COMPASS "Compound extremes attribution of climate change: towards an operational service"

France 2010

Extra-tropical cyclone ‘Xynthia’ in 2010

Extra-tropical cyclone ‘Xynthia’ was one of the most significant disasters to affect France in recent decades. The storm tore through the country in the early hours of 28th February 2010, causing 47 fatalities and some 2.5 billion euros in damages. Most of the fatalities and half of the economic damage resulted from storm-induced flooding on the western coast of France.

The windstorm and its impacts were not unusual in France, especially when compared to the 1999 storms. However, the coastal flooding was unexpected, a combination of both the rare synchronicity of storm surge and spring high tide and many long-standing deficiencies in disaster management and planning. Since the event, many actions have been undertaken to reduce exposure and vulnerability through improved flood protection (including temporary barriers), spatial planning, new emergency-response organizations and building adaptation requirements.

Flooding directly affected some 10,000 people and the windstorm cut off power to one million people, and the whole event resulted in some 470,000 insurance claims of an average 3100 euro.

La Rochelle
affected by flooding

Geographical context

Several European countries were impacted, but most were concentrated in France, including 47 out of 59 fatalities. The map (Figure 1) presents the departments (second-level administrative divisions) where fatalities had occurred, and the 25 departments with the highest insured economic losses (72% of the total). 41 fatalities and 54% of damages were concentrated in two departments, Charente-Maritime and Vendée (marked with blue hatching on map). Most fatalities and the highest economic loss occurred in La Faute-sur-Mer municipality in Vendée (Figure 2).

Figure 1 Main regions affected by extra-tropical cyclone Xynthia in 2010.
Background map: OpenStreetMap.

Figure 2 Principal municipalities that were affected in the Charente-Maritime and Vendée departments.
Background map: OpenStreetMap.

COMPOUND HAZARd

Windstorm and Coastal Flooding During Xynthia

Extra-tropical cyclone ‘Xynthia’ was an unusual event among winter storms that affect Europe annually. Despite being weaker than some other storms (particularly compared with 1999 storms), it was particularly damaging due to the coastal flooding component. The storm developed around 30^◦N, which is more southerly than typical winter storms. It developed during the record-breaking negative values of the North Atlantic Oscillation (NAO) index. It followed a very unusual trajectory from the Canary Islands through the Iberian Peninsula into France. The storm was also particularly fast-moving, covering 1400 km in 24 hours while it moved across Spain and France and then Germany and Benelux countries. Still, the important driver of coastal flooding was the synchronicity with a spring high tide, as the wind-driven surge was lower than in the 1999 storm. This allowed the water level to reach unprecedented heights.

During storm Xynthia, the French coast was struck by overlapping hazards – violent windstorms and coastal flooding triggered by the combination of extreme wind speeds, low atmospheric pressure, and a high spring tide – which occurred simultaneously in the same place and time, amplifying the disaster’s impacts.

The wind gusts during the storm reached up to 238 km/h at Pic du Midi in the Pyrenees. At the coastline, wind gusts were up to 160 km/h in Ile-de-Ré (island in the Bay of Biscay) and up to 130 km/h in coastal towns on the mainland. Inland, speeds of 160 km/h were also reached in the Deux‐Sèvres department. The sea level reached a record-breaking 4.51 m in La Pallice (harbour of La Rochelle), though the actual surge was only 1.58 m (estimated to be less than in 1924, 1940, 1941 and 1999 events).

The storm was first detected as a depression in the Atlantic Ocean on the 25^th of February, 2010. On the 26^th, the Canary Islands were impacted. On the 27^th the storm moved through the Iberian Peninsula into the Bay of Biscay. The highest red warning for four departments of France were issued on the 27^th at 16.00 pm. The storm made landfall in France on the 28^th around 2.00 am. By 6.00 am, it had moved to Paris and continued into Belgium, the Netherlands and Germany. It finally dissipated over the Baltic Sea before midnight on the 2^nd of March.

Flooding animation courtesy of Xavier Bertin, based on Bertin et al. (2014) A modeling-based analysis of the flooding associated with Xynthia, central Bay of Biscay. Coastal Engineering, 94, 80-94, https://doi.org/10.1016/j.coastaleng.2014.08.013.

Observed impacts

Xynthia caused extensive losses to people
and the national economy

The most detailed data is available for fatalities and insured economic losses (damage to houses, cars, businesses and agriculture). Direct economic losses were estimated at around €2.5 billion (combined flood and wind damage). Insurers received about 470,000 claims (35,000 flood; 435,000 wind), with a total claim value of roughly €1.48 billion (€745 million flood; €735 million wind).

Though the 25 departments with the highest losses were responsible for 72% of insured windstorm losses, 89% of flood losses were recorded in the top two departments.

