Review report of the manuscript entitled, “Calibrating estimates of ionospheric long-term change” authored by Christopher Scott, Matthew Wild, Luke Barnard, Bingkun Yu, Tatsuhiro Yokoyama, Michael Lockwood, Cathryn Mitchel, John Coxon, and Andrew Kavanagh.

Investigations on the ionospheric trends is an important area of research.  Researchers have been using hmF2 and foF2 from ionograms to address this issue. The authors of this work dwell on the corrections for possible uncertainties that may be introduced in the use of these parameters for assessing the long term trends given the fact that there is ambiguity in the rates of changes in the hmF2 in different studies reported in the literature. Thus, the current manuscript attempt to investigate the efficacy of deriving long-term ionospheric trends in hmF2 using empirical formulae and to investigate the extent to which this can reconcile the difference in trends derived from the global network of ionospheric monitoring stations.

Thus, the authors suggest using foF2/foF1 ratio to qualify the values of hmF2 for a mid-latitude location.  For that they use common data from the MU radar (from 1986-2020) and the ionosonde at Kokubunji (1957 – 2020)  to arrive at the frequency rations of F2 to F1 regions to “calibrate” the hmF2 values from ionosonde.

This is an interesting effort which needs to be encouraged. There are, however, several loose ends, incomplete information, which need to be addressed before this article can be considered further.

1. The formation of F1 layer has seasonal dependence. So, in order to provide an appropriate correction, which is more “accurate”, why not use the data from only that season?  It will reduce the number of points, however, the values considered for such a season over 35 years may provide a different (better?) correction?
2. Line 20: How much is the role of radiative cooling by CO2 in the thermospheric cooling?
3. Line nos: 92 and 96, correct the citations.
4. Is there any study that describe the relationship between stratospheric cooling with ionosphere? If it is there then provide the reference.
5. The title of section 1.2 can be different as it does not match with the text.
6. Line 126: define M3000F2.
7. In equation 2, what is M?
8. In eq. 5, define M_o.
9. Line 236: Insert ‘and’ before ΔM.
10. Equation 4 has been used for the estimation of hmF2. According to Bradley and Dudeney, (1973) this relation is not valid for the values of x_E < 1.7. Then, how the estimation has been carried out? In addition, what about those ionograms when the signal of E-layer is insignificant even in daytime?
11. Mention the R square values in Figure-2.
12. Correct the caption of Figure 2, such as, write +/- at the place of pm, and change the symbol by < for the case of SZA values.
13. In Fig. 1, there is a significant spread in the hmF2 values derived from ISR values which is not present in ionosonde derived values. Can author comment on this?
14. Line 341 – 342: daytime is when SZA < 90⁰. In these sentences it is mentioned incorrectly.
15. 5 is incomplete. The text points to the ISR data as well (388 – 389), but it does not exist in the Figure.  The figure caption too alludes to this.  ISR hmF2 also to be plotted in this figure, similar to the format in Fig. 3.
16. Lines 411-412: Explain how the percentage error is reconstructed using gradient and offset.
17. Lines 438- 439: This statement may be substantiated by an appropriate reference.
18. Lines 462 – 463: In the estimation of range bias, why inospheric profile is integrated only up to hmF2 as topside ionosphere also affects the radio wave propagation?
19. Lines 467-468: can one carryout a simple subtraction like this?
20. One can expect that the neutral wind changes with the magnitude of geomagnetic activity. Can author comment on this aspect?
21. According to these results, the composition effect that varies with geomagnetic activities, should be taken care of for the long-term study. It is mentioned at the line number 119 that regression model is not significant alone to remove the effect of geomagnetic activities. But, in Fig. 8, am index shows clear proportional behaviour with composition proxy. Then, how do author think that correcting the compositional variations in the hmF2 variation will give better result? Could the authors compare their results with the study of Xu et al., (2004) to validate/assess their method?

=== End of review  ====

Scott et al provide a compelling case that previous efforts to discern climate-driven lowering of the ionosphere have been hampered by a reliance on semi-empirical equations which incorrectly assume any F1 layer present has an insignificant effect.

They show F1 layers can cause biases in height estimates which risk masking any climate trends, or causing misattributions.

They attribute this to thermospheric composition effects, driven by geomagnetic activity.

The manuscript is of very high quality, and only needs a few revisions - should be minor in effort - to exclude some minor remaining doubts about potentially confounding effects, and make their case unarguable. As well as to make some of their reasoning a little more explicit, improving their work's accessibility to a wider audience. A more detailed breakdown of general and specific comments, and technical corrections is attached.