Simulation of a flash-flood event over the Adriatic Sea with a high-resolution atmosphere-ocean-wave coupled system

Abstract

On the morning of September 26, 2007, a heavy precipitation event (HPE) affected the Venice lagoon and the neighbouring coastal zone of the Adriatic Sea, with 6-h accumulated rainfall summing up to about 360 mm in the area between the Venetian mainland, Padua and Chioggia. The event was triggered and maintained by the uplift over a convergence line between northeasterly flow from the Alps and southeasterly winds from the Adriatic Sea. Hindcast modelling experiments, using standalone atmospheric models, failed to capture the spatial distribution, maximum intensity and timing of the HPE. Here we analyze the event by means of an atmosphere-wave-ocean coupled numerical approach. The combined use of convection permitting models with grid spacing of 1 km, high-resolution sea surface temperature (SST) fields, and the consistent treatment of marine boundary layer fluxes in all the numerical model components are crucial to provide a realistic simulation of the event. Inaccurate representations of the SST affect the wind magnitude and, through this, the intensity, location and time evolution of the convergence zone, thus affecting the HPE prediction.

Figure 1

Panels (a–i) 6 h accumulated precipitation observed in each ARPAV station (coloured dots), together with simulated precipitation (mm) as resulting from the different runs. Plots refer to the period 06–12 UTC, 2007 September, 26: max cumul is the sum of maximum hourly values associated with the storm cell; m.dist is the average distance (in km) between the storm location and Campagna Lupia (where the most intense rainfall was observed and highlighted with “OBS” in panel (a); panel l: Taylor diagram of the timeseries of hourly rainfall maximum associated with the storm cell. Panel m shows the numerical domains, topography and satellite SST (resolution 8.3 km) at 06 UTC.

Figure 2

Upper row: Sea Surface Temperature (SST) difference between the FLAT run and the other simulations (in °C) at 06 UTC, September 26. Middle row: Surface Heat Fluxes difference between the FLAT run and the other simulations (in W m⁻²). Bottom row: 10 m wind speed difference between the FLAT run and the other simulations over sea surface. Due to the marked similarity in terms of SST with the AO run, the results for AOW run are not shown here (see Fig. 3).

Figure 3

(a) Difference in surface roughness, (c) in SST, (d) in surface heat fluxes, (e) in 10 m wind speed between run AO and run AOW, (b) significant height of the wave (shaded) and the peak length of the wave fields (contours) in AOW run, (f) correlation between heat fluxes and surface roughness, (g) between heat fluxes and surface roughness, (h) between heat fluxes and 10 m wind, (i) between heat fluxes and SST.

Figure 4

Convergence line (0.003 s⁻¹ isoline) location at 05, 08, 10, 12, 15 UTC on September 26, for run RTG (a), ROMS (b), AO (c), and AOW (d).

Figure 5

Upper row: (a) 2-D radar reflectivity data (Vertical Maximum Intensity) in the HPE region; synthetic radar data simulated in (b) RTG, (c) ROMS, (d) AO and (e) AOW; (f) cross section of the storm radar reflectivity acquired by the radar in Teolo at 06 UTC, 26 September 2007; (g)–(l) reflectivity vertical cross sections along the axes of most intense convection (whose locations are in panels (b)–(e) for (g) RTG, (h) ROMS, (i) AO and (l) AOW.