This manuscript presents a comprehensive study of NOx emissions and ozone chemistry in east Texas in the warm season of 2019. It contains CAMx chemical transport model (CTM) simulations, comparisons of TROPOMI NO2 retrievals with different retrieval versions and averaging kernel/AMF applications, NOx emission estimates using exponentially modified Gaussian (EMG) function method and the flux divergence method, and satellite vs. model HCHO:NO2 ratio comparison. I have some concerns about the NOx:NO2 ratio and the NO2 effective lifetime used in the EMG- and flux divergence-based NOx emission estimates. The EMG method has been out there for over a decade, and the original assumptions about NOx:NO2 and NO2 effective lifetime seem too simplistic now, especially in this study where a full CTM simulation concurrent with satellite observations is available. The NO2 effective lifetime seems curiously low, making the EMG-based NO2 lifetime all over the place from the literature (from 6.7 h in Valin et al. 2013 to ~1 h in this and recent studies). Given the seesaw relationship between effective lifetime and emission rate, this also gives significant leverage to the emission estimates. I hope there could be some discussion and preferably quantitative assessment of the effective lifetime using the modeled NO2 reaction/removal rates. The other comments are mostly minor and also listed below.

Page 1, lines 27-28: satellite measures column density, not emission. de Foy et al. (2015, 10.1016/j.atmosenv.2015.05.056) showed better agreement using OMI NO2 data, and satellite NO2 have been widely used to quantify point emission rates. Given OMI has been proven to be able to do that in many other studies, it’s hard for me to believe TROPOMI with much smaller pixels cannot capture the plume. Even at the plume head the optical depth of NO2 is still far from optical saturation. Could non-equilibrium NO2/NOx ratio near the stack be a reason?

Page 2, line 7: the photolysis rate constant for NO2 is around 1e-2 s^-1 for a wide range of SZA, indicating an NO2 “photochemical lifetime” of only ~100 s. I think what was meant here was “chemical lifetime” that is dominated by NO2->NOz and usually a few hours.

Page 3, lines 3-5: WRF-Chem should be included as an example, because studies cited below used it.

Page 4, line 8: “and” should be “which” in “… and is also used for …”. Also please elaborate how data assimilation was done from GFS to WRF.

Page 5, line 5: please clarify if the power plant emissions data were only used for comparison or also used in the CAMx model simulation.

Page 5, lines 17-18: it seems inadequate to only mention LNOx briefly here, given a subsection is devoted to it later. Please provide more information such as how the timing/locations of lightning were incorporated in the model and how LNOx generation is parameterized.

Page 5, line 24: at least the SWIR should be separated with other bands as it is a separate spectrometer, i.e., UV-VIS-NIR and SWIR.

Page 6, line 8: it is TM5-MP model but TM5 in page 13, line 23, which I assume refers to the same model.

Page 6: TROPOMI HCHO level 2 product should be described similarly to NO2, maybe more briefly.

Page 6, line 34: the gridded TROPOMI NO2 averaging kernel should be described in more detail. From later mentioning it seems to be from gridded TM model outputs, which are not part of the TROPOMI product.

Page 7, line 14: please justify the usage of 100-m wind speed. The wind speed directly scales with effective lifetime and emission rate according to equation 2. By using different wind speed, one may practically scale the emission rate estimates.

Page 7, line 21: Phi should be a “Gaussian” CDF, not any generic CDF.

Page 7, line 27: the NOx:NO2 ratio of 1.32 has been widely used in EMG studies and alike. Later a 10% uncertainty was assigned to it (page 18, line 27). The Beirle et al. (2021) paper cited here actually used ozone mixing ratio for a more accurate estimation. The 1.32 number came from Seinfeld and Pandis that explicitly used [NOx] = 100 ppb,  [O3] = 100 ppb, and jNO2 = 0.015 s-1. Given this is a CTM study, it seems hardly justified to still use such a simplistic treatment.

Equations 2 and 3: please clearly distinguish the two “emissions”. One is emission rate in amount of substance per unit time, and the other one is (emission) flux in amount of substance per unit time per unit area.

Page 11, line 11: why were only the central 250 across track positions used?

Page 11, line 16-17: the sentence about lifetime scales with the length scale and equals to L/(2U) does not seem to be connected with the context. Please check if it was supposed to be there.

Figure 3: please explain what the red solid dots and the black dashed lines mean. The caption states on the right there are NO2 shape profiles from “two model simulations”, but I can only see one profile in each panel.

Page 12, line 14: “model out” should probably be “model output”.

