The study by Bhandari et al. builds on the group’s previous and companion work measuring and characterizing temporal patterns and source apportionment of non-refractory fine particulate matter (NR-PM) in Delhi. They have developed a “time of day” PMF method which improves on the traditional PMF algorithm primarily because it does not rely on the assumption of source chemical profiles being static over time. Instead, the authors use their new method to explore the possibility of diurnal and seasonal source profile changes in typical aerosol mass spec PMF factors such as BBOA, COA, and hydrocarbon-like organic aerosol (HOA). Most of the nuts and bolts of this method is covered in their companion paper, almost to the point that the AMT manuscript is an extended “methods” section of this manuscript. Naturally, the discussion in this manuscript is thus quite reliant on the methods presented in the AMT manuscript. Ideally, this manuscript should be considered for publication after the manuscript presenting the underlying methods has been peer-reviewed. As such, I will preface the remnant of my review with the following: this review of ACP manuscript assumes that the underlying methods presented in the companion AMT manuscript do not have any major technical issues.

This manuscript focuses on how the authors interpret their results, their advantages in informing source apportionment, and policy implications. Overall, I think this is a very informative approach developed by the authors. The manuscript is very thoroughly prepared and the results are presented clearly. I have a few comments that I tihnk should improve the readability of the manuscript.

A couple of general comments first:

1. Because several different PMF runs are being compared in this manuscript, I understand why the authors decided to tag each PMF run with the SYYTTTT code. But is the “YY” component really needed? As far as I can tell, all measurements occurred in 2017. I suggest removing the YY component and hyphenating to make the code more readily graspable i.e., S-TT-TT.

2. I think the authors need to be cautious in claiming that all differences between their “time of day” PMF approach and the conventional PMF approach can be attributed SOLELY to potential changes in source chemical signatures. After all, PMF is blind to chemistry and it simply tries to find a local minima in Q/Qexp for whatever dataset is provided to it. Time-varying instrumental uncertainties, relative ionization efficiencies, etc., can also play a role in how this solution is found. In fact, for the same input dataset and same number of factors specified, multiple slightly different PMF solutions can be derived based on how the matrix is rotated (“f_peak”), or how the first numerical step is taken by the algorithm (“seed”). Attributing variations between PMF runs to source chemical signatures without exploring (or at least acknowledging) these other possible sources of variations seems risky.

Specific comments:

L104: the term “NR-PM2.5” is used here for the first time. Please define. Also, if not already specified, please specify that the DAS ACSM measured NR-PM1, not NR-PM2.5.

L118: “… its ubiquity”. Please clarify what is being referred to here. Ubiquity of cooking sources? Or ubiquity of detectable COA? Its a hair-split, but good to clarify.

L135: “IGOR PET” – Igor is the software, which isn’t an acronym, so shouldn’t be capitalized. And PET is a super-acronym (the “P” stands for an acronym itself), which should be defined. Also, since PET is first mentioned here, the Ulbrich 2009 citation should be included here.

L164: at the end of “… course of the day”, I suggested adding a clause “especially OOA factors”, because the phenomena described later (reaction chemistry, gas-particle partitioning, changing meteorology, etc.) would affect the chemical signature of OOA factors more than primary factors.

L240: If I am understanding Figure 1 correctly, the sentence “clearly, POA concentrations exhibit larger variability than OOA concentrations” should end with the phrase “in winter”. Monsoon POA and OOA both seem to be similarly variable (0 to 50 ug/m3, from the looks of the y-scale).

Figure 1: I suggest a few tweaks to this plot – a) add Spearman R to each panel comparing “time of day” to “seasonal” PMF; b) adding “WINTER” and “MONSOON” to the panel headers, and adding a dividing line between (a,b) and (c,d) will make the seasonal distinction jump out better.

L291 – 304: comparisons are made to other studies that made measurements at 3m height near an arterial road, but it would be help to also include the DAS study parameters here (I know 4^th floor of the building, but just to compare numbers, please include sampling inlet height off of the ground, and distance to nearest major roadway).

L334: the nomenclature “winter-to-monsoon” sounds like a ratio of winter to monsoon levels. Instead, I suggest something like “monsoon-adjusted winter BBOA”.

L345: please verify precip data shown in Fig S1. Seems like an abrupt step change at 6 am. If there is an explanation for this, please include in Fig S1 caption.

Figs 2 and 3: I don’t see why these two cannot be combined into one figure, similar to Figure 1. Irrespective of whether authors choose to combine Figs 2 and 3, please consider adding a text label to the figures clearly specifying the season.

L385: The studies cited here for “Asian cooking” references were not conducted on Indian cooking styles. Some of them were on Chinese cuisine emissions (or measurements conducted in China). I would be careful about bucketing Chinese and Indian cooking together as “Asian” here, without at least acknowledging that these two are vastly different and there just isn’t literature on the latter as there is on the former.

