The Science of Acqua Alta: Why Venice Floods

What causes acqua alta in Venice?

Acqua alta is a compound event. The contributions that add up to a high water level are meteorological surges, seiches, tides, seasonal to decadal sea level variability, and long-term relative sea level change (Lionello et al., 2021, p.2635). No single one of them floods the city on its own at the levels visitors remember.

The meteorological part comes in several timescales: surges produced by planetary atmospheric waves lasting 10 to 100 days, storm surges from mid-latitude cyclones lasting a few days, meteotsunamis, and surges from mesoscale systems lasting a few hours (Lionello et al., 2021, p.2635). Two winds matter. Sirocco, a strong southeasterly, pushes water up the long axis of the Adriatic and raises the level across the whole lagoon. Bora, a cold northeasterly, produces a more localised set-up in the southern lagoon near Chioggia (Michielotto et al., 2026, p.846).

Geography does the rest. The Adriatic is a semi-enclosed, channel-shaped basin with an average depth of 126 metres, and it is a hotspot for extreme sea level events in the Mediterranean (Goglio et al., 2026, p.1). Venice sits at the closed end of it, and the margins are small: the lowest part of St Mark's Square is about 55 centimetres above present mean sea level, while the astronomical tide alone has an amplitude of roughly 50 centimetres (Lionello et al., 2021, p.2634).

How much of a flood is storm and how much is tide?

For the 12 November 2019 event the contributions have been separated centimetre by centimetre. Of the 175 centimetre hourly mean peak, about 80 centimetres is attributed to the surge and high-frequency contributions and 32 centimetres to tides and seiche oscillations (Goglio et al., 2026, p.7). Acqua alta at that scale is therefore mostly weather.

A seiche is a standing oscillation of the whole basin, and in the Adriatic it resonates with the tide: the first two oscillatory eigenmodes have periods close to those of the dominant tidal constituents, the diurnal K1 and the semidiurnal M2 and S2 (Goglio et al., 2026, p.3). When a surge, a high tide and an existing seiche line up, the total is far larger than any of them alone.

In November 2019 the tide was actually working against the event. The peak fell during the lower of the day's two high tides, with a tidal amplitude about 25 centimetres below the higher high tide a few hours earlier (Goglio et al., 2026, p.6). The same storm on the stronger tide would have been worse.

A third, smaller mechanism has now been measured. Baroclinic processes, meaning internal waves and a bottom-intensified density front centred near 44.8 degrees north, contributed a net 7 centimetres, about 4 per cent of the peak (Goglio et al., 2026, pp.1, 15).

What happened on 12 November 2019

A deep low-pressure system over the central and southern Tyrrhenian Sea generated a strong sirocco along the main axis of the Adriatic, and a smaller but intense local cyclone reached Venice at 20:30 UTC (Goglio et al., 2026, p.6). The sea level exceeded the 140 centimetre extreme flood threshold from about 18:30 to 22:00 UTC and peaked around 21:00 UTC (Goglio et al., 2026, p.5). The whole event was a single evening.

Different sources give different peak heights because they use different gauges. At the offshore Institute of Marine Sciences platform the level reached 182 centimetres, corresponding to a 175 centimetre hourly mean (Goglio et al., 2026, p.2). The in-city tide gauge record gives 189 centimetres, the second highest water height ever measured in Venice, behind the 194 centimetres of 4 November 1966 (Lionello et al., 2021, p.2634).

The flood covered 85 to 90 per cent of the historical city centre, and damage was estimated at 460 million euros (Goglio et al., 2026, p.2). It also landed on an already raised sea. Mean sea level over the whole of November 2019 was 67 centimetres, twice the annual mean of 34 centimetres, and the mean across 10 to 12 November reached 76 centimetres (Goglio et al., 2026, p.5). That month was the worst in the local tide gauge record, with 15 events above 110 centimetres and four above 140 centimetres (Lionello et al., 2021, p.2634).

What the acqua alta levels mean: 110, 140 and 194 centimetres

Venetian water heights are quoted above a local datum, the zero at Punta della Salute, and a few tens of centimetres change the day completely.

At 110 centimetres above that datum, flooding affects approximately 12 per cent of the city area, which the city authorities consider acceptable and handle with temporary measures such as raised footbridges (Michielotto et al., 2026, p.848). This is also the level at which the MOSE barriers are meant to be raised.

At 140 centimetres the event is classified as an extreme flood by Venice Municipality, and about 60 per cent of the historic centre goes under (Goglio et al., 2026, pp.1, 2). At the 175 centimetre hourly mean of 12 November 2019 the figure was 85 to 90 per cent (Goglio et al., 2026, p.2).

