The essentials in thirty seconds
One sentence comes back after every flood: “the département was on red alert, so it is normal”. It proves nothing: a colour on a map does not say how much water fell on a plot, nor whether the structure should have withstood it.
- A warning is a forecast, not a measurement. It rates a hazard expected over a whole département for the next 24 hours. It says nothing about the rain that actually fell on your roof.
- Counter-examples abound. On 3 October 2015, Cannes received rain of more than 100-year return period over one and two hours under an orange warning. In October 2023, the red warning for storm Aline ended in damage clearly lighter than in 2020. In September 2024, Cannes flooded under a mere yellow warning.
- The right question is the return period. Compare the rain measured closest to the site, over the critical duration of the structure, with the rain it was designed for.
- The law does not settle for a colour. A natural-disaster decree presupposes, for floods, a return period of at least ten years. By itself it does not establish the force majeure that exonerates a builder.
The basement is flooded, the flat roof overflowed, the car park took a metre of water, and the sentence arrives at the managing agent, at the contractor, sometimes in a report. The only question that settles it is the return period: does this rain, here, come back every two years or every hundred? This article is for experts, managing agents, project owners and owners who must write or challenge a conclusion after a flood. We separate what a source establishes, what can be calculated and what remains an order of magnitude.
A colour is not a measurement
Weather warnings are issued by Météo-France before the event. They announce, they measure nothing. Mistaking one for a finding is like taking yesterday’s bulletin for tomorrow’s reading. Its principle, set out on Météo-France’s rain-flood page (in French), comes down to three points: four colours, assigned by département, for a hazard expected within 24 hours. Each one disqualifies the colour as evidence in a claim file.
A forecast, hence uncertainty
The colour is decided from models, several hours before the rain. A Mediterranean storm can shift, intensify or fizzle out. The colour says what forecasters feared, not what happened.
A département, not a plot
The Alpes-Maritimes cover about 4,300 km², from sea level to over 3,000 metres. One colour covers both the port of Antibes and the upper Tinée valley. In a single episode, a valley can receive far more water than a coastal commune.
A 24-hour total, not an intensity
Météo-France’s reference points are expressed as daily totals: as a guide, for a lowland département, orange corresponds to 40 to 80 mm in 24 hours over a significant part of the territory or an area at stake. These points apply to lowland areas and do not transfer as they stand to the Alpes-Maritimes. Above all, a gutter, a gully or a downpipe overflows within minutes: the day’s total says nothing about the downpour that saturated them.
A hazard level, not a rarity
The colour rates a hazard to people and property, not the statistical rarity of the rain. The four episodes below show it: an orange warning accompanied rain of more than 100-year return period, a red warning was followed by limited damage, a yellow warning preceded a flooded town.
- An orange or even yellow warning can accompany locally exceptional rain.
- A red warning covers a whole département, including communes that are barely affected.
One last point of vocabulary: a warning (vigilance) is not an alert. The alert is the decision of the authorities, prefect or mayor, to make the population act. Writing “red alert” in a report already mixes two systems. No colour says how much water fell on the structure, and without that measurement there is no conclusion.
The return period: the only question that matters
The return period of a rainfall measures its rarity at a given place. A 10-year rain has, each year, a one in ten chance of being reached or exceeded. A 100-year rain, one in a hundred. This is the definition used by risk prevention plans for the 100-year flood.
Three misreadings to avoid
- “10-year” does not mean “once every ten years”. It is an average over a very long period. Two 10-year rains can fall in the same year, then none for twenty-five years.
- “100-year” does not mean “unforeseeable”. Statistics foresee it: its depth is known before it falls, which is why one can design against it.
- The return period is local. It is calculated for a station, from its own series of measurements. A 10-year rain in Lille would be an ordinary rain in Nice.
