the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Jul 2026
Multi-scale dynamics and mesoscale drivers of the catastrophic September 2022 Marche (central Italy) flood
Abstract. The catastrophic flood that hit the Marche region (central Italy) on 15 September 2022 has been analyzed using high-resolution Weather Research and Forecasting (WRF) model simulations. The convective rainfall event responsible for the flood was favored by the passage of warm, moist air masses that interacted with the Central Apennines. The convective cells, triggered by the rough orography of the area, formed quasi-stationary bands that caused high rainfall accumulation in limited areas. The formation of these rainbands was sustained by the interaction between the low-level flow and the local topography, which generated a persistent low-level convergence line extending downwind from Mount Amiata northeastward towards the Apennines. The observed vertical profile upstream of the orography reveals that the instability was initially suppressed by an inversion layer and was released only when the humid air arrived at low levels and eroded the inhibition. Although the model somewhat underestimates a peak of rainfall (simulating 160–180 mm versus over 400 mm observed), it correctly reproduces the position of the storm and its evolution, confirming the key role of the orography in anchoring the system and making it quasi-stationary. Furthermore, the simulations suggest that gravity waves generated by the mountains helped to sustain the vertical motion of the impinging air, and highlight the role of upper-level dynamics (passage of a jet streak). Sensitivity experiments reveal a strong dependence on initial and boundary conditions. A comparison of the control run (forced with Global Data Assimilation System (GDAS) analysis/forecasts) with a European Centre for Medium-Range Weather Forecasts (ECMWF)-driven simulation shows a strong sensitivity in the low-level dynamics. Although the large-scale upper-level forcing is almost identical, the ECMWF-driven forecast fails to reproduce the correct amount and distribution of rainfall due to a weaker surface pressure gradient, reduced moisture advection, and lower instability at the time of the event. Specifically, the simulation forced with the ECMWF data anticipates the arrival of the moist air mass, determining a timing mismatch between the low-level supply of moist air (and the consequent release of instability) and the upper-level forcing. This lack of synchronization shifts the rainfall away from the observed area.
- Preprint
(40157 KB) - Metadata XML
-
Supplement
(1838 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-2573', Anonymous Referee #1, 16 Aug 2026
- AC1: 'Reply on RC1', Matteo Berton, 25 Sep 2026
-
RC2: 'Comment on egusphere-2026-2573', Anonymous Referee #2, 04 Sep 2026
Review of: "Multi-scale dynamics and mesoscale drivers of the catastrophic September 2022 Marche (central Italy) flood" by Berton et al.
Recommendation: Minor revisions
The manuscript presents a set of 666-m WRF simulations of the September 2022 Marche flood. It combines observations, an analysis of the simulated environment, topographic sensitivity experiments, and a comparison of several initial and boundary forcing datasets. The paper is well structured, the analysis is generally clear, and the focus on the interaction between low level flow and the Apennines is valuable.
I recommend minor revisions. The main changes concern the calibration of conclusions from a simulation that underestimates the rainfall peak, the attribution of the forcing sensitivity, and the evidence supporting the gravity wave interpretation. Addressing these points should strengthen an already interesting study without changing its central message.
General comments:
- The rainfall evaluation documents important limitations of the reference simulation. It produces about 160 to 180 mm where the observed maximum exceeds 400 mm, misses the later weaker rainfall phase, and delays the main phase by about 1 to 2 h. The closest agreement is obtained with the 10 by 10 km neighbourhood method, which extracts a maximum from the model field rather than comparing the simulated value at the gauge. Please distinguish this useful displacement tolerant diagnostic from a direct station evaluation, report simple aggregate errors for both approaches, and restructure the Abstract and Conclusions accordingly. In particular, a retrospectively selected GDAS based reconstruction cannot establish operational predictive skill or the early identification of hydrogeological risk. It can support a more limited statement about the simulated precursors and mechanisms of this case.
- The experiments convincingly show that the simulation is sensitive to the forcing dataset, but the GDAS and ECMWF configurations differ in more than the vertical resolution of the input fields. A comparison of 500 hPa geopotential height and theta-e alone does not prove that the large scale forcing is irrelevant to predictability, nor does it isolate the lower level pressure gradient as the cause of the rainfall displacement. Please present these results as evidence of sensitivity among the forcing datasets and soften causal statements. A concise quantitative comparison of the relevant temperature, moisture, wind, and pressure differences at the initial and boundary times would make the interpretation more transparent.
