The revised manuscript has been improved but there are still weaknesses that must be taken into consideration before it is accepted for publication. See specific comments below.
The language of the manuscript needs further attention, as well as the flow of the text in some cases. Apart from the corrections suggested in the first review, there has been no attempt to improve it. I am sure that one of the co-authors could help in this context.
These are separated into two categories: First I list my previous comments (in parentheses) which, to my opinion, have not been addressed properly.
(General Comment). I still believe that the use of average ozone columns in RTD simulations does not allow assessing the uncertainty of the presented UVI simulations. I understand however, that repeating the calculations is not possible at this stage of the study.
Comment (4): P6, L6: The authors’ response in my comment on the spectral characteristics of the albedo used in the UV calculations has not been reflected in the text. Please add.
Comment (8): P6, L24: Another case of responding to a comment without passing the explanation in the text. Please explain for the sake of the readers, why using zonally averaged ozone and temperature profiles does not introduce important uncertainties.
Comment (14) was not taken into account, nor was a response from the authors. See below:
P7, L32- P8, L10: Please check this section and make the discussion clearer. I suggest focusing the discussion on differences between model and GB data and not on differences between GB and OMI because this is not the main subject of this paper. The changes reported in the abstract and the conclusions (-4 to 11%) are not discussed at all in this section. Furthermore it would be good to report in Table 3 the spread of the model differences to GB for each station (e.g. the standard deviation).
Comment (15): I haven’t identified in the text any “Specifications on the averaging process and on the limitation of this sensitivity analyses have been added.”, as quoted by the authors.
Comment (18): I insist on my previous comment that you cannot compare changes between two studies that are based on different periods, without mentioning that explicitly in the text. If you see Figure 3 of the Dhomse 2018 paper, it is obvious that total ozone was much higher in 1960 than in 1980 almost everywhere except in the tropics. Taking these differences into consideration you could estimate the additional change in UVI that is induced by the ozone difference. For example, in SH midlatitudes ozone is about 2.5% higher in 1960 meaning that UVI would be at least 3% lower in 1960 than in 1980. This change is comparable to the changes that Bais 2011 and the present study have reported.
Comment (21): The first sentence of the conclusions should indicate how good the UVI is reproduced by the model. Here is a suggestion: “We have shown that the use of CCMI model data with a radiative transfer model (TUV) can reproduce the current climatological values of clear-sky UVI derived from measurements, in most cases, to within ±5%.” This is my take on message from the results of Table 3.
P4, L8: (Mayer et al., 1998) is not a good reference for cloud enhancement. This paper discusses the apparent large increases of total ozone under optically thick clouds. Possible alternatives:
Marín et al., 2017, http://dx.doi.org/10.1016/j.atmosres.2017.05.008
Lovengreen et al., 2005, https://doi.org/10.1029/2004JD005372
P8, L9-10: It is now clear what you mean with “Here, a similar conclusion can be drawn,…”. What is the compulsion and to what it refers? The previous conclusion was about OMI being higher than GB observations. Please revise the text. Moreover, an overestimation of 10% was found for Palmer but there is not discussion on what could cause this discrepancy.
P8, L15: The altitude effect of UVI is the order of 7-8% increase per km altitude (McKenzie et al., 2001). This means that, for Mauna Loa, the relative difference between modeled for sea-level and measured UVI should be at least -25% while in Table 3 the difference is about 5% and of opposite sign. In Figure 1 both the median and mean based model results are higher than the measurements with one exception in June where the difference is in the right direction but much smaller than it should be. Therefore, the statement in line 17: “Nonetheless, at low and mid-latitudes the UVI differences observed are fully compatible... “, is invalid, and of course the comparison is not good for unidentified reasons.
P9, L5: The newly added sentence refers to latitudinal differences, but these are already taken into account in Figure 3. Consider revising to: “Due to seasonal and inter-model differences, this approach has limitations…”
P14, L11-12: These sentences belong to the introduction (already referenced). They are not conclusions of this paper, and as they are written create some confusion as to what means “reproduced” UVI.
P14, L23: According to Fig. 4, UVI does not return to 1960 levels at high southern latitudes when TOZ returns. It stays 5-8% above except for RCP 8.5 when TOZ exceed the 1960 levels. Please rephrase.
P14, L27: The statement: “In mid-latitudes, TOZ should increase between 1960 and 2100 in both hemispheres for all RCPs except RCP 8.5.”, is not correct: a) TOZ must be replaced with UVI. b) UVI increases between 1980 (not 1960) and 2100 only in the NH midlatitudes. In the SH midlatitudes the pattern is completely different, as it appears from Figure.
P15, L3-5: Please state to which time frame refers the quoted variability 0-3%. Surprisingly, in the next sentence the variability reaches 10%, without saying to which latitude or period it refers. Please clarify.
Table 3 caption: In the last line it is stated “For Barrow and Palmer station we selected the six months of their respective summer.” Since summer has only 3 months. It is better to explicitly name the months that were used.
Note: These are only a few examples of language and appearance issues.
P2, L12: I suggest rephrasing to: “Higher increases in UV index are projected …”
P2, L19: I suggest rephrasing to: “…of ozone to 1960 values, with a corresponding pattern of change observed on UVI, …”
P7, L7-8: As there is no other second-level heading, heading “3.1 Model Validation” should be deleted. I suggest to stop heading 3 at the colon to become: “3 Model Validation”.
P14, L14: Replace “ranges” with “ranging”
P15, L4: The word “Logically” does not make sense. Please remove. Also, replace “hemisphere;” with “hemispheres”
Figure 4: I suggest to maintain consistency in the vertical axes labels; some of them are drawn with one decimal, some without. Include a note in the caption, that the scale of the vertical axes is not the same for all panels. Same for Figure 6.
Caption Figure 6: The last sentence has been copied form Figure 4. But this figure does not show the four RCPs, but the three experiments. Please correct.
Caption Figure 8: The word “modelized” does not exist in English. Please replace with modelled (three occurrences). Please change also “from 2000-2010 values” to “from 2000 to 2010“. Same for Figure 9.
Axes titles in Figures 4,6,8,9: Remove the word “percent” since there is already a [%] sign.