Dear Phil Stevens,
The revised paper has been reviewed again by two referees. Referee #1 proposes to reject the paper mainly because of insufficient availability and quality of data, while Referee #3 recommends publication with minor revisions. The requested revisions, however, are substantial and concern the lack of NO data needed for the interpretation of the radical chemistry.
I think that both referees make some good points. I share the view of Referee #3 that the paper contains valuable information that is worth to be published. It is of general interest that a significant OH interference of roughly a factor of 2 has been observed in the Indiana University LIF-FAGE instrument in a forest environment by using a chemical modulation technique. This information is relevant because similar types of instruments have reported in the past unexplained high OH concentrations in environments with high isoprene and low NO concentrations. The present work also demonstrates the usefulness of the chemical modulation method that was introduced by Mao et al. (2012). Furthermore, the well characterized data set for HO2*, which is essentially the sum of HO2 and isoprene peroxy radicals, and the measured OH reactivities are also of interest in combination with interference-corrected OH data.
I share the view of Referee #1 and #3 that the use of a mean diurnal NO profile for the intepretation of the complete radical data set is problematic. There are large gaps in the measured time series of OH and NO, which show little temporal overlap between the records. The paper presents all arguments why it is not reasonable to use the averaged diurnal NO profile for OH model calculations which are then compared to the measured OH. First, NOx levels were much higher in the first period of the campaign, when most of the OH data were measured, compared to the second period, when most NO data were recorded. Second, missing NO could not be calculated from the available NO2 measurements, because the photostationary state was strongly perturbed by local NO sources. In such an environment, the lack of measured NO data prevents firm conclusions from comparisons between modelled and measured OH, as the production of OH is dominated by the reaction of HO2+NO. Following the recommendation of Referee #3 and in agreement with comments by Referee #1, I suggest that the authors revise their paper and restrict the model interpretation of OH and HO2* only to those periods when NO and NO2 are simultaneously available. Though this means that only a few days can be used for comparison between model and measurements, the conclusions are expected to be significantly more robust than in the current paper version. This may provide better constraints to answer the important question, whether we are still lacking unknown OH (or HO2) radical sources in current atmospheric chemical mechanisms. When your revise the manuscript, the suggestions by the referees should be taken into account. In addition, I also have a few specific suggestions that should be considered.
Editor's suggestions
• The magnitude of the measured OH interference should be mentioned in the abstract.
• The quoted literature on the chemistry of isoprene is not up-to-date. Since the LIM1 mechanism was published by Peeters et al. (2014), there have been new studies presented by Peeters (2015), Teng et al. (2017), Wennberg (2018), Berndt et al. (2019), Müller et al. (2019) and Novelli et al. (2020). For references and review of the differences in the mechanisms, see the recent paper by Novelli et al. (2020). It should be noted that the LIM1 mechanism and the isoprene mechanism in MCM v3.3.1, which are both used in the present study, are not the same, as they use different rate coefficients for the LIM1 chemistry. This should be clarified on page 9, line 12-13.
• Page 5, line 7. Does the laser power (0.5 - 4.4 mW) apply to a single laser beam, or to overlapping laser beams?
• In Figure 5, a considerable number of interference-corrected OH data points have negative values that are more than 3sigma less than zero. Why? Is it possible that the chemical modulation is overcorrecting the interference or is the precision of the OH data worse than indicated by the error bars? This point is also addressed by Referee #1 who is wondering about the negative OH concentration in the afternoon of July 20 (Figure S1). In Figure S8, this data has apparently been discarded and replaced by estimated values. The figure caption only mentions "Concentrations of OH in the afternoon were estimated due to the poor precision of the data." Please explain the reason and how the estimate was done.
• When the measured OH concentrations from the IRRONIC campaign are compared with different models, I recommend to compare the outcome with the results derived by Novelli et al. (2020) from chamber experiments. She investigated the photooxidation of isoprene by OH in synthetic air at ambient conditions and comparable low NO concentrations. The measured OH is compared with predictions by different chemical mechanisms including MCM v3.3.1 (Figure 5, 6) and LIM1 (Figure S5), which are also used in the present work.
• Figure S8. I agree with Referee #1. Since HO2* contains a well characterized fraction of isoprene peroxy radicals, I suggest to use interference-corrected HO2 concentrations for the OH budget analysis.
• Is there any explanation for the relative high midnight isoprene values (1ppbv)? For comparison, Tan et al. (JGR, 2001) report 10 times lower nocturnal isoprene concentrations in an isoprene emitting forest in summer 1998.
• The paper uses EDT time. When was local noon?
• Page 17, line 25: Typo. It must read "Rohrer".
• Table 1. Add footnote explaining HO2*.

Dear Phil Stevens,
The revised paper has been greatly improved and the comments of the reviewers and the editor have
been adequately answered. Before the paper can be published, some minor changes are needed for
clarification (see attached file).
Best regards,
Andreas Hofzumahaus

Dear Phil Stevens,
I am pleased to accept the revised manuscript for publication in ACP.
Best regards,
Andreas Hofzumahaus