Thanks for the opportunity for this review.

General comments 

I have read with interest the manuscript entitled “A new accurate retrieval of bromine monoxide inside minor volcanic plumes from Sentinel-5 Precursor/TROPOMI” submitted to the EGUsphere/AMT by S. Warnach and collaborators. The study presents a thorough empirical and quantitative investigation of the choice of different retrieval parameters to obtain column densities of BrO in volcanic plumes, which can be applied for global observations with the TROPOMI satellite instrument. In essence, the study investigates the systematic effects of the choice of different spectral intervals and the interference of HCHO, and, most notably, it presents an empirical background correction scheme using latitude-dependent information on O3 and cloud height and fraction parameters derived from independent algorithms applied to TROPOMI observations. Most of these ideas are present in previous studies, and for this reason, in my opinion the main merit of the study does not lie on its originality, but rather on the meticulous analysis to find optimal retrieval parameters and to quantify the expected uncertainty for results obtained at different regions of the world. 

The background correction scheme seems in a way similar to the DOAS algorithm, in the sense of separating “structured” information from “smooth” information, but it does so in the spatial domain instead of the spectral domain. The rationale behind is that the choice of an imperfect background, here proposed as the average radiance at a latitude band around the globe, leads to a bias that vary in space smoothly in relation to the strong variation caused by a volcanic plume. The further exploration of the co-location of volcanic SO2 leads to an even more accurate representation of the background and retrieval of the volcanic signal. 

GC1) It would be interesting to see a discussion on why the region selected for the background: a band around the equator, which includes many potential sources of volcanic or biogenic interference, does not lead to more noisy results. Is this just an effect of reducing random noise by averaging more pixels? Were the examples presented representative of general conditions? A map of uncorrected SO2 columns in the band used for the background presented in the examples could give a visual representation on how “clean” this background was. 

GC2) Because of the importance of the auxiliary information required for the background correction scheme, namely O3 column densities and cloud parameters, it would be good to provide a brief description of how those parameters were derived. 

GC3) The writing style could be more concise. There is room for improvement in avoiding the repetitive, and a bit confusing introduction of what is going to be presented on each section with reference to what has been presented before. Instead of saying for example “in this section we will do X, using what was done in Sect. Y”, I suggest presenting directly the new step without going back to the previous steps. The introduction to the sections, as done in the manuscript, adds little to readability. Shortening this will improve the flow of the text. 

By presenting concrete suggestion on optimal retrieval parameters for volcanic BrO from TROPOMI, this article can make an important contribution to operational global retrievals on volcanoes, complementing the existing capabilities of TROPOMI to detect SO2 from even weak volcanic plumes. 

Specific comments (SC) 

SC1) The title is a bit awkward. It indicates to present a retrieval of volcanic BrO from a satellite sensor. It should instead refer to an algorithm to retrieve volcanic BrO column densities from satellite sensor data. The authors may judge if the algorithm is only valid to this specific satellite sensor. 

SC2) The abstract could be shortened by reducing the information on the first two paragraphs to concentrate on the contribution of the study. 

SC3) Expand on importance of measuring BrO. How much can O3 be destroyed and by which mechanisms? How the columns of BrO or column ratios of BrO respect to SO2 can help to understand volcanic processes? 

SC4) The description of the DOAS method supported by Eq. 1 misses the essential feature of separation between high- and low-frequency components of the optical depth, that characterizes this method. 

SC5) Explain carefully the motivation to use the entire equatorial Earth-shine band (at all longitudes) for the background “reference” correction. This region includes quite different albedo regions (land, ocean), and many sources of BrO. The motivation of this choice is far from obvious. 

SC6) Provide essentials of the FRESCO algorithm to obtain cloud height and fraction products used for the sensitivity study. 

SC7) Provide essentials of the algorithm used to retrieve the HCHO product used for the sensitivity study. In particular, refer to how BrO interference was treated in such retrieval or if it was neglected? 

SC8) Section 6.1 introduces SO2 measurements, here some of the acronyms are spelled out for the first time. The first two paragraphs of this section could be moved to the Introduction. 

SC9) The spectral evaluation setup for SO2 could be included in Table 1. 

Technical comments (TC, followed by line number) 

TC1) L33- Spell out “sulphur dioxide” before chemical formula on first mention. 

TC2) L35- The payload is not designed to “determine” the composition, i.e., the instrument cannot determine the composition (this is determined by natural processes), but to measure certain properties of the atmosphere. 

TC3) L38- Define acronyms on first mention (GOME, SCIAMACHY). 

TC4) L49- Correct spelling of “measurments”. 

TC5) L60- Use consistent notation for all molecular species, i.e. “O3” instead of “ozone”. 

TC6) L63- What is lower altitudes and higher latitudes? Better to indicate “tropical”, “mid-latitude” etc., or even better to tell percentage of active volcanoes within +-30 deg, to back up this assertion. 

TC7) L81- Add “typical” before “volcanic BrO columns of small eruptions”. 

TC8) L83- Define “VCD” on first mention. 

TC9) L103- Use upper case for the name “Precursor”. 

TC10) L112- Correct spelling of “characteristica”. 

TC11) L113- Better to write “Selection of spectra” instead of “spectra… used” as sub-section title. 

TC12) L125- Use upper case and complete name for “Sentinel 5-Precursor Expert Users Data Hub”. 

TC13) L174- Correct “far off”. 

TC14) L229- Define “dSCD”. 

TC15) L243- Use upper case for “Pacific”. 