The probability over the life of the structure
The probability of suffering at least once a rain of return period T during n years is:
| Return period | Over 10 years | Over 30 years | Over 50 years |
|---|---|---|---|
| 10 years | 65% | 96% | 99.5% |
| 30 years | 29% | 64% | 82% |
| 50 years | 18% | 45% | 64% |
| 100 years | 10% | 26% | 39% |
The first row deserves a pause: during the ten years of the ten-year warranty, a building has almost two chances in three of receiving a 10-year rain. And over a fifty-year life it has more than one chance in three of meeting a 100-year rain. An event of that rank is no freak accident: it is a service load.
Read the other way round: after how many years does a building have an even chance of having already suffered a rain of return period T? The answer is about 0.69 times T:
| Return period | Time for an even chance |
|---|---|
| 10 years | about 7 years |
| 30 years | about 20 years |
| 50 years | about 34 years |
| 100 years | about 69 years |
A 100-year rain therefore has an even chance of occurring within 69 years: on the scale of a building’s life, it is not an event outside time.
An estimate, with its margin
A return period is not an exact number: it is extrapolated from a series of measurements that rarely exceeds a few decades, and uncertainty grows with rarity. A sound report therefore gives a range, for example “between 20 and 50 years”, not a figure to the nearest year. This margin takes nothing away from the method: between a 2-year rain and a 100-year rain the gap remains unambiguous, and it is the gap that decides a file.
One rain, several return periods
The same rain does not have one return period, it has one per duration. It can be ordinary over 24 hours and exceptional over 15 minutes, or the reverse. Saying “the rain was a 10-year event” without the duration is an incomplete sentence.
The Montana formula
Météo-France links rainfall depth to duration with the Montana formula, fitted station by station and for each return period:
h is the rainfall depth in millimetres, t the duration in minutes, a and b the station’s coefficients for the chosen return period. These coefficients are valid only over a range of durations: a sheet established for short durations is not extrapolated to 12 hours.
Each structure has its critical duration
A structure overflows when rain exceeds its capacity for as long as the water takes to fill it. This time, close to the hydrologists’ time of concentration, gives the rain duration to examine. Orders of magnitude:
| Structure | Rain duration to examine |
|---|---|
| Gutter, box gutter, outlet, downpipe | A few minutes |
| Flat roof, yard gully, channel | 5 to 15 minutes |
| Buried estate network, car park ramp | 15 to 60 minutes |
| Coastal gully, small catchment | 1 to 6 hours |
| Coastal river, groundwater | 12 hours to several days |
These durations are indicative and are recalculated for each site, but the hierarchy is constant: the smaller the structure, the shorter the duration.
The episode of 3 October 2015 shows the gap between scales. The State’s lessons-learned report (summary, presentation, in French) gives 109 mm in one hour and 175 mm in two hours in Cannes, with return periods above 100 years over both one and two hours, and 83 mm then 106 mm in Antibes. These figures describe the same evening but not the same structures: the first concerns the street network, the second the gully. Neither can be deduced from the other, and the peak over a few minutes, the one that matters for a gutter, is not in the documents we consulted.
Place matters as much as duration. Mediterranean storms are narrow: a station a few kilometres from the damage may have received double, or half. The return period is judged on the rain that fell at the structure, not on that of the best-known station in the département. “The rain was exceptional” means nothing: exceptional over what duration, at what place, against what series of measurements?
What the structure was designed for
A drainage structure is not designed for “the rain”, but for a specific design rainfall. The diagnosis consists in setting the rain of the loss against this design rainfall. As long as the second is not identified, the first proves nothing.
