This manuscript is well written. The authors conducted a fair intercomparison of online and offline INP-measuring instruments in the field. Despite the challenging environment at SPL, invaluable outcomes and lessons are reported in a neutral and unbiased manner. Furthermore, the authors include a list of limitations (e.g., deviation in sampling particle sizes etc.) and things to be further explored in this manuscript for more understanding of aerosol-cloud interactions (e.g., a need for online/direct deposition ice nucleation measurements), which are important messages to the INP research community. This reviewer agrees that more research is necessary to predict and explain the temporal variation of biological INPs (perhaps in a predominantly biogenic environment). While the authors found a predominant contribution of mineral and/or other inorganic particles to INP abundance in this study, they also note the need for in situ mixed INP detection and characterizations, especially for Soil & BBA INPs, which is important. The study topic is relevant to the journal scope of AMT. This reviewer supports the publication of this paper in AMT after the authors address several questions below.

===

Questions

[1] Figure 7: This reviewer wonders if using 3-hr INP median or log-average changes any conclusions of this intercomparison study. The ratio in Fig. 7 is computed by using time averages, which is reasonable. But, since the reported N_INP spans a log range at a majority of freezing temperatures examined in this study, the average can be biased by high N_INP values at the given temperature, such as the ones from FRIDGE-CS and CSU-IS. Perhaps, using the median may overall result in better agreement for NC State(F), NC State(I), and CSU-CFDC? The same average vs. median argument applies Figs. 8 & 9.

[2] Figure 5: This reviewer wonders why NC State CS(F) shows a lower detection of N_INP (~6 x 10^-3 L^-1) than NC State CS(I) (~10^-1 L^-1). The sampling air flow rate seems similar for these two methods as described in Sect. 2.2.2. The sampling interval was shorter for impinger sampling? Or it may be due to the difference in collected particle sizes (L836-839; L846-848; L855-858)? This reviewer is aware of a general statement in L865-870.

[3] P31L649-655: Low AE (<1) seen in 9/14-16 in Fig. 3b may be due to the predominance of large dust seen in Fig. 4? The authors also report that the submicron particles dominated during the study period (L-637-638). The effective aerosol scattering efficiency from SPL during this intercomparison campaign can be similar to what is reported in Testa et al. (2021)?

Refs.

Russell. P. B. et al., ACP, 10, 1155-1169, https://doi.org/10.5194/acp-10-1155-2010, 2010.

Testa, B. et al. JGR-A, 126, e2021JD035186. https://doi.org/10.1029/2021JD035186, 2021.

[4] P49L971-973: Can the authors clarify this part?

Comments

P37L749-750: This is good. Comparability of impinger and filter-based methods shown in this work implies that ambient particles collected on filters are well-scrubbed in liquid suspension for freezing tests on NC State CS, resulting in comparable N_INP to that from directly suspended impinger samples, for this field study at least.

P44L885-887: This recaps that the link between aerosol chemical composition and INP is not straightforward and underscores the importance of ice residual composition data.

P64L1249-1255: This reviewer agrees. The ultimate future INP instrument intercomparison may be performed on the aircraft platform in cirrus and/or pyrocumulonimbus cloud regimes with collocated aerosol instruments suggested by Burrows et al. onboard then.

Attached as a PDF