The authors offer a well-presented manuscript examining the ability of a convection-permitting (CP) model (ERA-Interim-driven COSMO-crCLM) to represent the reverse orographic effect at the northeastern Italian Alps area. The manuscript is well-written, and concise, with a good flow and sufficient discussion. The contribution of the manuscript is significant since it gives answers to an issue which may arise for many researchers dealing with CP models.

Every query or suggestion I had during the first part of the manuscript was explained or applied in the next sections, therefore I only have a few minor suggestions, mainly grammatical-syntax comments and typos.

Some minor/discussion comments:

It would be helpful to see a short literature review on existing CPM permitting models (probably in the Introduction), and comments on their performance. This would help you justify better the selection of the ERA-Interim-driven COSMO-crCLM.

Lines 17-19: “We introduce the use of a non-asymptotic statistical approach (Simplified Metastatistical Extreme Value, SMEV) for the analysis of extremes from short time slices such as the ones of CPM simulations” The word “introduce” is a bit misleading, since SMEV has already been introduced; maybe rephrase it to “We propose” or something like this?

Lines 55-57: “Over the Alps, but also elsewhere, CPMs tend to generate more precipitation at higher elevations than in reality, thus reducing the bias with respect to observations compared to RCMs (Lind et al. 2016, Reder et al. 2020).” This sentence is confusing to me, it sounds like CPMs overestimate precipitation at higher elevations than in reality, but at the same time, they reduce the bias compared to RCMs. Could you rewrite this?

Lines 143-144: “We considered only rain gauges with at least 9 valid years during the period 2000-2009,” Could you explain here why you chose this period?

Lines 153-154: “More details on the used physical parameterisations can be found in Leutwyler et al. (2016).” Give two-three sentences on the basics of the process.

Some suggested syntax changes:

Line 25: “SMEV’s capability”

Line 26: “promises further applications”

Line 45: “In CMPs,”

Lines 51-53: “In areas with a complex terrain, the possibility of explicitly resolving convection along with a more detailed representation of orography and surface properties are crucial elements for correctly capturing the initiation and development of convection”

Line 269: Do you mean “A spatial pattern” instead of “organization”?

Lines 361-363: I think a verb like “show” is missing from that sentence: “The consistency of the return level estimates obtained from the full record and from the 10 yr record, and the small increase in the associated uncertainty show that, once its assumptions are verified, SMEV is a reliable statistical method for the analysis of extreme precipitation from short time slices.”

Line 415: “n” in italics

Line 480: “100 yr, and parameters of…”

Figure comments:

Figure 2, Figure 4 and rest of the figures showing linear regression: do you want to also show the coefficient of determination R2?

Figure 4: “(SC_CPM), and all CPM”

Figure 4: “the linear regressions lines shown as a solid line, are expressed as..”

Figure 4: Could you change color for the observations, it is the same as CPM

Figure 4: You do not focus on the orographic effect for daily but still can show the slope for the 24-hour case

Figure 7: remove “,” from: “grid, CPM”

Figure 7: “are significant” instead of “result significant;”

The manuscript provides an assessment of whether a high-resolution (2.2 km) regional climate model (COSMO-crCLM) can represent the "reverse orographic effect" - an observed effect whereby short-duration (e.g., hourly) extremes in precipitation decrease with increasing altitude. The study is performed over northeastern Italy including some of the Italian Alps - a region of complex orography. The authors make use of a statistical technique: simplified metastatistical extreme value (SMEV) to reliably determine return times longer than the duration of their simulations (10 yr). They find that the model does produce a reverse orographic effect, albeit with reduced magnitude compared to rain gauge observations.

The manuscript is well written, with clear figures and a logical structure. The interpretation is all backed up by the figures. Uncertainties are assessed using a boostrapping method (random sampling of years with replacement). Potential sources of bias, from underestimation of the rain gauges to differences in the elevations of the gauges and the co-located grid cells, are considered. The discussion further considers why the model might be underestimating the effect, discussing the effective resolution of the model as well as the representation of subgrid processes such as turbulence. All of the points that arose as I was reading the manuscript were addressed in the discussion.

I have therefore recommended publication with technical changes only. Below are some very minor comments and clarifications.