On the building: 3 litres per minute per square metre
Gutters, box gutters, downpipes and collectors are sized under NF DTU 60.11 part 3 (August 2013), which keeps the intensity of 0.05 litre per second per square metre, that is 3 litres per minute per square metre of roof, as recalled by the French building federation in its sizing file (in French). We could not re-read the text of the DTU, which is paid: the safety coefficient applied by part 3 is to be read in the standard. Converted into the meteorologists’ units, this intensity gives:
| Expression | Value |
|---|---|
| In litres per minute per m² | 3 |
| In millimetres per minute | 3 |
| In millimetres over 6 minutes | 18 |
| In millimetres per hour | 180 |
What this gives against the Cannes episodes. Sustained for one hour, the DTU intensity would represent 180 mm, more than the 109 mm of the wettest hour on 3 October 2015: at that scale, the 2015 rain represents 61% of the design intensity. In September 2024, under a mere yellow warning, Météo-Paris reports about 16 mm in 6 minutes, that is 160 mm/h and 89% of the DTU intensity (a specialist-press figure, to be confirmed against Météo-France data at a 6-minute step). A correctly calculated, clean and well-fitted gutter should therefore not overflow in an ordinary episode, even under a red warning. If it does, the causes lie elsewhere: insufficient section, counter-slope, outlet too small, clogged strainer, crushed downpipe, no overflow.
The following table lists the values found in the sources cited at the end of the article, for several duration windows. The percentage compares the average intensity of each window with the DTU design intensity (180 mm/h):
| Episode | Window | Depth | Average intensity | Share of DTU intensity |
|---|---|---|---|---|
| Cannes, 3 October 2015 | 1 h | 109 mm | 109 mm/h | 61% |
| Cannes, 3 October 2015 | 2 h | 175 mm | 87.5 mm/h | 49% |
| Antibes, 3 October 2015 | 1 h | 83 mm | 83 mm/h | 46% |
| Antibes, 3 October 2015 | 2 h | 106 mm | 53 mm/h | 29% |
| Cannes, 23 September 2024 | 6 min | 16 mm | 160 mm/h | 89% |
| Cannes, 23 September 2024 | 12 min | 18 mm | 90 mm/h | 50% |
| Cannes, 23 September 2024 | 1 h | 26 mm | 26 mm/h | 14% |
The reading is clear: the shorter the window, the closer the intensity comes to the design one. The one- and two-hour windows of 2015 remain far from the DTU, whereas the 6 minutes of 2024 come close to it. It is therefore the 6-minute data, rarely quoted in the press and in summaries, that decides the fate of a gutter: it should be requested from Météo-France rather than concluding on an hourly total. The 2015 rows (Cannes and Antibes) come from the lessons-learned report, the 2024 rows from specialist press.
Two reservations: a peak of one or two minutes can exceed the 6-minute average, and the station is not the plot. These reservations are lifted by measurement, not by assertion.
On networks: a chosen return period
Buried networks and retention structures are calculated for a return period set by the project owner or by the local authority. Standard NF EN 752 grades design flooding frequencies by stake; we could not re-read its table, which is to be consulted in the edition in force. Locally, the stormwater zoning, the sewerage regulations or the local urban plan often impose their own return period for on-plot retention. These documents are read before any conclusion: they say what the building permit required.
On watercourses: the reference flood
Risk prevention plans use the 100-year flood, or the highest historical flood if it is greater. The mapping of areas at significant flood risk distinguishes, according to the 2012 ministerial instruction, a frequent event of 10 to 30 years, a medium event of 100 to 300 years and an extreme event of about 1,000 years (see also our page on PPRI flood zones).
Three possible verdicts
| Rain of the loss, over the critical duration | Reading |
|---|---|
| Clearly below the design rainfall | The structure failed in normal service: design, execution or maintenance |
| Close to the design rainfall | Grey area: check margins, state of maintenance, external inflows |
| Clearly above the design rainfall | Exceedance: then examine the intended behaviour under overload |
The third row does not close the file. A well-designed structure overflows where it was intended to: facade overflows, raised thresholds, a path for the water to the public space. An overflow towards the inside of the building, even under a rare rain, remains a design question. On flat roofs, see our flat-roof infiltration case.
What the law says, and what it does not
To our knowledge, no text makes the colour of a weather warning a criterion of liability or compensation. The two mechanisms that count after a flood, natural disaster and force majeure, reason in terms of the rarity of the event. What follows is a technical orientation, not legal advice: qualification belongs to the judge and to lawyers.
