|Comments to the revised version of amt-2022-51 by Choo et al.|
The revised version includes a large number of improvements. Thanks to the authors for the revision, especially for their efforts to answer the review comments.
There are some aspects remaining that require correction. However, I am sure the authors can address these. After revision of the remaining issues, I strongly recommend publication of the study in AMT.
Overall, apparently a new NO2 data version has been used in the revised version. It is very shortly mentioned in one of the answers that NO2 VCD has been recalculated using, e.g., a different polynomial. Substantial differences in the NO2 values occur for the comparison with the other data products (i.e. changed values for the correlation, slope, and number of data points). Especially the data points in collocation with OMI are now 37 instead of 53, the reason for which is unclear. For the GeoTASO-Pandora comparison, now a slope of 1.15 is given instead of the original 1.48. Not required for the manuscript maybe, but as an answer in the discussion, it would have been good to explain this a bit more prominently, that a new NO2 data version is used in the revised version.
Comment 1 (VCD stripe in Fig. 5a):
The surprising VCD values close to the Boryeong power plant (latitudinal stripe at 126.4°E) have not been improved and not commented, see Fig. 5a. It is not clear, how the stripe of large VCDs evolve from the given SCD and AMF. The SCD shows well the strong plume from the Boryeong power plant, while the VCD is dominated by artefacts. Even if the AMF calculation is never perfect, it should be clarified where this problem evolves from. Even if the data cannot be improved, it is recommended to state the cause of the artefact and a reason for not correcting it.
(The explanation and Figure R1 in the AC2 supplement would be more helpful if they were in better quality (the maps are somewhat hazy and blurred), and ideally with the same aspect ratio and grid lines as Fig. 5.)
Comment 2 (treatment of GeoTASO resolution):
One open question is the resolution of GeoTASO and how this is advertised and used in the manuscript. The authors’ explanations seem to motivate that the fine resolution is not necessary (e.g., emphasizing the reduction of random uncertainty at coarser resolution), in contrast to the advertisement of the GeoTASO 250 m resolution. The fine resolution is suitable for detection of the largest NO2 values. Larger spatial pixels always dilute the signal. Clearly, the 0.01° resolution also shows the industrial enhancements and variability of NO2, although the 250 m resolution is even more suited for smaller scale variations. In any case, it is acceptable to use 0.01° resolution throughout the manuscript, however, in that case the 250 m should not be advertised in the abstract without any restriction. At least some modification of the text is recommended, if the authors decide not to show any native 250m resolution NO2 data. From the abstract:
“For the first time, we examine highly resolved (250 m × 250 m resolution) tropospheric NO2 over the Seoul and Busan metropolitan regions, and the industrial regions of Anmyeon.”
Could be changed to something similar to this:
“For the first time, we examine highly resolved tropospheric NO2 over the Seoul and Busan metropolitan regions, and the industrial regions of Anmyeon. The native 250 m × 250 m resolution data of the instrument is binned to 0.01° resolution for the analysis of NO2 presented here.”
p. 10, ll. 249-252: Here, the authors include some modified text, and this states one of the disadvantages of the binning: “… larger VCDs at 250m resolutions do not necessarily lead to larger VCDs at wider resolutions.” So some larger signals are diluted, but this is a trade-off with better signal quality.
Overall, if the native data is not the data shown and analysed, the reasoning for which data resolution is used should be stated very shortly once at the beginning with the main aspects for the choice, advantages and disadvantages, making clear that the final choice is finding a balance between the well-known influences (more details and stronger variability in the finer resolution, better data quality while still retaining a large portion of the variability at slightly coarser resolution).
Comment 3 (variability of AMF parameters on GeoTASO scale):
In the error analysis, the authors do not take into account the error that results from the fact that the values of AOD, ALH, SSA and SR used for AMF determination might not fit to the small GeoTASO ground pixel. Although the uncertainties might be suitable for the data products used, the question on how representative these values on a larger spatial grid are for the finer GeoTASO pixels, is not answered. The authors now mention that there is an additional error resulting from this aspect, but some further clarification would help.
p.19, ll.406-408: Here, it should be clarified which grid box is referred to. For example: “…variability of the parameters within the respective CTM or satellite grid box.”
Please add a sentence directly after the above in the sense of:
“If values such as surface reflectance are assumed constant over larger areas, the fundamental spatial variability in this input data increases the uncertainty of the AMF and hence of the determined NO2 VCD on the respective finer spatial scale.”
An idea of the variability of the surface reflectance of GeoTASO measurements within a MODIS grid box can be gained by looking at the relative intensity of the GeoTASO spectral data within an area, for which constant SR is assumed. Even if no specific calculations or refinements of the input data are made for the current manuscript, this influence can be roughly estimated.
p.7, l. 186: Considering the response to the earlier question regarding the GDF spread, the numbers are still surprising. A spread in GDF is expected, but in the author’s response the FWHM at 438.4 nm varies only between 0.8788 nm and 0.8985 nm. Does the GDF vary that strongly (0.7 to 1.0 nm as stated) at different wavelengths? This is a bit surprising, if variation is that small at 438.4nm. The table given in the supplement to AC2 is not needed in the manuscript.
p. 8-9, ll. 197-202 (Figure 3 and caption): Description and label of the x-axis (wavelength / nm) is missing. This information needs to be added.
p.9, ll.224-227: The discussion of the NO2 VCD retrieval still misses some explanation. If the NO2 SCD used in the equation is the original SCD from the DOAS retrieval, part of the stratospheric SCD is already taken into account, because it is part of the reference measurement South of Jeju as well as of the current spectral measurement. The earlier comment here (originally ll. 182-185 in the original manuscript version), did not ask for new data or new model calculation, but simply for mentioning the correct relevant factors in the text. Since the retrieved SCD is the difference between the actual measurement and the reference measurement, the stratospheric, free tropospheric and background NO2 from the reference region are effectively subtracted from each measurement. Only the changes in time and space of these values are still included in the final NO2 product. These aspects of the retrieval and the assumptions behind should be stated in the text. The original question addressed the changes of these values, especially the changes in the stratospheric part over time. The stratospheric NO2 during the morning reference is different from the stratospheric NO2 during the GeoTASO measurement time. This change in stratospheric NO2 could be mentioned (and potentially quantified) in the text, as well as the potential influence of changes in free tropospheric and background NO2.
p.19, l.409: The values AMF_true and AMF_new are not explained in the text. Maybe “centre” or “mean” would be rather suitable than “true” and “perturbed” would be rather suitable than “new”?
Why is the change in AMF divided by the arithmetic mean of AMF_true and AMF_new, while the AMF_true is the point of reference? This way, changes to larger AMF (AMF_new > AMF_true) always cause smaller relative differences than changes to lower AMF. This is not clear.
p. 8, l.190: Probably this point was overlooked in the revision. It would be recommended to add a reference (a citation) for the applied H2O cross section.
p.19, ll. 404-408: There are also earlier papers, that analyse the NO2 profile influence on the AMF error.
p.18, ll. 384-385: Typo in the value for SSA. Is the uncertainty really only 0.001? In that case the value should read 0.980 (0.001). The same number of digits is required for value and uncertainty. Possibly, there is a typo in the value of the uncertainty, which is probably much larger?
p.21, l. 439: some typos, sentence should read “… It is thought that the reason for the low slope …”
p.21, l.438: Typo in GeoTASO, here it says “GeoTAOS”.
p.22, l. 443: Same typo in GeoTASO.
Please change sentences with “I” as subject to “We” or even better to passive speech. (e.g. p.22, l. 455 and others).
It is recommended that the final manuscript is checked for language and grammar improvements.