The manuscript presents reanalyses model simulation results on convective-permitting scales over Italy with hourly resolution. The study is important since understanding precipitation and provides a method of studying events at hourly scales. Extreme precipitation trends on an hourly basis are important for assessing flood risk, especially as these trends are expected to increase with a warming climate, even in drying areas. The paper is generally well written, and the figures are well presented, however there are a few major concerns, along with some minor comments, that I hope the authors will address in order for the article to be suitable for publication.

General comments

A major concern is that the title and abstract of the paper describe extremes, while a lot of the results and figures describes the full dataset compared to the extreme analysis. The Result section should be altered to highlight the extreme analyses better and maybe reduce the description and figures around the full dataset to better suit the journal. Another major concern is that with the weight given on extremes in titles and abstract, there is only one threshold for extremes used which also is set very high, especially when considering hourly data. Above the mean of 37 datapoints which give only 18-19 events per grid cell if the events are evenly distributed around the mean (a quick calculation sets this threshold around the 9.995 percentile). The threshold applied to define extreme precipitation is exceptionally high, and with so few events included, the resulting trend estimates are highly uncertain. The robustness of the study would be improved if additional analyses were carried out using several lower thresholds, allowing for a more comprehensive assessment of trends.

Specific Comments

Page 2. L27-29: Even drying areas experience more extreme precipitation events.

Page 5. Section 2.2  Median is easier than the 50^th percentile, and used earlier in Introduction.

The Method section needs to be improved to better understand the results presented. Section 2.2 should be rewritten to increase readability. The thresholds and smoothing are first presented, and then described again in the later paragraph, maybe rewrite for better readability. Line 162 – 165 More specifically, … this sentence in especially hard to follow. More than half in which instances?

I’m also curious how time is handled as an event usually lasts more than one hour. Is there any clustering in time?

Section 2.4 Is this Gaussian filter the same as used in section 2.2? If so, does the justification of this radius apply to the earlier smoothing, and move this part in section 2.2?

Table 2, The table could be improved if the names were included in addition to the short names.

Page 9. L196 ERA5 is already introduced as the driver of the dataset.

Page 10, l207-208 Explain better here: How many events would this give over a typical grid cell on the coast or in the mountains?

Page 20. L 358-359: This analysis would also benefit from a lower threshold, as two executive hours above RX1hour is extremely rare.

Page 22. L 423-424, This sentence could be misinterpreted, there could be trends here that were not found because of issues with the reanalysis. Could there be a false positive trend, or a false insignificant trend?

Please find my comments in the attachment.

Please find attached my comments