TC16) L435- Better to use the noun-phrase “Investigation of systematic…” instead of the continuous verb form “Quantifying the systematic…” in the title of a sub-section. 

TC17) L472. Similar than previous comment. 

TC18) L577- Does the value indicated as typical for TROPOMI correspond to one or four standard deviations of SO2? And is this the standard deviation of the residual? 

Figures 

Fig. 1) Indicate which references cross sections were used before convolution (reference to authors), preferably as a legend or caption to the figure. 

Fig. 2) Add labels to y-axes. 

Fig. 3) Reference to Fig. 4 and Fig.5 in the figure description is not appropriate because one needs the three figures to understand the meaning. Better to explain briefly the reason of the two regions, e.g., with relation to the study of the cloud and stratospheric O3 interferences. 

Fig. 4 (and all following maps) Add units of “deg” to all axes showing lat and lon. 

Fig. 5) The O3 background varies very drastically with latitude. It would be good to discuss if the reason for this steep gradient can be found in terms of the general features of O3 dynamics (e.g., presenting a plot of typical O3 latitudinal gradients for comparison). 

Fig. 6) This sequence of figures, presented in the sensitivity studies, is difficult to understand. It would help to add text in the caption to guide the reader towards a conclusion. The pattern cannot be understood at first glance. What complicates matters is that the scales for VCD are different. 

Tables 

Table 1) “Shift and squeeze” and “ISFR” should be classified as instrumental corrections and not “pseudo absorbers”. Define “ISFR” and its parameters in the table’s description or as a footnote. 

Table 3) It could be limited to include only new information not presented already in Table 1. 

Warnach et al. presented. A new scheme to retrieve BrO with improved precision and accuracy. This is made possible thanks to a combination of improved DOAS settings and sophisticated bias correction. This is an important study in view of the systematic investigation of volcanic BrO plumes measured by TROPOMI. The paper is well structured and a pleasure to read. The approach is scientifically sound. I recommend publication in AMT after addressing my (minor) comments below.

Section 2

-In the section describing TROPOMI, there are sub-sections 2.1.1 and 2.1.2 which are very short. The author could consider removing the sub-sections structure (but keep the text in section 2.1).

-Section 2.1.2. Here FRESCO and MICRU are mentioned as available cloud products, but the author does not consider OCRA/ROCINN which is the S5P operational cloud product. Is there a reason for not considering OCRA/ROCINN. For the O3 VCD, why was the operational total ozone column product not used?

-Equation 1 does not include the broad-band terms. Please add those to the equation.

-Equation 3 is incorrect as it should in principle imply the cosine of the angles (SZA, VZA). Please update the equation.

-Section 2.3: it is mentioned that the full zonal band (from 20S to20N) is used as reference region. However, the author is not really justifying its choice. This is a very large region which covers many volcanoes (with potential contamination by strong plumes SO2 and BrO, and to some extend HCHO). Please clarify why it is an advantage to use such an extended reference region compared to a smaller region (e.g., in the equatorial Pacific).

Section 4

-Section 4.1. on the effect of clouds. From Fig4, it is not clear to me whether the observed effect of clouds is an artefact or not. For large CF and elevated clouds, the BrO VCDs are lower which is compatible with a possible cloud shielding of the tropospheric BrO column. I am not sure why this should be corrected.

-Fig5: it is stated that:” the sign of the relation is inverted and high cloud fraction results in elevated BrO VCDs” but without the information on cloud height it is difficult to know whether there is a significant difference as for the effect of clouds, compared to the tropics. Please add the cloud height map and expand the discussion.

From FigA1, it seems that the 323-360nm range is the one with the strongest cloud impact. I find it hard to justify that this range is the one retained for the final SCD retrievals.

Section 4.2.

-for low CF, the retrieved cloud height is uncertain, not to say ill-defined. How do you manage this in your correction?

-Regarding the lat-ozone correction, it is mentioned that ‘the latitude band can be adjusted for each volcanic plume’ (line 351). I guess this is a future implementation wish or is this really what is implemented. Also, it is not clear how frequent the correction parameters are updated. Is this done separately for each calendar day? Please clarify.

-Figure 11 is an interesting plot but it is not clear from where the model factor of 6.22e12 is coming. Could you elaborate?  Also, in the evaluation of the systematic uncertainty of the retrievals (summarized in Table4), the authors assume that a good estimate of these systematic influences can be obtained from the SCD std over many pixels. The validity of this approach is not clear because random uncertainties are still present and contribute to the std (these are not reducing as the square root of the number of pixels, as they do for the estimated mean). Also, systematic uncertainties like the one related to the BrO cross-section uncertainty is not accounted for.

Typos and suggestions

-p4, l111: ultra-violett -> ultra-violet

-p4, l112: characteristica -> characteristics

-Single sub-section like 3.1 is not necessary. Instead I would write just after l225: “To illustrate the fit performance, the global BrO VCD map for 1 October 2018…”

-Fig4c: it would be good to only show the cloud heights above a certain CF threshold, to better appreciate low/high clouds.

-FigA1: it would be good to have the same color bar limits for all 8 subplots, otherwise it is difficult to compare the results.

-line 310: ‘..the for the wavelength..’ -> ‘..the wavelength..’

-line 445: ‘..increasing increasing..’->’..increasing..’

-Table 3 is redundant. I would propose to remove it. Table 1 could better highlight the preferred settings (e.g., with the corresponding text in bold).

-line 532: ‘the latitudinal background correction is applied’ here you mean the ozone-cloud correction, right?