L390 and several other places: the m/z 55:57 signal ratio is used often to infer cooking influence, but it is not clear what is a reference value for this ratio to compare against while making this inference. For example, L416: why is the 55:57 value of 1.1 “low”? This is where those reference values from Robinson et al. and Mohr et al. would be helpful. I suggest that when the Robinson 2018 study is cited first here, also include the 55:57 ratio from that study, so as to “set the stage” for upcoming discussion on 55:57 ratios. Also, Claudia Mohr’s 2012 study (https://acp.copernicus.org/articles/12/1649/2012/) was an earlier one that used the 55:57 ratio to identify cooking emissions. It should be cited here along with Robinson, and example values of 55:57 ratios from both should be mentioned for reference.

L653: “with larger ratios of contributions of m/z 55 to m/z 57” please include a number or range here to quantify what “larger” means.

Reviewer Comments on manuscript titled “Contributions of primary sources to submicron organic aerosols in Delhi, India” by Bhandari et al. submitted to ACP Discussions

This work by Bhandari et al. utilizes their companion study (Bhandari et al., 2022 AMT Discussions) that proposes a  time-resolved method for source apportionment using the underlying approach of positive matrix factorization (PMF), also referred as “time-of-day PMF” and demonstrated statistical improvements over the traditional PMF (uncertainty owing to static mass spectral profiles). Delhi, India is one of most polluted megacities on Earth, with inarguably one of the highest Primary Organic Aerosol (POA) concentrations anywhere. This study critically focusses on Delhi to quantify the contributions of different POA components: BBOA (biomass burning), COA (cooking), and hydrocarbon like organic aerosol (HOA, from anthropogenic fossil fuel combustion) by applying “time-of-day PMF” (diurnal profiles as a result) on two seasons (winter and monsoon 2017) using OA measurements from an Aerosol Chemical Speciation Monitor (ACSM). They utilize the EPA PMF tool with the underlying Multilinear Engine (ME-2) as the PMF solver, and conduct detailed uncertainty analysis for statistical validation of their results. Assuming that the companion AMT manuscript will eventually go through final publication without any technical modifications in its method (given it’s the bulk of the “Method” section of this paper as well), I think this work is very significant for better design policies to mitigate pollution in Delhi or National Capital Region (NCR, in vicinity of Delhi) caused by relevant primary sources of organic aerosols as analyzed in detail with the time-resolved component in this study.

I will suggest publication of this work, after the following comments are addressed by the authors:

General comments:

• Will suggest the authors to include updated citation of the finally published companion Bhandari et al., 2022 AMT paper as its it’s the bulk of the underlying principle/Method of this section. That would be ideal before publishing this work. Any modifications/edits to the companion AMT paper on its final publication should be accommodated in current work (ideally before a final draft is accepted, if possible or as an addendum later).
• The current study needs to further decipher the identification of different markers for the presence of cooking organic aerosol (COA) based on the variability in cooking fuels or technology in Indian context (more regulated liquified petroleum gas connections vs wood or residual burning using open stoves- also presents a pragmatic contrast within different COA sources in Delhi), which is currently missing in the current discussion (Lines 385-395). It’s understandable if it is beyond the scope of current study, but should be in that case, mentioned as a limitation of the current study that needs further exploration.

Specific (minor) comments:

Line 70: Please refer ‘… and co-workers’ as ‘… et al. (YEAR)’ consistent with other instances in the manuscript text (rephrase Lines 70-72 accordingly, edit other such instances in the manuscript accordingly).

Lines 96-97: Add space between ‘µg’ and ‘m^-3’ (applies to similar other instances in the manuscript text).

Lines 99-101: Same point about citation being consistent: Rooney et (2019) and Rooney (2019) in consecutive lines, although it’s the same reference. Try to keep citation of a paper consistent throughout the manuscript.

Lines 104 and 133-134: Please clarify the full form of any abbreviation at its first use in the manuscript, i.e. non-refractory (NR) and ACSM in this case at Line 104 instead of Lines 133-134. Similarly, apply for any similar instances in the manuscript.

Line 185: Rephrase “have been described previously” to “have been described in previous literature”.

Lines 291-304 (Section 3.1.1): More discussions/hypothesis and/or details are needed on why there exists inconsistency in HOA average contributions to OA compared with previous studies? For instance , is difference in meteorology in different years between different studies a factor as well besides the difference in profile of emission sources at site(s) between different studies?

Line 351: “relatively high volatility of BBOA”: oxidized BBOA is low-volatility OA and more explanation needed here on why monsoons won’t exhibit much of low-volatility oxidized BBOA?

Figs 2 and 3: Add “Monsoon 2017” and “Winter 2017” labels respectively to Figs. 2 and 3. (Applies to other figures also)

Section 3.1.3: more clarity is needed on the rationale or necessity of doing “Winter-to-Monsoon” and “Monsoon-to-Winter” weighing on diurnal PMF

 Figures 4,7 and 10:  Author(s) should consider combining/rearranging parts of Figures 4,7 and 10 based on if they are for Winter 2017 and Monsoon 2017. Also discussions pertianing to these figures in Section 3 can also be further synthesized to improve readability.