|I recognized that the authors made some efforts to improve the presentation of their results, and they now include more details regarding used datasets and their modeling approach. These are helpful.|
I still have several major concerns as follows:
1. The authors tried really hard to dig some trend in BVOC emissions during 1998-2013 based on information in Table 6. I do not see a clear increasing trend from the contents in Table 6. I am not convinced that results from these independent studies are directly comparable to produce a trend estimate. This is because these studies used different approaches and/or focused on different areas, and uncertainty of BVOC emissions from this study is unknown (also see my major concern 4 regarding impact from WRF modeling and evaluation). The authors argue that the increasing forest cover contributed to an increase in BVOC emissions, but they do not discuss the impact of meteorology and climate on BVOC emissions during this period. Does a growing forest cover decrease air temperature, which would compensate the emission changes due to the increase in source area and intensity?
2. The authors state that the BVOC estimates represent year 2013. The WRF fields and LAI data are based on 2013, but three of the five experiments used FROM land cover, which is based on year 2010. So please use more accurate language about the year your BVOC emissions represent.
3. The authors should be careful when drawing conclusions that BVOC emissions are insensitive to LAI but sensitive to PFT. Based on my observation, the three LAI products over forest areas differed by ~1.0 in maximum in warm seasons, when the absolute values are 3-5. The spatial patterns of the LAI from three products look consistent as well. The relative differences among PFT data are much larger according to Figure 6. Is the smaller sensitivity to used LAI products (than to PFT products) due to the smaller discrepancy among used LAI products? Can the absolute sensitivities be normalized by LAI or PFT discrepancies?
4. WRF simulations: I wonder why the authors evaluated WRF T2 using hourly observations (Table 3 in author response file) but did not want to include those results in the revised manuscript. There are some differences between those results in that Table and current Table 2. In fact, the current presentation of WRF evaluation is not satisfactory. Specifically,
1) All evaluation statistics are reported for the whole year whereas the BVOC emissions are presented month by month, and an obvious seasonal cycle is seen from the MEGAN BVOC emissions. A clearer way to present your model performance is to show time series plots of the statistical metrics.
2) The authors speculate that the largest cold biases are due to unrealistic land use data in WRF, why don’t you show clearly how WRF land use input looks like over the Beijing area? And why not try to correct them before running WRF? Are there other factors that could contribute to model biases (e.g., FNL, physics, terrain representation, vertical resolution)?
3) The authors should consider adding discussions on how these T2 biases may have introduced uncertainty to their BVOC estimates (e.g., adding a sensitivity run, or cite previous studies on these topics suitably). How does this uncertainty affect their conclusions on BVOC emission trends?
4) Radiation evaluation is only available at Beijing station, where BVOC emissions are low. Not quite useful information.
L13-14: “Forest coverage rose from 20.6% to 35.8% from 1998 to 2013 in Beijing according to the National Forest Resource Survey (NFRS).” I suggest to write a concise sentence to emphasize accurate representations of land cover for recent years is important to estimating biogenic emissions and addressing air quality issues.
L21: what is the “baseline” calculation? There is no prior definition in abstract for this.
L22: delete “the” before “isoprene”
L25: “The different PFT inputs strongly influenced the spatial distribution of BVOC emissions, which is determined by the spatial distribution of PFTs.” ->consider shortening it.
L24: why is mountain mentioned here? Perhaps the authors meant the terrain influences on meteorology that drives BVOC emissions? If so, say that clearly.
L25: “it is necessary to investigate and evaluate the impact of biogenic sources on local air pollution.” -> needs to be reworded. You are not quantitatively evaluating BVOC impacts on air pollution in this study; you only produce BVOC emissions for future evaluations of their impact on air quality.
L7: clearly say 31 layers included 4 layers of soil from the Noah land surface model. In most publications, WRF model layer does not include soil layers.
L20-23: Causes of these negative biases are not well explained. See also my major comment 4.
Sect 2.2.2: Some information on the temporal frequency of these PFT and LAI products is necessary. I didn’t notice a little information until the results section about the temporal frequencies of some of these LAI products.
L5: consider a different word than “regarded”
L9: spell out “TM”, also for the “ETM” in P11
L5: “Additionally, the emissions differ more between experiments in summer than in winter, indicating that the emission estimates for summer are more uncertain than those for winter.” It is not surprising because summertime emissions are high. Be clear that you refer to the absolute values rather than a normalized metric by absolute emission amounts.
L9-11: starting from “As shown in…”, something is messed up. These are not complete sentences.
Finally, the authors concluded that the use of cross-walking table is important (in abstract and conclusion). Why are related results only be shown in the supplement?