
Urban flash flood in Zagreb
23 July 2020 20:00 UTC–25 July 00:00 UTC


Extreme rainfall on the evening of 24 July 2020 left the Croatia capital Zagreb under water.
08 December 2022
23 July 2020
By Natasa Strelec Mahovic (EUMETSAT), Tanja Renko, Izidor Pelajic and Tatjana Vujnovic (DHMZ)
In a short period of just a few hours on the evening of 24 July 2020, 60 to 90mm of rain fell in the city of Zagreb, causing severe urban flooding. At Zagreb-Gric meteorological station, located in the very centre of the city, two-hourly precipitation of 77mm was recorded. The return period for that kind of intensity at the station is more than 400 years.
Flooding caused damages at many facilities, including the children's hospital, already previously badly damaged in the catastrophic earthquake in March, and one firefighter died.
The situation was well forecast by various NWP models and, using satellite, lightning and radar data, forecasters from Croatian Meteorological and Hydrological Service (DHMZ) were able to issue timely warnings for thunderstorms, rain and flash floods for north-western parts of the country (Figure 1). Hydrological warnings were also issued, since hydrological models forecast high probability of flash flooding.
Synoptic situation
The infrared images in Figure 2 follow the severe convective storms that developed on 23 July and overnight into 24 July, first over north Italy where they brought large hail and strong winds in many places.
On 24 July storms started to develop on the northern Adriatic coast, with hail recorded over Istria peninsula . Later very severe storm development, triggered by the Dinaric Alps, caused heavy rainfall, hail and severe lightning over Bosnia and Herzegovia, with flash flooding and severe wind damage in the city of Tuzla. The severity of the storms was indicated by many overshooting tops, as well as cold-ring and cold-U/V features occurring on top of the storms.
Synoptic forcing was provided by an upper-level trough passing over the Alps on the afternoon of 24 July. The trough moved eastwards, as can be seen in the animation in Figure 3.
Looking at the wind at 700hPa one can notice that the wind field suggests a presence of a small-scale cyclone, moving ahead of the trough. Notice also a small short-wave trough forming at 500hPa and passing over north west Croatia, which, after aligned with the 700hPa cyclone, gave an additional synoptic 'push' for the storm development.
The ingredients for deep moist convection, moisture, instability, lifting mechanism and wind shear, were all in place and the setup was supportive for the storm development. Moderate to high values of CAPE were present all over the area, mostly with very little or negligible convective inhibition, as seen in the Sqew-T log-P diagram of Zagreb radiosounding at 12:00 UTC (Figure 4).
Lift was very well supported by the approaching trough, with topography of the Alps and the Dinaric mountains playing an important role. Due to a southwesterly flow of moist and very unstable air towards the Dinaric Alps, divergence in the upper levels and significant cyclonic vorticity advection, vertical motion was quite strong, helping the storms to consume available energy. Deep layer shear (between 10 and 20m/s) helped the storms organise, especially during the afternoon over Bosnia and Herzegovina, and later on over continental Croatia.
Storm over Zagreb
The initiation of the storm over Zagreb happened quite late in the day, when the forecasters thought the warning level they issued might have been too high and the precipitation signal overestimated. This delay in the development may have been caused by the anvils of the storms over the north Adriatic, which cut the insolation and slowed the initiation. However, with the arrival of the trough new cells started to develop behind a more intensive and damaging storm system over northwest Croatia. Comparison of the satellite and radar depiction of the storm can be seen in Figure 5.
Image comparison


Figure 5: Comparison of storm appearance in IR10.8 and radar image. Satellite image on 24 July at 19.30 UTC, radar image from Bilogora, Croatia radar at 19:45 UTC to account for scanning time difference. The difference in the location of maximum signal is caused by the parallax shift of the satellite pixels.
Despite rather cold cloud tops in the satellite image, the radar signal was not showing very high values, probably due to a low, and predominantly liquid, cell core. This can be verified by the fact that there was not much lightning and no hail was recorded, just extreme precipitation.
The precipitation intensity could also be followed in NWC SAF products. The animation in Figure 6 shows the Convective Rainfall Rate based on Cloud Physical Properties (CRR-Ph) product. Precipitation rates are well depicted, notice that for the storm over Zagreb the product shows the highest values at 19:30 and 19:45 UTC (comparable to the highest radar signal, at the time of the strongest showers).
Additional content
Urbana poplava u gradu Zagrebu tijekom noći s 24. na 25. srpnja 2020 (DHMZ, in Croatian)
Urban flash flood in Zagreb Croatia, July 2020 (Emergency Management)
GRANDINATA Hailstorm in Arma di Taggia, Italy - July 24, 2020 | Arma di Taggia venerdi 24 luglio (Tetova News/YouTube)
Maltempo a Milano | Thunderstorm in Milan, Italy - July 24, 2020 (Tetova News/YouTube)
Nevrijeme u Tuzli: Ulice pod vodom, poplavljeni automobili, polomljeno drveće... 24 07 2020 (RTVTK/YouTube)
Vehicles submerged as Croatian capital hit by heavy flooding (Pozitive Idea/YouTube)
VIDEOS: Zagreb Battles Big Floods after Storm Rips through Croatian Capital (Total Croatia News)
VIDEO: Big floods in Zagreb as storm hits (Croatia Week)