|Second review of “Neutral atmosphere temperature change at 90 km, 70°N, 19°E, 2003–2014” by Holmen et al.|
1. It is clear that quite a bit of massaging of the data has been necessary to get the temperature time series used in the paper. As far as I can tell, the authors have clearly laid out the steps that they took and the various uncertainties added at each step. I think readers are suitably warned about the reliability of the resulting temperature data and, therefore, I do not have anything more to add about it. Others who analyze similar types of data will be interested in the steps to follow.
2. It seems that my previous review was not clear enough about my concerns with the solar & trend analysis. It is simply not possible to separate these two sources of change during this period. The short data record would be a serious problem for any analysis that is limited to one solar cycle or less. (Note: despite being 11 years, your data span is substantially shorter than the full cycle between the maxima of SS 23 and 24, hence less than one solar cycle.) Because of the strong difference in the solar flux at the beginning and end of your period, there is a marked trend in the F10.7 flux itself. The atmospheric temperature trend (or change; whichever you call it, it is the same problem) and the response to solar variability cannot be separated. This has nothing to do with which analysis technique you use, it is not possible because the two terms you are using as predictors (solar flux and time) are significantly correlated.
3. As per the above comment, the paper should not be published with the trend and solar analysis (Sections 4, 5.2, and 6). If you want to show the deseasonalized data along with the F10.7 flux, that would be OK. If so, it is necessary to acknowledge the ambiguity between possible solar cycle and trend variations. For example, time information could be included in Figure 8 by color-coding based on year. As shown, the figure is not an adequate representation of the solar response and therefore the fits (red and blue lines) are not meaningful. Further analysis to actually define these two responses separately should not be included because it cannot be scientifically supported.
1. (p. 15292, revision near top) Do you mean decreasing concentrations of O3?