally distributed populations. Have the authors tested for normality? The F-test is typically used. How did the authors treat skewed distributions? Please discuss.
In the supplement, several relationships are clearly non-linear, such as the temperature-NPF probability for a few sites, but the authors insist to use a linear fit (probably for consistency). May I ask the authors to note on the supplemental graphs in which cases linearity is not followed. The authors are better aware of the statistical significance of the related graphs than the reader is. In the case of Denmark Rural (S2b), it is not evident whether the deviation from linearity is statistically significant or not.
Please define in the methods section what is weak, strong and very strong correlation in this work. It will assist the reader further.
The effect of SO2 on NPF that the authors are discussing in Section 3.2.1 has been presented before. Please check the references below. These works relate particle acidity to NPF. Experimentally it has also been verified at the site named GRERU in this work.
Line 247. “The remaining data” is better use of English than the “data left”
If the authors prefer the term NPF probability it is fine. But please use it throughout the manuscript. The caption in Table 2 is a bit confusing.
Jung, J., P. J. Adams, and S. N. Pandis (2006), Simulating the size distribution and chemical composition of ultrafine particles during nucleation events, Atmos. Environ., 40, 2248–2259, doi:10.1016/j.atmosenv.2005. 09.082.
Jung, J. G., S. N. Pandis, and P. J. Adams (2008), Evaluation of nucleation theories in a sulfur-rich environment, Aerosol Sci. Technol., 42, 495–504, doi:10.1080/02786820802187085.