Recently Mikhailov and Perrone (2020; hereafter MP20) proposed the mechanism of neutral atomic oxygen reduction to explain the density decrease during the storm recovery phase, and claimed that there is no need to introduce the storm-time overcooling concept. Unfortunately, their methodology, model applicability and hypothesis all have problems, resulting in conclusions that are not meaningful. The reasons are given as follows.
The MethodologyUsed in MP20 is Questionable, and in Fact, Incorrect, and the Results Lack of Proper Uncertainty Estimate and Verification MP20 determined simultaneously several thermospheric parameters including [O], [O 2 ], [N 2 ], exospheric temperature Tex and ion vertical drifts by matching the calculated ionospheric parameters using a 1-D ionospheric model with ionosonde observations during the stormtime. The only ionosonde parameters used are F 2 and F 1 peak electron densities (foF2 and foF1). These results arrive with ambiguity and inadequate uncertainty estimate. In a commentary paper, Zhang et al. ( 2018) pointed out that their approach, developed in the authors' earlier papers (also similar to what is used in MP20), is questionable; however, the reply by Perrone and Mikhailov (2018) did not seem to have addressed some of the key issues raised in the Zhang et al. (2018) commentary. For example, the ambiguity/uncertainty among the derived parameters is not quantified. This problem arises because a foF2 increase, for example, could be related to [O] increase, [O 2 ] and/or [N 2 ] decrease, changes in the neutral winds, changes in ion drifts and changes in ambipolar diffusion, and/or a different combination of these parameters. Zhang et al. (2001) demonstrated this ambiguous behavior even when full Ne profiles were used. It is interested to note that Perrone and Mikhailov (2017) compared the differences between neutral mass densities from CHAMP and their method that found the deviations are larger than 10% for approximately 45 out of 70 samples. We question whether this kind of accuracy is typical during storms and is adequate to explain a small neutral mass density reduction of less than 5%-10% (see Tables in MP20). Another major problem with the technique is that the derived neutral parameters are not physically self-consistent, but are modifications to the background MSIS model. The complicated, coupled relationship between density, temperature, and composition in the thermosphere exhibits large nonlinearity, which is not well represented by the empirical MSIS model, particularly, during storms.