Natural disaster: a ten-year threshold
The natural-disaster state is recorded by an inter-ministerial order, commune by commune. For floods the criterion is statistical: a ministerial answer published by the National Assembly (in French) explains that where the return period is greater than or equal to 10 years the intensity is considered abnormal. Recent orders (for example the order of 5 June 2024) state their decisions in these terms, through rainfall totals or flows with a return period above 10 years. Two lessons:
- The administration looks at the return period, not the colour. A commune under a yellow warning can be recognised, a commune under a red warning can be refused.
- Ten years is not much. The threshold corresponds to a rain that a building has almost two chances in three of suffering during its ten-year warranty. The order opens a right to compensation; it does not show that the event was out of the ordinary.
Force majeure: unforeseeable and irresistible
The builder is presumed liable for damage falling under article 1792 of the Civil Code. He is released only by proving an external cause, including force majeure. Article 1218 defines it as an event beyond the debtor’s control, which could not reasonably have been foreseen when the contract was concluded and whose effects cannot be avoided by appropriate measures. Unforeseeability is therefore assessed at the date of commitment, and intense Mediterranean rains are, at that date, known to statistics.
Case law refuses the shortcut and looks at the event. In a judgment of the first civil chamber of 26 January 1999 (no. 97-10.028, noted in the Répertoire de droit civil, to be checked against the judgment), a flood may be a case of force majeure where the atmospheric disturbances were exceptionally violent, but not where the exceptional nature of the rain is not shown; natural-disaster status is irrelevant there. More recently, the second civil chamber (17 June 2021, no. 17-18.082, reported by ANIL, in French) held that exceptional, unforeseeable and irresistible bad weather exonerated an owner after a landslide, the decision noting that the weather had been the subject of two natural-disaster orders: the orders are a fact, not the basis. The question is therefore neither “was there an order” nor “what colour”, but “would a compliant structure have held?”.
What the judge expects from the expert
If the decree is not enough, the colour of the warning weighs even less. The expert does not have to write “force majeure”. He or she must supply the four figures that allow it to be decided:
- The rain measured closest to the loss, over the critical duration of the structure.
- The return period of this rain, with its margin.
- The design rainfall that the rules required of the structure.
- The real capacity of the structure as built and maintained.
A report that concludes “exceptional event” without these four figures is fragile against the opposing party. See also the deadlines of French building warranties.
Red without major damage, yellow under water: four Riviera episodes
In nine years the Alpes-Maritimes have seen very different combinations of announced colour and rain that fell. None allows one to be deduced from the other. The values below are those we could find in the sources cited at the end of the article; they sometimes differ from one source to another.
| Episode | Warning | Rain recorded | Outcome |
|---|---|---|---|
| 3 October 2015, western coast | Orange | 175 mm in 2 h in Cannes | 20 dead |
| 2 October 2020, storm Alex | Red | More than 500 mm in 24 h at Saint-Martin-Vésubie | Valleys devastated |
| 20 October 2023, storm Aline | Red (from 4 am to 10 am) | Up to 170 mm at Saint-Martin-Vésubie in the morning | Damage clearly lighter than in 2020, but real in the Vésubie |
| 23 September 2024, Cannes | Yellow | Up to 40 mm, including about 26 mm in one hour | Town flooded, no casualties |
3 October 2015: the exceptional under an orange warning
The State’s lessons-learned report (in French) gives 175 mm in two hours in Cannes as a record value, with return periods above 100 years over one and two hours. The toll is twenty dead and about 610 million euros of damage according to one summary. The warning was orange: the deadliest rain of the decade on this coast fell under a colour that nobody invokes as an excuse. The prefecture documents the episode.
2 October 2020: red, and a département cut in two
Alex is the episode where colour and event coincide: the Alpes-Maritimes were under a red rain-flood warning. The Météo-France bulletin of 2 October describes a major Mediterranean episode with a 100-year return period for the département. The Saint-Martin-Vésubie station, swept away by the water, gave 500.2 mm in 24 hours from a neighbouring rain gauge, a département record; other values, from 513 to 663 mm, are cited for other stations (Cerema, Météo-Paris). The previous regional record cited by Cerema was 460 mm at Lorgues in 2010. But these totals concern the valleys: for a coastal building damaged that day, the département-wide red still says nothing, the rain that fell on the spot is needed.