Distribution of insured economic loss from wind damage during Xynthia (only 25 departments with highest losses).

Distribution of insured economic loss from flooding in the Charente-Maritime and Vendée departments during Xynthia (black numbers) and fatalities (blue numbers).

Principal impacts of extra-tropical cyclone Xynthia in 2010

Storm Xynthia caused extensive damage in France: about 560 km²—mostly agricultural land—was inundated; 47 people were killed (41 due to flooding and 6 due to wind); roughly 10,000 people were directly affected by flooding, and power outages impacted about 958,000 customers.

The six fatalities from the windstorm were highly scattered all over France, but flood fatalities were highly concentrated. 29 out of 41 fatalities occurred in a single municipality (La Faute-sur-Mer), of which 28 occurred in an area of just 0.3 km² (compared to 560 km² flooded area and 12 km² of the whole municipality).

Details of insured losses

Data from the Fédération française des sociétés d’assurances (FFSA) and Groupement des entreprises mutuelles d’assurance (GEMA)

Insurance claims (number).

Total

Flooding: 19000, Windstorm: 339500

Insured loss (million euro).

Total

Flooding: 450, Windstorm: 430

Insurance claims (number).

Total

Flooding: 10500, Windstorm: 22500

Insured loss (million euro).

Total

Flooding: 60, Windstorm: 35

Insurance claims (number).

Total

Flooding: 1700, Windstorm: 33500

Insured loss (million euro).

Total

Flooding: 26, Windstorm: 112

Insurance claims (number).

Total

Flooding: 3800, Windstorm: 39500

Insured loss (million euro).

Total

Flooding: 209, Windstorm: 158

Insurance claims (number).

Total

Flooding: 35000, Windstorm: 435000

Insured loss (million euro).

Total

Flooding: 745, Windstorm: 735

Reported flood extents resulting from the Xynthia storm

Qualitative impacts and responses

Relatively little is mentioned in context of the windstorm, as it was not particularly impactful (compared to particularly 1999) and remains overshadowed by the unprecedented magnitude of coastal impacts. Post-flood analyses list several reasons for the scale of impacts

The windstorm was correctly forecast well in advance, but storm surge warning was not precise and actionable

Upon warning, no action was taken by local authorities, who considered windstorm the larger risk, rendering evacuation not feasible (except campsites, due to windstorm rather than flooding). However, local authorities had no possibility of knowing the possible severity of flooding.

Many dikes, seawalls and beaches were badly maintained, partially due to unclear titles or private ownership. Some dikes were also still unrepaired after the 1999 storm surge, except Gironde department which was most affected in the previous event. Long-term investments in coastal protection have also protected most of the town of Châtelaillon-Plage.

The coastal zone experiences decades of rapid growth in housing stock. In addition, many houses were built in violation of a 1995 spatial planning law, which aimed at stopping construction in areas at risk of flooding, but local authorities still often authorized housebuilding even in the most flood-prone areas.

Between 1960s and 1980s, houses were built typically with the bedroom on the first (upper) floor, reducing vulnerability especially in terms of mortality. However, newer houses were built with only a ground floor to imitate the more traditional style of the 1950s and earlier houses.

Ground-floor-only houses were preferred by older people, who had lower chances of escaping floodwaters, further exacerbated by possible exit through windows often being blocked by electric shutters in conditions of power failure due to windstorm. Consequently, the median age of Xynthia fatalities was 75, while there were no fatalities between the ages of 15 and 42.

The flood occurred in wintertime, when few seasonal homes were occupied, which reduced the number of fatalities, though it didn’t decrease economic losses. However, the occupancy wasn’t at its lowest possible due to winter school holidays, and 14 out of 41 fatalities were non-residents. All five children who died in the flood were on holidays.

Population and housing data for the most affected La Faute-sur-Mer municipality from national statistics institute INSEE reveals the specifics of risk and its dynamics typical in the region, but particularly acute in this location. Firstly, it is noticeable that seasonal housing far exceeds permanent housing. Further, this ratio increased from 4:1 in 1968 to almost 7:1 before the flood, and the overall number of houses increased nearly 4-fold. At the same time, the population had only increased 70% in the same timeframe. After the flood, exposure reduction was implemented in the two affected departments of Charente-Maritime and Vendée by (mostly) voluntary purchase and demolition of 1162 houses in the highest-risk zones. This was paid by the national government to the amount of nearly 318 million euro. In La Faute-sur-Mer, the reduction in the number of seasonal houses by 2016 was 36%, permanent houses by 28% and population by 34%. However, since then the population and housing have started to increase in numbers again.

