These comments refer to a revised submission of a paper previously reviewed in ACPD.
In brief, the work uses the surface ozone concentrations from 6 of the ACCMIP models – for a recent time period (2000) and for 2100 under the RCP2.6, RCP4.5 and RCP8.5 scenarios – to calculate changes in AOT40 values and in a potential vegetation risk factor due to ozone (IO3) between 2000 and 2100 for the three climate scenarios. The variations in the values of the metrics and in the simulated future changes in the metrics are discussed in terms of differences between the 6 models, global geographic patterns, and differences between the future scenarios.
Scientific comments:
I raised a number of comments on the previous draft relating to lack of clarity with the methodology for calculating AOT40 and IO3 values, together with some concerns about validity of some of the statements/conclusions derived from their calculations. The specific methodological queries have largely been addressed, although in the revised text the authors have now introduced new lack of clarity in how they assigned particular IO3 values to each model grid square. In L160 it is stated that vegetation was grouped into the three categories of ‘conifer’, ‘crops’ and ‘deciduous trees.’ What are tropical rain forests classified as? What are grasslands and dry shrub classified as? These don’t seem to map readily onto the three specified vegetation classifications. I am also not clear how a % change in IO3 can be derived if the AOT40 value in a particular grid square was initially zero (L425). An IO3 value may become non-zero in a new scenario, where previously it was zero, but it is not possible to assign a % change to a change from zero.
Whilst the authors may have presented a lot of careful and numerically-correct calculations, I still wonder whether the use of various ‘globally-wide’ approximations means that one has to be very cautious about taking the results, particularly for changes in IO3, too literally. Example of these ‘global’ approximations include the calculation of AOT40 values from 08.00 to 20.00 everywhere globally regardless of latitude and time of year, the calculation of AOT40 for a full-year, and the assignment of all global vegetation to just 3 vegetation classes. The authors have responded that because they are interested in examining relative changes in impacts on vegetation these issues are not so important, but surely if growing season length (and extent of daylight hours) at a particular location is relevant for ozone-induced injury then how the annual distributions of ozone concentrations change in the future under different scenarios (and also how growing seasons change under different scenarios) will have a quantitative impact on the extent of ozone-induced injury in the real world compared to these modelled worlds.
Much of the conclusions section remains more like general discussion (particularly in its extensive reference to prior literature) than ‘take home’ conclusions for the reader. There are also a few issues with some of the text in this section: (i) the sentence in L529-530 is not a complete sentence; (ii) the statement in L548-549 that AOT40-based critical levels “will be exceeded over many areas…” is not clear: to which RCP scenario(s) is this sentence referring? It is also potentially a bit misleading because even though AOT40 may be exceeded in certain areas, actually for two of the three future scenarios investigated AOT40 values will be less exceeded in the future than in recent history; (iii) L569 refers to sensitivity of grasslands to ozone injury but yet the authors’ methodology does not refer to how IO3 is calculated for grasslands.
Please explicitly state somewhere in the text that tabulated and quoted values of global, NH and SH mean surface ozone, AOT40, and IO3 are derived from averaging values over the global/NH/SH land areas only, not over their full respective geographic domains. (At least, I presume that this is the case.) |