This paper investigate the mechanism of short-term surface ozone anomalies in the urban areas over the Qinghai-Tibet Plateau. The topic and presentation are fine in general. A revision according the following comments should be provided.

The description for producing meteorological normalized concentration in Section 3.3 is quite vague. For example, (1) how the final result be sensitive to the four-week period window? If it is chosen to reflect the seasonal variability, is it really considered to be superior/useful than traditional deseasonalization methods? (2) p6, l17, "This selection process was repeated 1000 times to generate a final input dataset." How is the final input being generated exactly? By using sample mean, median, or anything else? Since the figure results presented in this study are not as variable as I expected from the main text, I am not quite convinced that a random resampling method would lead to such smooth results. A time series plot of original data for each station is desired.

A significant portion of thIs study is devoted to the discussion of ozone extreme values. To provide a more systematic discussion, an to facilitate better communication, I suggest the authors should quantitatively work on the percentile variation instead (e.g. the 5th and 95th), as suggested by following references: 

Cooper, O. R., Gao, R. S., Tarasick, D., Leblanc, T., & Sweeney, C. (2012). Long‐term ozone trends at rural ozone monitoring sites across the United States, 1990–2010. Journal of Geophysical Research: Atmospheres, 117(D22).

Munir, S., Chen, H., & Ropkins, K. (2012). Modelling the impact of road traffic on ground level ozone concentration using a quantile regression approach. Atmospheric environment, 60, 283-291.

Chang, K. L., Schultz, M. G., Lan, X., McClure-Begley, A., Petropavlovskikh, I., Xu, X., & Ziemke, J. R. (2021). Trend detection of atmospheric time series: Incorporating appropriate uncertainty estimates and handling extreme events. Elem Sci Anth, 9(1), 00035.

Wells, B., Dolwick, P., Eder, B., Evangelista, M., Foley, K., Mannshardt, E., ... & Weishampel, A. (2021). Improved estimation of trends in US ozone concentrations adjusted for interannual variability in meteorological conditions. Atmospheric Environment, 248, 118234.

Minor suggestions:

p4, l6-7, the data quality control procedures should be briefly stated.

p6, l1-2, this part seems to come from nowhere.

p8, l4, Yin et al. (2017)

p10,l25, what does it mean for "seasonal cycles of surface ozone anomalies"? Should the anomaly is deseasonalized already in Eq (5)? If it refers to remaining seasonality variation, can it imply that the methodology in Eq (5) is not appropriate?

This study examined surface ozone variation over the QTP regions and quantified the role of anthropogenic emissions and meteorology in ozone changes from daily scale to multiyear scale using the random forest model. I think the topic of this study is within the scope of ACP journal. The dataset and method applied here are reasonable. However, I feel the results are overall descriptive and more in-depth analysis should be done to improve the current manuscript.

The authors said that the implication of this study is for ozone control. However, the mean ozone level over QTP is far behind the national AQ standard. Is ozone an air pollution issue over QTP? They summarized the ozone nonattainment days but listed them as Table S1 in the supplementary. This table should be move into the main text.

The authors quantified the anthropogenic and meteorological contributions to ozone changes at different time scale, but they failed to explain them further. In Section 6.4, they only discussed the anthropogenic or meteorological roles generally. But it is important to know if it works for all time scales. Or the authors could focus on some time scale to discuss.  

P9L37: I don’t understand why it is needed to quantify anthropogenic and meteorological contribution to ozone changes at the diurnal scale.

P4L27-28: Has MERRA2 data verified by surface measurements over QTP?

P6L18-19: change “zone” to “ozone”. Please check over through the text.

P10L19: Could you specify the dominant meteorological variables responsible for ozone nonattainment events?

P11L1: This section talks about ozone changes over multiyear time frame.

P11L12-13: It is not clear how these ozone changes were driven by anthropogenic emissions.

P12L30-31: Please make it clear how ozone changes are consistent with emission changes.