20 October 2023: red, and much lighter damage
Storm Aline placed the département under a red rain-flood warning from 4 am to 10 am, with schools closed. The press (Reporterre, La Gazette France, in French) reports damage clearly lighter than Alex, but significant material damage in the Vésubie valley, with readings of up to 170 mm at Saint-Martin-Vésubie. A basement flooded that day in a commune that received little rain cannot rely on the colour.
23 September 2024: yellow, and Cannes under water
On that Monday morning Cannes flooded while the département was under a yellow thunderstorm and rain-flood warning, with forecast totals below the orange thresholds (Météo-Paris, Futura, in French). The readings cited reach up to 40 mm in total, including 26 mm in one hour and 18 mm in 12 minutes (Météo et Radar, Brut), and about 16 mm in 6 minutes according to Météo-Paris: intense but short rain, no casualties, with vehicles swept away and shops and cellars hit. It is the most instructive case: rain that is nothing like a record floods a town, so the cause is not primarily in the sky. It is in sealing of surfaces, network capacity, maintenance of gullies, the level of thresholds.
What these four dates teach
- The colour does not predict local rain, in either direction.
- Damaging rain often returns to this coast: four notable episodes in nine years, all between late September and October.
- A Riviera builder cannot plead surprise: the Mediterranean episode is part of the project’s input data, like wind.
Where to find the figures
A return period is demonstrated with two datasets: the rain of the event and the statistics of the place.
The rain of the event
- Météo-France stations. They give totals at a fine time step, down to 6 minutes for automatic stations. This is the reference measurement, provided the station is close and at a comparable altitude.
- Radar rainfall estimates. They combine the radar image and rain gauges to estimate rain per square kilometre, and fill the gaps between stations.
- Local authority networks. Some authorities run their own rain gauges and level sensors. These data are requested in writing.
- Private rain gauges and testimonies. They guide, they do not prove. A full bucket is not a measurement.
The statistics of the place
- Return-period sheets and Montana coefficients. Météo-France produces them per station, for several durations and return periods, with a confidence interval.
- Regionalised estimates. The SHYREG method provides rare rainfall depths at any point of the territory.
- The grounds of natural-disaster orders. Published in the Journal officiel, they state the criterion used per commune. It is an indication, bounded at the 10-year threshold.
- Existing hydraulic studies. Risk prevention plan, stormwater master plan, water-law file of the estate: they often contain the design rainfalls of the area.
Three losses, three readings
The three cases below are constructed to illustrate the method: the values are chosen for the calculation and describe no real file, with no commune or date. They show how the same method leads to three opposite conclusions without ever going through the colour.
The box gutter that overflows under a red warning
A co-owned building suffers facade infiltration on a red-warning day, and the contractor invokes the event. The neighbouring station recorded at most 9 mm in 6 minutes, that is 1.5 litres per minute per square metre: half the DTU design rainfall (3). The box gutter should have held with a margin of two. The inspection finds a single outlet, no overflow, and a clogged strainer. The colour was red, the rain was ordinary, the defect lies in design and maintenance.
The car park flooded through its ramp under an orange warning
An underground car park takes 40 cm of water. A radar rainfall estimate gives 55 mm in one hour over the district, that is, in this example, a return period estimated between 20 and 50 years, while the stormwater rules required 30-year protection. This is the grey area. The expertise moves to the structure: section of the ramp channel, height of the threshold, state of the sump pump, water coming from the road. The conclusion will depend on these findings, not on the weather alone. See our basement car park damp case.
The house by the gully under a yellow warning
A ground floor is invaded by a gully that left its bed, with no orange warning. A radar rainfall estimate nevertheless gives 130 mm in two hours over the small catchment, beyond the local 100-year rain in this example. The structures were compliant and maintained: the exceedance is established by figures, even though the colour would have led to the opposite conclusion.
