Dear Authors

Thank you for submitting a revised version of your manuscript. Both reviewers have provided very positive feedback on your paper, including also valuable and thoughtful suggestions for changes. I very much appreciate the effort you put into answering the reviewers' requests for changes and I find your answers convincing. However, I would like to encourage a more explicit transfer of the following reviewers' reflections into the manuscript by asking for the changes listed below. In addition, I have a few small requests for more precise formulations :
1) L.3 ", and sometimes are reaching the coastal fog and low clouds" -> can you be more precise? e.g. "Each year between June and October, in some synoptic settings, absorbing biomass burning aerosols (BBA) are overlying the stratocumulus clouds in the adjacent Southeast Atlantic, thereby modulating the time of the day when FLC dissolve. In favorable synoptic conditions, this layer of BBA reaches Namibia and its desert, where it interacts with coastal fog and low clouds ." This would help a bit because up until here a non-expert doesn't know what the impact of BBA is on the cloud layer.
2) I do agree with Rev. 1 about their point on the artificial effects due to the 12h satellite data assimilation on water vapour and TCW, you may therefore add their suggested reference at L. 150.
Vicencio, J., C. Böhm, J.H. Schween, U. Löhnert, and S. Crewell, 2023: A comparative study of the atmospheric water vapor in the Atacama and Namib Desert, Global and Planetary Change, 104320, https://doi.org/10.1016/j.gloplacha.2023.104320.
For me personally this reference would help to understand what the implications of only assimilating satellite obs in the 12h time window are at this specific location of the text, while Hersbach et al. 2020 is a more general reference for the data assimilation window.
3) I do also agree with Rev. 1 on the transient nature of BBA influence: "Coming back to my suspicion that the strong pollution mainly occurs as episodes. This counteracts the assumption of a linear model to predict FLC times." Can you mention this specific limitation of the chosen prediction framework more explicitly in the conclusions. Also the final sentence from the Abstract sounds a bit like a dead end. Can you add a sentence that would show the way forward? LES modelling? Use of trajectories or combination with other tracers?
4) Semi-direct (both reviewers): is now defined in the intro but in the abstract I also stumble over it and would very much benefit from a short explanation. Either avoid semi-direct and directly say that you mean impact on thermodynamics and stability through interaction with radiation or, explain what semi-direct is in a relative clause: "In this study, a novel 15-year data set of geostationary satellite observations in the Namib Desert is used together with reanalysis data in order to better understand the semi-direct impacts of BBA on the dissipation of FLCs in the Namib desert, i.e. through adjustments of the atmospheric stability by the interaction of the aerosols with radiation.
5) L. 78 and 80: I agree with reviewer 1 that it is unclear at this stage (and later on) how you can disentangle the BBA effects and meteorology
6) L. 118-121: please shortly mention why this is the most frequent dissipation time.
7) L. 300-301: Are these transient events not well captured or not captured at all? And what do you suggest as a way forward to address this point in the future?
8) Rev.2 "How good is this column-integrated BCAOD reanalysis, since there are aircraft measurements from the field campaigns in the region, I wonder if this product can be validated against observations?" and the next remark by Rev.2 about ACAOD from NASA: I understand the motivation for using BCAOD reanalyses data in your study, but this choice should be justified explicitly in Section 2.2. Mentioning that other more observationally-based products can be used in particular for case studies and to evaluated the reanalysis product in more detail would be helpful for the reader.
9) Rev. 2's point about why not using an approach that would accommodate for the non-linear nature of the prediction problem. Please shortly explain the rationale behind using a ridge regression model at the beginning of Section 2.5.

Technical comment: °S and °N should not be in italics.

After these final corrections, the paper will be accepted and published in ACP.

Kind regards
Franziska Aemisegger