Cloudburst in Denmark - Just Another Tail Event?
October 22, 2015| By Jan Højgaard
One of the main purposes of a reinsurer is to cover tail events, characterized by high volatility and low frequency. Some tail events, such as the occasional large industrial loss, are predictable, while others present an opportunity for learning new lessons.
On July 2, 2011 Copenhagen was hit by a cloudburst. As it turned out, it was the second largest natural peril in Denmark's modern times and insured losses in Copenhagen City Centre were second only to the 1999 Windstorm Anatol.
A cloudburst is a sudden downpour of unusually large quantities of rain in a limited period of time in a given area. For the Danish Meteorological Institute (DMI), 15mm of rainfall within 30 minutes would be defined as a cloudburst. The insurance industry would define a heavy cloudburst as 40-50mm in a 24-hour period or 1mm per minute (1 millimeter per square meter =1 liter).
Cloudbursts are a normal occurrence in Denmark during the summer, and they generally cause limited insured losses - both in terms of amount and space. Losses are usually handled well within the retention of Nat Cat programs; consequently, exposures have not been modelled and have played an insignificant role in the pricing of Nat Cat programs. Meanwhile, we have become wiser.
Now, four years later, we can draw some conclusions about its size, initiatives undertaken by both authorities and insurers, and how the industry should manage the risk in the future.
Here are the main factors that contributed to the loss:
- Huge volumes of rainfall over central Copenhagen: the Botanical Gardens' meteorological station recorded 135.4mm in two hours; other stations recorded 4.5mm in 60 seconds, 31mm in 10 minutes, and 63mm in 30 minutes. These are the highest intensities ever measured in the city.
- It was the “perfect hit” in terms of location. The city center has few green recreational areas where the water could be absorbed, and the drainage system couldn’t cope with this type of event. Water flooded basements and many shops and restaurants below street level were severely affected. For example, a few computer centers that were located in basements (due to the cooler environment), suffered large losses. More than 90,000 claims were recorded. The insured loss reached DKK 6,200 million, which was more than USD 1 billion at the time. In comparison the annual reported losses due to cloudbursts ranged between DKK 135 million and DKK 997 million for the whole country in the 2006 to 2010 period.
- Copenhagen suffered severe infrastructure disruptions: 10,000 homes had no electricity for up to 12 hours; 50,000 homes had no heating for one week; major thoroughfares were closed for three days.
- Contrary to storm systems, which can be tracked and predicted weeks and days in advance, forecast systems have difficulties in predicting cloudburst cells due to their small size and rapid developments. The development of this very large cloudburst was not predicted by the DMI weather radars.
- Although the city council did have disaster plans, not all were initiated. For instance, the practice at the time dictated not to open the floodgates to the harbour for fear of pollution.
- Although insurers are generally well-organized and have standard routines to cope with catastrophic events, the number of claims presented a problem.
- Insurers have contracts with claims service providers - for example, for pumping out water and providing dehumidifiers - but they were often overburdened by the high demand. Mold from the resulting moisture meant that many insureds suffered increased damage and that insurance companies experienced a substantial loss amplification factor.
- Some policy wordings, particularly for private lines, weren’t clear enough on what was covered.
Could this happen again? In 2014 a cloudburst of similar intensity hit the Copenhagen and Malmø region. Luckily, the cloudburst only affected the northern outskirts, missing the city centre. The most significant downpour took place a few miles off the coast. Despite not being directly comparable in terms of origin and build-up (the 2011 cloudburst came in from the east and the 2014 one was from the south of Copenhagen), it would only have taken a few degrees change in direction to create a substantially larger loss.
The DMI expects the average temperature and rainfall to further increase over the next decades. Cloudbursts are created by atmospheric and surface conditions (granularity and sea surface temperature) and by what is called the urban heating effect. We have observed an increasing frequency of heavy cloudbursts in the Greater Copenhagen area over the last five years, and there is reason to believe that the general increase in temperature will contribute further to this trend in the years to come.
Several initiatives to mitigate the effects of cloudbursts have been taken by a number of entities, and some initiatives are already in force, particularly on the insurance side. Others, particularly concerning infrastructure, will be realized over a number of years.
- The DMI now routinely warns the public and the authorities when conditions for a cloudburst arise, thereby raising awareness and increasing preparedness.
- The city council has adapted contingency plans so that flood gates now will be opened. This measure clearly had a positive effect on the 2014 loss.
- A 580m rainwater tunnel will be constructed to lead the water directly to the ocean (Østerbro) - to be completed by 2017.
- In 2015 a plan was announced for re-dimensioning River Harrestrup Å within 10 to 20 years.
- Insurers have introduced a range of revised terms ranging from higher deductibles to sub-limits and restrictions of coverage. Although these measures undoubtedly had a positive effect in 2014 on the loss factor for insurance companies, and can be modelled to some extent, the actual effect in a similar event is still difficult to substantiate.
We know that cloudbursts can occur and that they still have the potential to produce large insured losses. The 2014 event was a clear reminder. Initiatives will help mitigate similar losses in the future, but some of them will take a long time to implement. If a similar event strikes Copenhagen in the next decade, it is likely that we’ll see smaller but still substantial losses, which may affect the lower end of catastrophe programs; we still do not have reliable predictive cloudburst models in Denmark.
Readers in parts of the world that experience more substantial amounts of rainfall may wonder what the fuss is about. The issue is that the climate seems to be changing in Denmark, and society, policyholders, insurers and reinsurers need to adapt to changed conditions. The Danish insurance market is now in that process. Reinsurers will continue to cover the tail risk, including unforeseen events, and we, as a result, will have to charge a premium that is commensurate with the increasing frequency of cloudbursts.
Forsikring og Pension Beredskabsstyrelsen: Nationalt Risikobillede (NRB)
Danish Meteorological Institute