General: 

• divide symbol (÷) seems to be used where a dash should be used.
• slice/s - I would normally prefer the word period/s
• Fig. vs Figure, consistency.
• I believe collocate should be colocate or co-locate. 

Specific:

• Title: suggest changing to "How well does a convection-permitting regional climate model represent the reverse orographic effect of extreme hourly precipitation?"
• L14: Recent observational studies...
• L21: northeastern (no capital)
• L45: become -> are becoming
• L45: CMPs -> CPMs
• L77: spell out what the "reverse orographic effect" is: i.e. short-duration extremes decrease with increasing altitude (or similar).
• L152: GPU -> GPUs
• L153: More details on the physical parameterisations...
• L153-4: Also, it would not hurt to describe some of the key parametrizations here, such as the microphysics (1-moment or 2), and the turbulence paremtrization, so that the reader does not have to dig into the references. 2-3 sentences perhaps on these.
• Fig. 1b: Cannot see OB at low elevations. Perhaps plot as blue line (staircase - i.e. with flat tops/vertical lines) on top of SC_CPM?
• L218: left-censoring - what does this mean?
• L244: AM defined as Annual Maxima, i.e. already plural. This means AMs should probably not be used later. (Very pedantic.)
• L274: 30% km-1 - please remind reader of the definition without having to refer back to Sect. 3.3.
• L275: Add details of regression R^2/fmse?
• Fig. 2b: suggest adding a vertical dashed line at elevation = 100 m, or showing all points below this as open circles, to visually show they are not included in the regression.
• L284: Figure 3a, b) instead of c)?
• L286: (b) and (d)
• L305: missing full stop.
• Figs. 4, 6: should have a blue triangle in key for SC_CPM (looks green on zooming in to PDF for me)
• L348: "The slopes test significantly different at the 5% level." Clarify - slopes of what are different to what?
• L355: Figure S3 reports the uncertainty in the observed 1 h duration...
• L362: ...uncertainty indicate that...
• L382: it seems to me as if GR_CPM shows less dependence on return period than SC_CPM is worth saying something about. Perhaps one sentence saying this?
• L392: ...found in a CPM.
• L396: , with the lower values of the interquartile ranges...
• L398: ...would be double that of the observations...
• Sect. 5.3: suggest renaming this to "Bias assessment of differences in CPM and rain gauge elevations" (as a short subsection), and moving 5.3 to just before L430.
• L433: undercath (typo)
• L472: ...estimation of hourly return levels...
• L487: ...in the case of strong wind.

The manuscript assesses the representation of the extreme precipitation by the convection-permitting-scale dynamically downscaled regional climate model COSMO-CLM, driven by ERA-Interim reanalysis. The focus of the study lies on the reproducibility of the "reverse orographic effect" consisting of a decrease of short-duration extreme precipitation with the increase of the elevation in the complex-orography context of northeastern Italy. To limit drawbacks in terms of underrepresented climate variability within the short temporal segment of 10 years considered and related large uncertainty on the estimation return period longer than the available period, Authors take advantage of the Simplified Metastatistical Extreme Value (SMEV). This approach relies on the assumption that a suitable statistical model describing the ordinary events may be identified and related distribution can be used to define the distribution of yearly maxima and to capture the probability of occurrence of extremes. Uncertainties are characterized using a bootstrapping method and sources of bias from rain gauges located at different elevations have been taken into account.

The manuscript is overall well written, and results regarding the adoption of the statistical SMEV approach and physical mechanisms behind the presented results (i.e., subgrid processes behind the model underestimation of the reverse orographic effect) have been comprehensively presented and properly discussed.

It follows only minor comments.

Line 45: CPMs instead of CMPs.

Line 60: Please correct the doubled reference.

Lines 205-207: This statement is not supported by evidence. Please provide a real demonstration through some plots in defense of the applicability of Weibull tail approximation to the right distribution tail of ordinary events of your datasets.

Lines 214-215: Please better introduce this section.

Line 252:: Correct the numbering of the section.

Line 348: "The slopes test significantly different at the 5% level." Not clear what “different” is referring to.  

Why place subsections 5.1 and 5.2 in the discussion section and not in the results section?

Figures 4 and 6 it quite challenging distinguishing green and blue markers.