The authors have considered all my comments on the previous version of the manuscript and for most part have satisfactorily addressed them through revisions in the manuscript. While the revised manuscript is much improved, I feel there still remain a few aspects of the discussion that could benefit from additional clarification and/or substantiation. The following comments are offered.
1) L202-204: Please indicate at what model levels (altitudes) the ICs for O3, NO, NO2 were adjusted by the OI method – I believe it is only at the surface, but for completeness it should be explicitly stated.
2) The discussion related to the default BCs as a “control run”, its representativeness for the conditions being simulated, and inferences drawn relative to the case with “dynamic” BCs still raise some questions. While it is okay to convey it as a control run, I feel some of the inferences drawn and the suggested importance of the use of the default BCs need to more carefully conveyed. If the intent of the “control” run is to infer the pollution burden solely associated with emissions within the LIS domain, wouldn’t a case with no inflow (i.e., zero boundary conditions) be more appropriate? The static default LBCs appear to be so unrepresentative for the small LIS domain with surrounding regions of relatively high emissions, that anything with even slightly better representativeness would yield improved performance for the aggregate statistics by reducing the systematic biases induced by the default LBCs. I am thus a bit wary of some of the broad conclusions that are drawn on the suitability of one over the other for different seasons.
3) For instance, the discussion starting on L304 states that the “influence of BCs is less important during the cold season” – I would exercise some caution in making such broad sweeping statements based on the limited domain extent and conditions examined here. In fact, there are many studies documenting the importance of BC specification during times when “local” pollution production is slow (as in cool seasons). At the very least, the statement should be qualified to convey the specific and limited conditions it is based on. By the same token, I do not believe the importance of specifying space and time varying BC, especially for small geographic domains as the LIS region should be diminished just because deriving them from larger scale simulations is considered computationally burdensome for real-time applications (as suggested in the author response).
4) I don’t believe the conclusion on L311 that influence of dynamic BCs is “more significant during the high pollution time” is robust. Dynamic BCs could also be influential in conditions when local production may not be high. The dynamic BCs help better represent the changes in baseline pollution at a location based on changes in large scale forcing and thus should be important at all times. As stated, I think the statement can be easily misconstrued and should either be qualified for specific conditions, or further substantiated, or deleted. It appears that for the cases examined in this study the baseline levels in the LIS were enhanced by regional transport.
5) L323: I think this sentence is somewhat trivial in that it should be expected that a finer resolution simulation will show “more detailed distributions” than a coarser resolution simulation – perhaps it would useful to point out specific features of interest that are better resolved and point the reader to the appropriate illustration supporting that. How does one assess if the more detailed distribution is better?
6) L324: please point to the observational evidence that suggests that “O3 over the LIS is lower than its surroundings”. Is this a persistent feature or only on specific days?
7) L326: based on Figure 2 it is not readily apparent which specific feature of the O3 distribution along the coast agrees better with the observations – it would helpful for the readers if that association is more explicitly pointed out.
8) L327: Given the many differences (in addition to model grid resolution) between these 3km resolution runs and the NAQFC configuration, can one emphatically claim that “This proves the high-resolution simulation can better reproduce the pollutant variability in coastal urban areas”. While I do strongly believe that finer resolution simulations are needed for coastal urban areas, I do not think that the analysis presented necessarily proves that need in a robust manner – the statement appears to be a little too strong for the evidence presented.
9) L462: “2014 NEIs” should be “2014 NEI”
10) L581: How does one ascertain that the grid resolution solely resulted in the NO2 spatial pattern differences between these runs and the NAQFC?
11) L637: it would be useful to point the readers to the illustration that supports the statement that 3km resolution improves simulation of peak values and timing.
12) In response to my query on whether emissions of species other than NOx were “refreshed”, the authors response states that “regional O3 production is more controlled by NOx than VOC” – while this may be true for regional O3 distributions, does it also hold for the NY metro area and the LIS region. I do acknowledge, that changing emissions over the past several decades has changed NOx and VOC sensitive regimes in the area, and would think that capturing both changes in NOx and VOC emissions would be important. It is also important to note that several recent studies suggest perhaps higher VOC emissions from volatile consumer products than represented in the inventories. Thus, it seems that updates to VOC emissions may also influence O3 simulations in some urban areas. It may thus be worthwhile to at least acknowledge the need for updates similar to NOx for other precursor species, while also acknowledging the challenges in doing so.
13) L402: the sentence “In large metropolitan areas, OI adjustments result in spikes in large metropolitan areas indicate the model errors at the time ..” is awkwardly phrased and needs to be rewritten.
14) The additional tables 4 and 5 only re-enforce that the two adjustment techniques did not alter performance much, not necessarily explain likely reasons. I am not sure the various possible reasons outlined in the response, coherently suggest the reason either. If O3 production is low in "NOx saturated" regions, why compare O3 there rather than in downwind regions where the production is enhanced? Is the suggestion that in downwind regions the urban emissions in the plume are "well-mixed" such that the spatial adjustments in emissions in the urban core have little effect? If so, it may be worthwhile speculating so. |