Figure 4: in the third panel, dividing the tropospheric column into “below 2km”, “lightning NOx”, and “other” seems a bit strange, as it mixed up vertical layers with sources. An implicit assumption here is that LNOx is exclusively constrained from 2 km to tropopause. Although this could be a good assumption, it is better to add some clarification in the text.

Page 13, lines 8-10: “while the urban NO2 will artificially decrease” appeared twice.

Page 13, lines 17-20: these hypotheses also appeared in the abstract and in page 16, lines 7-10 but in slightly different forms. Please consider consolidating and reconciling them. It also a bit strange as hypotheses were formed but not followed by testing or design of testing.

Page 15, line 6: as correlation is mentioned in the text, the associated number should be correlation coefficient, r, not r², which appears to be coefficient of determination. It is denoted as R² in Appendix A though.

Figure 6: the caption seems incorrect as “bias-corrected” and “downscaled” TROPOMI NO2 was never mentioned.

Page 16, line 10: high solar zenith angles sounds the opposite. Should it be either low SZA or high solar elevation angle?

Page 16, lines 7-10: first, it is questionable how much the “effective lifetime” fitted from EMG can represent real atmospheric chemistry. The 0.5 hour fitted from Fig. 8 right look especially unrealistic. Since this study presents a full CTM simulation, could the reaction rates in the model shed some light on the reliability of the “effective lifetime”?

Page 16, lines 7-10: second, TROPOMI only measures column amount, the factors given (wind, OH, VOC, high sun) do not seem to matter how well TROPOMI measures NO2 column amount.

Page 18, lines 14-20: it would be useful to clarify photochemical and chemical lifetimes of NO2 here, and how the effective lifetime from EMG is related to them.

Page 18, line 23: DISCOVER-AQ in Texas in September 2013 provided lots of vertically resolved NOx measurements already.

Page 19, line 13: please provide the power plants NOx are biased low by how much.

Figure 11: please explain what the numbers in the plot mean.

This study evaluates NOx emissions and ozone production in Texas, which covers several interesting questions, including model-satellite comparison, top-down NOx emission estimates using EMG approach, and HCHO/NO2 for identifying the ozone production regimes. Overall the results are clearly presented, and the figures are informative. There are several interesting findings, which could well be three independent studies, but my concern is that the findings are a little disconnected, and it is not easy to tell the major take-aways from the study. I’d suggest the authors better frame the core questions, and show how each section is connected. Below are some major comments: 

1. It is not clear to me why the authors compare two different versions of TROPOMI NO2. Satellite retrieval products are updated routinely. The newest version should certainly be better, and the older version is already replaced with newer one. I don’t think there is much scientific value of such comparison. Evaluation of different versions should be in the technical document of TROPOMI NO2. I’d suggest the authors stick to the newest, publicly available version.
2. Related to previous question, I’d suggest the authors use newest version TROPOMI HCHO and NO2 data in Section 4.2 to evaluate ozone sensitivity. The authors have shown better performance of v2, but switch to v1 in 4.2. It is also not clear to me why the authors use an outdated version of TROPOMI HCHO (v1.1) while a newer version is already available. 
3. It’s also not clear to me why satellite HCHO is less sensitive to the application of AK. While it may not help improve the spatial patterns of HCHO, application of AK may resolve the overall difference between modeled and satellite HCHO. It’d be great if authors can show a figure or two to illustrate this point.  

Specific comments:

1. Figure 3 (right): What does the red line mean? Which shape profile?
2. Figure 4: There is no explanation of the right figure in the caption. Why is it continuous? Are these generated for selected pixels? 
3. Page 16 Lines 5 to 10: The authors show TROPOMI NO2 is significantly lower than modeled NO2, and they attribute this to issues with TROPOMI. While it may be true, but could this also be due to model unable to capture the sub-grid processes?
4. Page 20 Line 12: This sentence is confusing. 
5. Table 3 shows the derived NOx emissions are sensitive to NO2 lifetime. How much confidence do you have with the NOx lifetime in EMG approach? Does it agree with the NO2 lifetime simulated from model? And how does the seasonal variation in NO2 lifetime affect the calculation of emissions? 
6. Figure 12: It’s interesting to see there is almost no diurnal cycle with HCHO. I think both biogenic emissions and photolysis rate vary diurnally. It’d be interesting to illustrate why HCHO is relatively constant, and what this would mean for ozone production. 
7. Figure A1: There seems to be a large discrepancy between CAMx and observed NOx and NOy. I’d suggest the authors investigate why CAMx is biased too low, and how this would affect the interpretation of other findings.