VRTINEC·News

Event analyses

A Genoa low in two waves: rain in the west, snow down to Rateče and a November thunderstorm (16–19 Nov 2026)

Event analyses · 19 November 2026

  • padavine
  • ciklon

From Monday to Thursday a low formed over the Gulf of Genoa twice in a row. A moist south-westerly airflow soaked western and south-western Slovenia the most, a thunderstorm with graupel formed over the Karawanks on Wednesday afternoon, and on Thursday, as the low moved away, it turned cold enough for snow to reach the valley floor at Rateče.

The week began mild but moist. A broad upper trough lingered over the western Mediterranean, and on its forward side warm, moist air flowed towards the Alps. In such conditions a low over the Gulf of Genoa often forms repeatedly – this time twice. Each time the airflow over Slovenia strengthened and the rain spread again from west to east.

How it unfolded: two pulses of the Genoa low

Animation (GIF version): mean sea-level pressure, precipitation intensity and the track of the low from 15 Nov 19:00 to 20 Nov 07:00 CET. Animation: ECMWF IFS and ECMWF AIFS (open data, CC BY 4.0). Plotting and computation: VRTINEC.

The first low closed off over the Gulf of Genoa on Monday afternoon, 16 November. It did not travel far: for two days it circled between the Ligurian Sea and the Gulf of Genoa, slowly deepening, and was deepest on Wednesday morning. With the centre staying to the west, a south-westerly airflow blew over Slovenia the whole time. It forced moist air up against the Trnovski gozd, Nanos, Snežnik and the Julian Alps, so rain there was the heaviest and lasted the longest.

During Wednesday the first centre filled, but only briefly. In the evening a second low formed over the Gulf of Genoa, this time with a sharper upper trough pushing colder air from the north-west. The second low did not linger: on Thursday morning it slid across the northern Adriatic, was deepest over central Croatia in the afternoon, and then moved towards Hungary. Colder air followed from the north, and in the evening the rain retreated to the south-east.

Time (CET)Position of the centreCentral pressure
Mon 16 Nov 16:00western Ligurian Sea (44.25° N, 7.75° E)1009.3 hPa
Tue 17 Nov 04:00western Ligurian Sea (44.0° N, 7.75° E)1008.0 hPa
Tue 17 Nov 16:00western Ligurian Sea (44.0° N, 8.0° E)1002.3 hPa
Tue 17 Nov 19:00Gulf of Genoa (44.0° N, 8.75° E)1001.7 hPa
Wed 18 Nov 01:00western Ligurian Sea (44.25° N, 8.0° E)998.3 hPa
Wed 18 Nov 07:00Gulf of Genoa (44.0° N, 8.5° E)996.2 hPa
Wed 18 Nov 22:00Gulf of Genoa (44.75° N, 8.25° E)999.9 hPa
Thu 19 Nov 07:00western Ligurian Sea (43.5° N, 7.5° E)994.4 hPa
Thu 19 Nov 10:00Gulf of Genoa (44.0° N, 9.75° E)993.8 hPa
Thu 19 Nov 13:00northern Adriatic (44.75° N, 12.75° E)990.4 hPa
Thu 19 Nov 16:00central Croatia (45.25° N, 16.25° E)989.5 hPa
Thu 19 Nov 19:00western Hungary (46.0° N, 19.25° E)990.2 hPa
Fri 20 Nov 01:00north-western Romania (48.0° N, 24.0° E)990.0 hPa

Track of the low according to ECMWF (positions on a 0.25° grid, so approximate). Between Wednesday morning and evening there was no closed centre. Source: ECMWF IFS and AIFS (open data), computed by VRTINEC.

How much fell

Total precipitation 16 Nov 02:00 – 19 Nov 20:00 CET (90 h) at 89 automatic stations with at least 95 % data. Source: ARSO (automatic stations, 10-min data), processed by VRTINEC.

