Twenty years ago, while the planet was getting ready for transitioning to year 2000 and trying to solve the Y2K bug, the (re)insurance industry in Europe got caught by surprise by windstorm Lothar. Even today, 1999 remains a historic windstorm year, with catastrophic storms Anatol (December 3), Lothar (December 26) and Martin (December 28) all happening within a period of less than a month.
Lothar tracked across northern France, southern Belgium and central Germany and into Poland; Martin tracked through southern Europe – affecting France, Spain, Switzerland and Italy. Between Lothar and Martin, 140 people were killed, and losses ran over €14.2 billion economic losses, approximately €7.7 billion of which was insured. If the three events happened today, they would cost approximately €20 billion (US$23.3 billion) to the (re)insurance industry.
At the time, I was still living in Geneva with my parents. I remember waking up the day after Christmas and seeing fallen trees in our garden and our telephone line was cut. It was very dramatic and since then, no other windstorm has caused that kind of damage in this region.
In commemoration of the twentieth anniversary of windstorms Anatol, Lothar and Martin, I have asked my colleagues at RMS to share their experience of the storms.
What Were You Doing on December 26, 1999?
Robert Muir-Wood, chief research officer at RMS.
Christmas Day and Boxing Day (December 26) in the U.K. are public holidays in which almost everyone is off work. Trains don’t run, the television news service is on skeleton staff, with minimal content. Around 5 p.m. on December 26, I received a call from a colleague at RMS in California, Paul VanderMarck, asking if I had seen the news from Paris? I had not.
I immediately went online to see what I could discover from French news sites, and stayed online for the next 10 hours, piecing together accounts from many local French and German websites, recounting extraordinary levels of wind damage to roofs, power lines and forests from Lothar. One of the first metrics available to reflect the severity of windstorm, and the ultimate cost of repairs to damage, unfortunately comes from casualties – and most of those killed were in cars or outside.
While the storm had hit Paris without warning – early in the morning, before many people were up, late morning the storm had gone on to cause many casualties as it passed into Germany. More than two million households were without power. This was clearly going to be a multi-billion Euro loss event: comparable in windspeeds to the great 87J storm and in loss magnitude to 1990 windstorm Daria. And another even larger intense storm was on its way heading for western France…
Stephen Cusack, senior director in Model Development at RMS, was working in the Numerical Weather Prediction group of the U.K. Met Office at the time. He told me how sobering windstorm Lothar was from a numerical modeling perspective:
The U.K. Met Office had made a huge investment into modeling after the forecast bust for 87J in October 1987, yet the forecasts, like most from around the world, failed to predict the exceptional strength of Lothar.
The main forecasting lesson from Lothar is quite universal and timeless: automated models can be very good where training samples are large, but expert human input would be very useful during rare events.
Farid Ait-Chaalal, also a hazard modeler working on RMS Europe Windstorm Models, was living in Paris in 1999. He remembers vividly the storm as it came through the city:
The wind and the sound of objects falling woke me up. I looked through the windows and I saw some objects flying. Roof tiles, I think. It was in the 5th arrondissement of Paris.
Windstorm Lothar caused tremendous damage, not only to property, but also to forestry. Unlike property damage that can be repaired fairly quickly, damage to forestry can change the landscape of an entire region and take many years to recover. In the case of forestry damage, windstorm Lothar allowed more research to improve forest management so they can better sustain windstorms.
In France, 140 million square meters were felled, with a further 13 million square meters lost in Switzerland. Farid recounts how four years later, the Vosges region, on the eastern part of France, had not yet recovered from the storm. He took these pictures near the Lac de la Maix.
Pierre Wiart, a model developer for RMS at the time and now client director in Australia, was in Caen, Normandy, during windstorm Lothar. He also recalls how exceptional the wind gusts were, and tells me how he was struck by the landscape changes due to fallen trees between Caen and Paris. For months, logs were piling up along the roads because people did not know what to do with them.
However, windstorm Lothar did not only affect France, but also Switzerland and Germany. I now live now in Zurich, where windstorm Lothar also caused significant damage to forests and buildings there, as we can see on the picture below, published in a local newspaper.
A stone commemorating the storm has also been erected in the middle of the Zürichberg, a hill in the north of Zurich.
In the (re)insurance world, it was also the opportunity for catastrophe modelers to get a better understanding of European windstorms and hone the vulnerability modeling for the European market. Michael Drayton, consultant and storm surge specialist at RMS, and Navin Peiris, senior director in Model Development, highlighted lessons learned from the storms, especially how the three storms have been a crucial part of the calibration of damage curves and understanding of post-event loss amplification.
Part of the learning happens by collaborating with industry partners, which allowed RMS to improve vulnerability modeling. This is a continuous process, and year by year, as more events happen, this process and the damage curves are updated.
Another part is through reconnaissance trips. After the storms, RMS sent employees to Denmark and France to evaluate damages. They published a report on storms Lothar and Martin, which is still available to clients today. One of the authors, Barbara Page, is a senior director for RMS:
A few months prior to the storms, I can remember visiting one European insurer asking if they would be willing to share exposure data with us. Their opinion at the time was very much that “There is no windstorm risk in Europe.” Shortly after that, on December 3, windstorm Anatol blew straight through the cover of several primary insurers in Denmark. This was a once in a lifetime windstorm event for Scandinavia.
When Lothar and Martin hit France just two weeks after, I was at home running RMS RiskLink analyses during my Christmas break, on a clunky old desktop computer with a dial-up modem, trying to identify best-fit stochastic events by matching up very sparse wind speed data with wind footprints from our stochastic event set. Because Lothar hit Paris, it was a challenge to find representative wind observations at all. On top of that, we had at the time no automated process for selecting, sifting and correcting wind observation data. There were a lot of judgment calls taken.
Many things at RMS were born out of this painful experience – not least of which was our professional event response service, which now plans in advance how to deal with catastrophes as they occur, and sets up relationships with key meteorological organizations to obtain the best available observation data.
Twenty years later, many people still remember this series of storms, whether they are working for the (re)insurance industry or simply because they had been woken up during the night by flying tiles. Eventually, events like windstorm Lothar stay in people’s mind and awakens curiosity, which might lead into a career in meteorology or catastrophe modeling. Today, a new generation of modelers and researchers benefit from the experience of those who were already around back then.
Windstorm is a key peril in Europe. RMS constantly invests in the research and development, to continue to improve our understand of this peril and provide the (re)insurance industry with the best Europe Windstorm Models:
- In 2015, RMS released the Europe Windstorm Models version 15.0, including latest knowledge of long-term and inter-decadal variability of view of windstorm risk.
- In 2016, RMS updated the clustering methodology, to include new methodology and analyses of long-term historical datasets.
- In 2018, RMS released a new set of 18 historical storm footprints, to complete the existing catalog of 135 historical storms.
- RMS is currently working on a new version of the Europe Windstorm Models, which will leverage latest numerical modeling, data and knowledge, to complete the high-definition modeling solution in Europe.
Interested to read more? Check our blog article written on the occasion of the fifteenth anniversary of Lothar and Martin.