One exception to seasonal housing being generally unoccupied due to winter can be noted. Many mobile homes in the area were inhabited by students who couldn’t afford to rent flats in La Rochelle. The mobile park in Aytre was completely flooded, but the young people were able to escape to the roofs and survive. This is in contrast to older inhabitants of permanent houses that came under water, especially in La Faute-sur-Mer. However, six fatalities of the windstorm include two inside a camper that was washed away by waves from a pier.

Resident population (1954-2022) and number of houses (1968-2018) in the most affected municipality, La Faute-sur-Mer (census data from INSEE). No housing data after 2018 is available due to the merger of the municipality with L’ Aiguillon-sur-Mer.

Attribution analysis indicates a contribution of climate change to windstorm losses, though with considerable uncertainty. This indicates the need to frequently reevaluate the minimum standards for buildings – European construction regulations state that they are required to withstand wind with a return period of 1 in 50 years.

Attribution results

Climate change is shown to have increased the impacts of both coastal flooding and windstorm. As the coastal flood levels were influenced especially by well-quantified sea level rise since 1950, the confidence of this finding is very high (90% confidence interval is +10% to +14%). For the windstorm, the uncertainty using the hadGEM-3A approach is high, as the 90% confidence interval spans from –71% to +44% influence of climate change relative to preindustrial levels (for the tsEVA approach, the uncertainty was not quantified).
The statistical approach with a 1950 baseline both indicates a greater influence of long-term changes in climate (22% versus 7%), but with considerable spatial differences (see map below). This indicates the strong influence of the choice of method and baseline in attributing climate impacts.

Windstorm

Windstorm losses are heavily driven by GDP growth, with much lower contribution of population growth or increase in relative asset value. Spatial distribution of assets in context of the Xynthia windstorm changed very little since 1950 but has noticeably moved towards lower-impact zones since 1850.

Change in windstorm damage due to observed long-term changes in climate since 1950 using the transformed-stationary EVA approach.

Flood

The attribution of flood impacts was more comprehensive in terms of the number of impacts and drivers used. Relative to the 1950 baseline, climate change contributed to a 10-12% increase in impacts across all three types of impacts. Economic loss was substantially increased by economic growth (633%), but population growth contributed much less (45-48%). Spatial distribution of population and assets, as well as change in economic structure, is estimated to contribute little to the losses. Similarly, changes in flood protection levels are estimated to have very limited influence on the magnitude of losses.

Most importantly, changes in vulnerability (related to the level of non-structural adaptation and preparedness) are estimated to have compensated substantially for increases due to climate change and exposure growth (56-66% decrease). However, the uncertainty in estimating past vulnerability is high, similarly to the protection levels. Therefore, confidence in this finding is low, especially in context of the model underestimating factual flood impacts in the first place.

Reported fatalities (persons)

Absolute factual impacts

Reported population affected (thousand persons)

Absolute factual impacts

Reported economic loss (million 2020 euros)

Absolute factual impacts

Stakeholder and policy relevance

In the two affected regions, extensive seawall and dike renovations as well as smaller technical adaptation measures have been carried out since the flood and will continue to be implemented at least until 2030, at a cost of some 1.5 billion euro. Within the ‘agglomeration communality of La Rochelle' (which combines the city with several smaller municipalities) several adaptation measures are currently implemented.

According to the civil protection office of the La Rochelle agglomeration, they include:

Prefectural flood maps and flood prevention plans
Use of Xynthia flood level as baseline for planning plus 20 and 60 cm sea level rise
Requirement of floor levels in new or extended buildings to be above potential flood levels
Supporting adaptation of existing buildings in the Xynthia + 20 cm zone
Reducing new construction in the Xynthia + 60 cm zone
Early warning procedures plus a voluntary response force
An 800-meter temporary barrier to protect La Rochelle harbour (deployment time: 7 hours with 28 persons)

Trees painted blue to signify Xynthia water level and 20 cm sea level rise. This part of La Rochelle was actually flooded to that level in 2010. (Photo D. Paprotny)

Trees painted blue to signify Xynthia water level and 20 cm sea level rise. This part of La Rochelle was actually flooded to that level in 2010. (Photo D. Paprotny)

Xynthia flood marker in the La Rochelle historical harbour. (Photo D. Paprotny)

Xynthia flood marker in the La Rochelle historical harbour. (Photo D. Paprotny)

Public evacuation centre in La Rochelle. (Photo D. Paprotny)

Movable flood barriers in the La Rochelle harbour. (Photo D. Paprotny)

Movable flood barriers in Châtelaillon-Plage. (Photo D. Paprotny)

France 2010

La Rochelle
affected by flooding

Geographical context

Windstorm and Coastal Flooding During Xynthia

Île de Ré

La Pallice (La Rochelle)

Deux-Sèvres (inland)

Pic du Midi (Pyrenees)

Paris

Xynthia caused extensive losses to people
and the national economy