Eight rules for a defensible report
A defensible flood report compares a measured rain with a design rainfall, over the same duration and at the same place. The eight rules that follow all stem from that sentence.
- Keep the colour out of the conclusion. The warning may appear in the chronology, with its time of issue. It never appears in the reasoning on cause.
- Name the source of the rain. Station, identifier, distance to the loss, altitude, time step. Or radar estimate, with the product and grid. Rain without a source is a rumour.
- Choose the duration before reading the figures. The critical duration follows from the structure: minutes for a gutter, an hour for a network, half a day for a watercourse. Choosing it afterwards is choosing the conclusion.
- Give the return period with its margin. “Between 20 and 50 years”, with the series used and its length. Never “37 years”, never “exceptional”.
- Write down the structure’s design rainfall. DTU, standard, stormwater rules, permit requirement, engineer’s note. If it cannot be found, say so: that is already a finding.
- Measure the structure as it exists. Sections, slopes, number of downpipes, overflows, state of maintenance. Real capacity is calculated on what was built, not on the drawings.
- Reconstruct the path of the water. Where does it come from, where did it get in, where should it have gone? Part of the water often comes from the neighbouring land or the road and is not the responsibility of the structure examined (see rainwater draining toward my house).
- Separate the finding from the qualification. The expert establishes whether the rain exceeded the design return period or not. “Force majeure” and “natural disaster” are words for judges and the administration.
The documents to request before concluding
- Rain records at a 6-minute step from the nearest station, with its identifier, distance and altitude, or the radar product and its grid.
- The return-period sheets of that station, for the durations relevant to the structure.
- The building permit, the engineer’s note, the stormwater rules and zoning applicable at the date of the project.
- The execution drawings and maintenance logs: sections, slopes, number of downpipes, overflows, dates of cleaning.
- Dated photographs of the loss, with the water-level mark, and the chronology of interventions.
A model conclusion paragraph
Here is the form of a defensible conclusion, to be completed with measurements and never with adjectives:
For the reader of a report, managing agent, insurer or owner, the reading grid fits in five lines:
| What you read | What to demand |
|---|---|
| “The département was on red alert” | The rain measured closest to the structure |
| “Exceptional rainfall” | A return period, over a named duration |
| “Two months of rain in one day” | The 10-year and 100-year depth over the same duration |
| “The commune is classed as a natural disaster” | The comparison with the structure’s design rainfall |
| “No structure would have withstood it” | The calculated capacity of the compliant structure |
A report that relies on the colour of the day?
We are an independent building expertise firm: we sell neither works nor treatment. After a flood we look for the rain that actually fell, compare it with the structure’s design rainfall and write what that gap allows us to conclude. If you have received a report that relies on the weather warning, send it to us for a review. Free quote within 24 hours.
Have a report reviewedWater infiltration diagnosisSee our rates
Also on this site
- Water infiltration: diagnosis and expertise in the Alpes-Maritimes and the Var ;
- Basement car park damp in a new build, Nice West ;
- Roof-terrace water infiltration: flow rate ;
- Flood zone (PPRI) in Nice: extensions and new construction ;
- French building warranties: deadlines.
Glossary
- Weather warning (vigilance)
- warning issued by Météo-France, in four colours, by département, for a hazard expected within 24 hours. It is neither a measurement nor an alert.
- Return period
- average interval, over a very long time, between two exceedances of a given value. A rain of return period T has each year a probability 1/T of being reached or exceeded.
- 10-year, 100-year rain
- rain of return period 10 years, 100 years. Always tied to a duration and a place.
- Intensity
- rainfall depth per unit of time, in millimetres per hour or litres per minute per square metre. One millimetre of rain equals one litre per square metre.
- Montana coefficients
- a pair of parameters a and b that links, for a station and a return period, rainfall depth to its duration.