The record remains 4 November 1966, which reached 194 centimetres and stayed above 110 centimetres for 22 consecutive hours (Lionello et al., 2021, p.2634). Duration is the part the numbers hide: a peak passes in an hour or two, while 1966 kept the city flooded for most of a day, and it was the event that led to the Special Law for Venice of 1973 and eventually to MOSE.

How often does Venice flood, and is it getting worse?

Flooding has become much more frequent, and the reason is the sea level the tides start from rather than stormier weather. Water height maxima above 120 centimetres increased from fewer than two events per decade in the first half of the twentieth century to 40 events in 2010 to 2019, and at the lower 110 centimetre threshold from 4.2 to 95 events per decade (Lionello et al., 2021, p.2638). Counted another way, 52 high-water events exceeded 110 centimetres between 2000 and 2009, and 95 between 2010 and 2019 (Michielotto et al., 2026, p.848).

The increase is attributed to relative sea level rise, with no robust evidence that the meteorological conditions behind extreme water heights have intensified (Lionello et al., 2021, p.2638). Land subsidence supplied about half of the long-term 2.5 millimetres a year, and accelerated the rate to around 5 millimetres a year between 1950 and 1970, when groundwater was being pumped from beneath the lagoon; regulation has largely controlled that cause since (Lionello et al., 2021, pp.2634, 2638). Recent rates are higher again: 4.9 plus or minus 0.3 millimetres a year for 1991 to 2020 (Ferrarin et al., 2024, p.7) and about 4.4 plus or minus 1.5 millimetres a year in current estimates (Michielotto et al., 2026, p.848).

The season is narrow. Acqua alta events typically occur in autumn (Goglio et al., 2026, p.1), and barrier activations usually fall between October and February, with exceptions such as May and late August in 2023 (Michielotto et al., 2026, p.848).

How accurate is the acqua alta forecast?

Well enough to act on, and not well enough to relax. Forecast systems aim at a three-day range because several days are needed for timely management of the MOSE system (Goglio et al., 2026, p.4). Re-forecasts of the 2019 event initialised on 10 November predicted it three days ahead with errors of 18 and 25 centimetres against the observed 175 centimetre peak, about 10 and 14 per cent (Goglio et al., 2026, p.5).

At the time the operational systems did worse. All available forecasts underestimated the maximum height of the 12 November 2019 compound event by up to 45 centimetres, and on 8 December 2020 the city flooded with MOSE available but not operated, because the forecast underestimated the water level (Lionello et al., 2021, p.2638).

The weakness is the weather input rather than the sea model. For 2019 the modelled wind speed was almost 10 metres per second below scatterometer observations, and pressure at the offshore tower was about 5 hPa lower than the weather model field (Goglio et al., 2026, p.6). How often the weather is sampled matters more than how fine the grid is: hourly forcing reproduced a 157 centimetre peak, while six-hourly forcing dropped the modelled peak to 136 centimetres (Goglio et al., 2026, pp.6, 18).

Errors run in both directions. Predicted maxima issued before closures have often substantially exceeded the levels actually observed in the city (Michielotto et al., 2026, p.848).

What MOSE can and cannot do

MOSE is a set of 78 hollow flap gates across the three lagoon inlets. They rest flush with the seabed and are raised by pumping air in when levels are predicted to exceed 1.10 metres above the local datum (Michielotto et al., 2026, p.848). In practice they have to be closed in advance, generally when the level in Venice is between 60 and 80 centimetres depending on the weather (Ferrarin et al., 2024, p.4).

It works. On 3 October 2020, the first closure under real hazard conditions, the peak outside was 130 centimetres while the lagoon and St Mark's Square were held at 70 centimetres (Lionello et al., 2021, p.2634). Since then the system has been activated more than 150 times, averaging 8.6 hours per activation and about 146 hours a year, a frequency driven partly by forecast uncertainty and precautionary management (Michielotto et al., 2026, pp.848, 850).

The limits are arithmetic. Closures of three weeks a year are unlikely before the 2040s and virtually certain before 2100 even under a low-emission scenario, while six-month closures are likely before 2100 under high emissions, against projected relative sea level rise of 30 to 110 centimetres by then (Lionello et al., 2021, p.2638). Isolation has a cost: closure hours could rise fourfold, cutting water exchange through the inlets by more than 40 per cent in winter months (Ferrarin et al., 2024, pp.11, 15). An optimised activation rule could have cut closures by roughly 30 per cent (Michielotto et al., 2026, p.851).