The pattern is familiar from similar autumn events: the most fell where the moist south-westerly airflow first met the mountain barriers. The wettest belt ran from Snežnik across the Javorniki and Nanos to the Trnovski gozd and the Idrija hills – the most at Sviščaki, followed by Zadlog and Otlica. Towards the north-east there was less and less, as the air behind the mountain barriers had already dried out; the least fell in Štajerska and in places in the north, in the lee of the mountains.

The twenty wettest stations, 16 Nov 02:00 – 19 Nov 20:00 CET. Source: ARSO (automatic stations, 10-min data), processed by VRTINEC.

The rain did not fall evenly. In the first part, up to Tuesday noon, there was only light rain, mostly in the west. In the second part, with the first low over the Gulf of Genoa, the south-westerly airflow strengthened and it rained persistently along the western barriers, much less inland. The most – more than two thirds of the total – fell on Thursday, as the second low crossed the northern Adriatic: rain was then heavy across the whole western half of the country, including the Logatec and Idrija areas and the Ljubljana basin.

PartPeriod (CET)Station meanMaximum
Part 116 Nov 02:00 – 17 Nov 13:001.6 mmSviščaki 14.0 mm
Part 217 Nov 13:00 – 19 Nov 01:008.1 mmSviščaki 54.1 mm
Part 319 Nov 01:00 – 20:0026.4 mmZadlog 52.5 mm
Total16 Nov 02:00 – 19 Nov 20:0036.1 mmSviščaki 113.7 mm

80 stations with at least 95 % data in all three parts that also have a model point. Source: ARSO (automatic stations, 10-min data), processed by VRTINEC.

Thursday: snow down to the valley floor at Rateče

As the second low moved off towards Hungary on Thursday, colder air followed from the north. In the upper Sava valley this coincided with the heaviest precipitation: heavy precipitation cooled the air further, because melting snowflakes and evaporation take up heat. The snow line therefore dropped quickly. At Rateče rain turned to snow around 14:30, the temperature fell to about 0 °C and the valley floor turned white. Most models kept the snow line a few hundred metres higher, so snow in the valley came as a surprise to many. Higher in the mountains more fell, about 19 cm at Vršič.

The thunderstorm over the Karawanks and graupel in Begunje

On Wednesday afternoon, 18 November, a distinct thunderstorm cell formed over the Karawanks and upper Gorenjska. Around 16:15 it was over Jesenice and Begunje. In the light of the setting sun the cumulonimbus anvil stood out sharply, with mammatus beneath it. In Begunje, graupel fell and briefly whitened the ground.

Cumulonimbus with anvil and mammatus, 18 November 2026. Photo: Robert Sitar.
Graupel in Begunje, 18 November 2026.

Why then, and why there? Ahead of a cold front, very warm and moist air still lay over the lowlands. In Črnomelj the temperature was around 20 °C, and dew points were around 10 °C. Aloft, however, the air was already much colder, so the atmosphere was unstable. A strong south-westerly airflow covered Slovenia, strengthening and turning to the right with height. This produced strong shear in the layer from the ground up to about six kilometres. Along the Karawanks the airflow met the ridge and was forced upward, which was decisive for the cell to form.

This is shown clearly by soundings from the C-LAEF 06 UTC run. In the valley itself, at Begunje, there was almost no buoyant energy near the ground at 15:00 and 16:00, and a strong cap as well: air from the valley floor could not rise on its own. On the Karawanks slope, about 1,600 metres above sea level above Begunje, there were 300 to 370 J/kg of buoyant energy and no cap. The cell could therefore be released over the ridge and then move down into the valley.

C-LAEF sounding for Begunje na Gorenjskem, 18 November 2026 at 16:00. Near the ground there is almost no buoyant energy, and the cap is strong. The hodograph turns to the right with height. Source: C-LAEF, 06 UTC run (GeoSphere Austria), plotted by VRTINEC.
C-LAEF sounding on the Karawanks slope above Begunje (about 1,600 m), 18 November 2026 at 16:00, at the point of maximum buoyant energy. There is no cap, so the updraught can be released already over the ridge. Source: C-LAEF, 06 UTC run (GeoSphere Austria), plotted by VRTINEC.

