The natural catastrophe regime does not cover damage caused by a natural agent, it covers damage caused by the abnormal intensity of a natural agent. The word carrying the whole weight of that sentence is abnormal, and it does not mean severe. This is the most constant source of misunderstanding in the field, the one that produces indignant appeals and loses months: an insured whose workshop is destroyed takes it as obvious that its loss is abnormal, since it had never happened before. Yet the criterion measures nothing of what it went through.
What the criterion measures is the statistical rarity of the meteorological or geological event itself, compared with a series of observations. For flooding by overflow, one looks at the rain that fell on a catchment and the return period attributable to it given a station's readings and the available series. For flooding by groundwater rise, one looks at the water table level against its history. For drought, one looks at a soil moisture index, season by season, compared with the same seasons of earlier decades. In all three cases the measurement bears on the agent, never on the damage.
That distinction produces two situations that look unfair and are coherent. The first: ordinary rain falls on a catchment already saturated by three wet weeks, an undersized drainage network backs up, forty businesses are under water, and the criterion is not met because the rain, taken by itself, was in no way exceptional. The second, symmetrical: rain of a high return period falls on drained ground, the order is issued, and only three insureds are found to rely on it. The amount of the damage enters the reasoning at no point, and trying to make it enter is the costliest methodological error in the field.
From this follows the way to read a refusal of recognition, which is a technical document and not a general appreciation. A refusal states at which station the measurement was taken, on which quantity, and at what return period it arrives. Each of those three elements can be argued, and it is the only thing that can usefully be argued. The station may be twenty kilometers away and on the far side of a ridge, the quantity chosen may be a twenty-four hour total where the damage came from an intensity over two hours, and the reference series may be short. A file reopens on a rain gauge, never on photographs of the damage.
One must also know that the regime's perils are not all judged on the same type of criterion, which changes what has to be produced. Flood and drought are judged on quantified thresholds, with public data. Ground movement other than drought, avalanche and earthquake are judged on geotechnical or seismological expert reports, where the demonstration is descriptive and where the adversarial process takes another form. An insured producing a geotechnical report in a flood file is spending money on a document that will not be read.
One case deserves to be isolated because it accounts for a growing share of refusals: urban runoff. Water rushing down a sealed slope, saturating a network and entering through a door is not the same thing as a river leaving its bed, and the demonstration of abnormal intensity is harder there because the cause of the damage is partly human. The subject is not academic. When a refusal rests on a defect in public works, a badly maintained retention basin or an undersized network, the insured's route is no longer the natural catastrophe regime but the liability of the works manager, and steering it there early saves it a year.
The method therefore comes to three questions, asked before anything is written. Which quantity was measured, and is it the one that caused the damage. Where was it measured, and does that station represent the place of the loss. And on what reference series is the return period computed. If all three answers hold, the refusal is solid and the client's energy must go elsewhere, towards the ordinary-law covers in its contract or towards a liable party. If one gives way, the commune can file a supplementary request, and it is the commune that files it, never the insured.
A stationary thunderstorm falls on an area on June 8. Two neighboring business parks are hit. In commune A, seven companies suffer 480,000 euros of damage; in commune B, six kilometers away, forty-one companies suffer 2.1 million euros. Three months later, the order recognizes commune A and refuses commune B. The refusal states a total of 62 millimeters over twenty-four hours at the station nearest B, return period estimated at eight years, below the threshold applied. The companies in B point out that their damage is four times that of A and that the water came from a communal retention basin whose valve was jammed. What is that appeal worth?
The appeal as formulated is worth nothing, and that must be said early so that forty-one companies do not lose a year on it. The difference in damage between A and B is irrelevant: the criterion measures the rarity of the rain and not the amount of the damage, so that one commune can be ruined by an ordinary event and another recognized for a rare event that broke nothing. Citing 2.1 million against 480,000 amounts to producing a quantity the commission does not look at. What can be argued, by contrast, lies in the refusal itself and it must be taken at its word on its three elements. The station: it is the nearest, which does not mean it is representative, and a stationary storm is precisely the phenomenon that can dump twice as much six kilometers away; if a private rain gauge, a network operator's reading or an agricultural station exists on the site, the commune can produce it. The quantity: the total used is over twenty-four hours whereas the damage from a stationary storm comes from an intensity over one or two hours, and that is the most frequently winning gap. The reference series, finally, can be checked. But the appeal's second element is the more important, and it does not belong to this regime: if the water came from a communal basin whose valve was jammed, the determining cause is a defect in the upkeep of public works and not the abnormal intensity of a natural agent. The route is then the manager's liability, with an entirely different timetable and burden of proof, and the state of the valve must be formally recorded before it is repaired. The two routes do not exclude each other and are prepared in parallel, but they are not argued before the same body.
- 01The criterion measures the rarity of the natural agent, never the severity of the damage: the amount of the loss enters recognition at no point.
- 02A refusal is argued on three elements and three only: the measuring station, the quantity used, and the reference series behind the return period.
- 03Ordinary rain on a saturated catchment can ruin forty businesses without opening the cover; that is coherent, and it has to be explainable.
- 04Perils are not all judged alike: quantified thresholds for flood and drought, expert reports for ground movement, avalanche and earthquake.
- 05When the determining cause is a defect in public works, the route is the manager's liability, and the state of the works is recorded before it is repaired.