- The terrain reduction experiments support a contribution from the Apennines to the location and amplification of the simulated rainfall. However, the present evidence is less decisive for the stronger conclusion that gravity waves are generated and are essential to anchor the system. Alternating vertical motion in one cross section is consistent with a wave response, but it is not a complete diagnosis. The Froude number also uses a fixed 1000 m mountain height across a complex barrier. Please either add a direct wave diagnostic and assess the sensitivity of the Froude estimate to the chosen barrier height, or use more qualified language throughout the manuscript. A figure showing the spatial and temporal evolution of the Froude number would also help support this interpretation. The method used to modify the terrain and any associated changes to static fields should be described clearly.
Specific comments:
- The final namelists, preprocessing configuration, and postprocessing scripts should be archived in a public repository rather than being available only on request.
- The 12:00 UTC Pratica di Mare comparison shows simulated MUCAPE of 1731 J kg−1 against 3092 J kg−1 in the sounding. This is useful information, but it should be described as a substantial underestimation rather than qualitative agreement. A compact comparison of the humidity and wind profiles would help assess whether the model represents the ingredients relevant to the later convergence and moisture transport.
- The modified physics experiment changes the microphysics, convection, radiation, and land surface schemes simultaneously. It therefore shows sensitivity to this physics package rather than the contribution of any individual scheme. Please state this explicitly when discussing Fig. 14.
- For the Froude number calculation, please identify the precise upstream averaging area, the temporal sampling, and the treatment of terrain height. The statement that the flow was definitely able to surmount the barrier should be softened because this bulk index is only one approximation for such an irregular, nonuniform terrain profile.
Citation: https://doi.org/10.5194/egusphere-2026-2573-RC2 - AC2: 'Reply on RC2', Matteo Berton, 25 Sep 2026
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 236 | 117 | 40 | 393 | 56 | 22 | 21 |
- HTML: 236
- PDF: 117
- XML: 40
- Total: 393
- Supplement: 56
- BibTeX: 22
- EndNote: 21
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General Comments:
The manuscript provides a comprehensive diagnosis of the convective storm and flash flood that struck the Marche region on 15 September 2022, utilizing high-resolution WRF model simulations. The study effectively highlights the stationarity of the rainbands, the role of upper-level divergence combined with low-level moisture advection, and the extreme sensitivity of the results to initial/boundary forcing conditions.
The manuscript is well-written, clear, and logically structured. The authors do a good job of diagnosing why the model struggles—specifically noting the premature decay of the convective system after 17:00 UTC and the timing mismatch in certain boundary layer forcings. However, even under the optimal setup (GDAS forcing), the model severely underestimates total accumulated rainfall (~180 mm simulated vs >400 mm observed at Cantiano and Frontone). In addition, there is a key theoretical inconsistency regarding the flow regime discussion over the Apennines.
Major Comments:
1. Methodological Interpretation of Orographic Flow (Froude Number):
The discussion employing the classical Froude Number (Fr) to describe airflow dynamics over the Apennines is methodologically flawed. The standard criterion (Fr < 1 for flow splitting/blocking) is derived for isolated, three-dimensional topographic obstacles. For an extended, continuous 2D mountain barrier like the Central Apennines, lateral deflection is heavily constrained by the broad scale of the ridge. Consequently, stable air can frequently be forced over the barrier, rather than simple horizontal splitting. I recommend either removing the classical Froude number argument or revising the text to reflect the specific dynamics of elongated 2D ridges (referencing relevant literature on 2D mountain flow regimes).
2. Discussion on the Precipitation Underestimation and Convective Decay:
While the authors acknowledge the underestimation of the peak rainfall (180 mm simulated vs 419 mm observed at Cantiano), the manuscript would benefit from a deeper physical discussion on why the simulated storm decays prematurely after 17:00 UTC. Is this decay driven by excessive cold-pool propagation, insufficient surface latent heat fluxes, or microphysical parameterization limits? Expanding slightly on these potential physical mechanisms would provide greater insight into why NWP models systematically fail to reproduce the extreme tail (>400 mm) of such localized convective events.
Minor Comment:
To enhance readability and provide a clear overview for the reader, please include a summary table listing the maximum predicted 24-h precipitation across the different sensitivity experiments (NCEP GDAS, HRES ECMWF, NCEP GFS, ERA5) alongside the observed rain gauge maxima (e.g., Cantiano, Frontone).