In the same model, shear in the 0–6 km layer at 16:00 was about 33 m/s in the valley and even stronger over the ridge, about 44 m/s. Storm-relative helicity (SRH) in the 0–3 km layer was about 190 m²/s² at Begunje and about 120 m²/s² on the slope. This is an environment in which supercells are possible. That does not confirm that the cell itself was rotating; a radar signature of rotation would be needed for that. The figures are also only an approximation, because this C-LAEF dataset has only eleven pressure levels.

Among the models, only C-LAEF in the 06 UTC run captured this thunderstorm reliably. D2-RUC and ICON-CH1 caught it only in part.

Which model forecast the precipitation best

Model precipitation forecasts compared with measurements at 80 stations, sum of three parts. Source: ARSO (10-min data), model data and computation by VRTINEC.

Because the event lasted almost four days, we split it into three parts. Each part is computed from a single run starting at the beginning of the part; the forecasts of all three parts were then summed and compared with the measured total at each station:

  • Part 1 (16 Nov 02:00 – 17 Nov 13:00 CET, 35 h): runs of 16 Nov 00 UTC,
  • Part 2 (17 Nov 13:00 – 19 Nov 01:00 CET, 36 h): runs of 17 Nov 12 UTC,
  • Part 3 (19 Nov 01:00 – 20:00 CET): runs of 19 Nov 00 UTC.

ICON-CH1 has short runs, so each part is stitched from two consecutive runs (details in the notes on the graphic). MAE is the mean absolute difference of the total per station; the bias shows whether a model over- or underestimated the precipitation on average.

Among the high-resolution models, HARMONIE DINI came closest overall, followed by ICON-D2 and C-LAEF. Among the global models, ICON-EU was best, and this time it was even slightly better than all high-resolution models. Most high-resolution models underestimated the precipitation; only C-LAEF and ALADIN-SI overestimated it. The largest differences were at stations along the western barriers, where the most fell: models that did not strengthen the lifting of air along the Trnovski gozd, Nanos and Snežnik enough missed there by several tens of millimetres.

By part: C-LAEF was closest in parts 1 and 2, and HARMONIE DINI in part 3. In part 3, with the second low and the heaviest rain, the models diverged the most: ALADIN-SI, ECMWF IFS and GFS overestimated the rain, while ICON-CH1 and ICON-2I underestimated it.

Season ranking: the rain of 16–19 November is the fifth event in the live verification season ranking, where models are ranked by their mean error (MAE) over all events of the season. There is no change at the top, but lower down the order shifted somewhat. Among the high-resolution models, C-LAEF stays on top (mean MAE 10.9 mm), followed by HARMONIE DINI (12.3 mm), ALADIN-SI (12.8 mm) and ICON-CH1 (13.2 mm). Moving up: HARMONIE DINI (4 → 2) and ICON-D2 (7 → 6). Moving down: ALADIN-SI (2 → 3) and ICON-CH1 (3 → 4). Among the global models, ICON-EU stays on top (mean MAE 11.8 mm), followed by ECMWF AIFS (13.4 mm), global ICON (13.4 mm) and ECMWF IFS (14.2 mm). Here the order did not change. ICON-D2-RUC with its own short runs is still shown only for comparison and is not ranked.

Sources: ARSO automatic station measurements; models WRF 2 km (VRTINEC), ALADIN-SI, C-LAEF, ICON-2I, ICON-CH1/CH2, ICON-D2, HARMONIE DINI, ICON-EU; ECMWF IFS, ECMWF AIFS, GFS and ICON via Open-Meteo; C-LAEF soundings, 06 UTC run (GeoSphere Austria), plotted by VRTINEC; animation ECMWF IFS and ECMWF AIFS (open data, CC BY 4.0), plotting and computation VRTINEC; cloud photograph: Robert Sitar; photograph of graupel in Begunje: author unknown.

Gregor Skok