- Time of concentration
- time water takes to travel from the farthest point of a surface to its outlet. It sets the most penalising rain duration for the structure.
- Design rainfall
- rain adopted to size a structure, defined by an intensity or by a return period.
- Radar rainfall estimate
- estimate of the rain fallen at each point, obtained by combining weather radars and rain gauges.
Frequently asked questions
Does a red warning prove the rain was exceptional?
No. A warning is a hazard forecast at département scale, issued before the rain. It does not measure what fell on a commune, still less on a plot. Only a rain measurement compared with local statistics establishes that it was exceptional.
What is a 10-year rain?
It is a rain whose depth, over a given duration and at a given place, has a one in ten chance of being reached or exceeded each year. It does not return every ten years on a fixed date. Over ten years, the probability of suffering at least one is about 65%.
Does a natural-disaster decree exonerate the builder?
Not by itself. Force majeure requires an unforeseeable and irresistible event, assessed on the facts: the builder must show that a compliant structure would not have held. That demonstration rests on rainfall and capacity figures.
How do I find the return period of the rain that flooded my building?
First the rain measured near the building is needed, from a station or a radar estimate, over the duration that matches the structure. It is then compared with the rainfall depths of different return periods, established by Météo-France for the reference station. The result is expressed as a range.
My gutter overflowed during a storm: is that normal?
Rarely. Roof drainage is sized for 3 litres per minute per square metre, that is 180 mm per hour. An overflow most often signals clogging, insufficient slope, too small a section or too few downpipes, but a peak of a few minutes can approach this intensity: measurement settles it.
After how many years does a 100-year rain have an even chance of occurring?
After about 69 years: the probability of at least one rain of return period T over n years is 1 − (1 − 1/T)^n, and it reaches 50% for n close to 0.69 times T. On the scale of a building’s life, it is not an event outside time.
References
Pages consulted on 16 November 2026, often through excerpts: each figure should be rechecked on the full document before being used in a report.
- Météo-France, Le phénomène pluie-inondation (orange reference point for a lowland département).
- Lessons-learned report of 3 October 2015, summary and presentation of the lessons learned from the 3 October 2015 storms in the Alpes-Maritimes ; Alpes-Maritimes prefecture page.
- National Assembly, answer to written question no. 10051 (return period of at least 10 years criterion) ; order of 5 June 2024 recognising the natural-disaster state.
- Météo-France, daily bulletin of 2 October 2020 ; Cerema, weather presentation on Alex ; Météo-Paris, Storm Alex.
- Reporterre, intense episode in the south-east ; La Gazette France, Aline causes damage (October 2023).
- Météo-Paris, surprise floods in Cannes ; Futura-Sciences, Cannes ; Brut, 23 September 2024 ; Météo et Radar, Cannes (September 2024).
- French building federation, DTU 60.11: sizing of rainwater drainage ; AFNOR, NF DTU 60.11 P3 (index P40-202-3, in force).
- ANIL, Natural disaster and force majeure (Cass. 2nd civ., 17 June 2021, no. 17-18.082).
Cited without being re-opened: NF DTU 60.11 P3 and NF EN 752 (paid standards, whose content we did not re-read), Civil Code (articles 1218 and 1792), Insurance Code (article L. 125-1), Cass. 1st civ., 26 January 1999, no. 97-10.028 (known from a secondary commentary), the 2012 ministerial instruction on areas at significant flood risk, the Montana formula. Unchecked values: Météo-France’s red threshold (not used here), the 16 mm in 6 minutes peak of September 2024 (specialist press), critical durations by structure (orders of magnitude) and the three illustrative losses (constructed values). The probabilities and the DTU conversions are calculated and were recomputed on 16 November 2026.
Survey carried out by Frédéric Faucouneau, independent building expert, Antibes, Alpes-Maritimes (06), Var (83), Bouches-du-Rhône (13). Last updated: 16 November 2026. Find out more
Last updated 16/11/2026 - Cassini Expertise, independent building expert, departments 06, 83 and 13.