Flood modelling has come a long way - but has a long way to go
Flooding accounts for both one third of national catastrophes, and one third of associated economic loss. Over the next twenty years these proportions, much like the water level itself, will continue to rise.
Yet due to its complexity, catastrophe flood modelling has developed relatively late compared to other perils.
At a recent Catastrophe Modelling Conference, experts said that a critical mass of events, reaching back to the European flood of 2002 (15bn euros of economic loss – only 2bn of which was insured) woke people up to the dangers of flooding, and the virtues of modelling.
Since then, understanding in the marketplace has increased exponentially. In the last three years alone, the number of flood modellers in London has swelled from a few dozen to almost 500.
Recent events – and the variance in the findings of the models themselves – suggests that each and every one of them is needed.
The flood model waters are muddied by a number of factors.
“Despite wide range of methodologies, models rely on one thing: information.
The correlations between rainfall and flooding, and water level and flooding are not absolute.
Nor is relationship between damage intensity and water depth. Pollution, duration of exposure and velocity are also key components.
There is the not so small matter off-flood plain losses, including blocked drains, which can account for as much as 80% of total losses. No model can, as yet, capture the intricacy of this variable.
Flood defences are also extremely difficult to map on a national scale because of the locality of their impact, and the enormous variance in their effectiveness.
And business interruption is notoriously difficult to assess, let alone predict.
As a result, the range of uncertainty for flood models can be as high as 20%.
“The insurance industry must continue develop proactive, not reactive solutions to manage the risk.
Nonetheless, industry experts are confident this figure can be reduced.
Despite wide range of methodologies, models rely on one thing: information.
Access to and use of high resolution data is critical. The unwieldy act of postcoding flood risk is giving way to modelling based on a grid resolution of just 50 square metres. The problem is that digital terrain models are extremely expensive to use, and the data available is old because of commercial confidentiality.
It is believed that greater reference to claims data can help offset this gap, as will increased interaction between scientists and authorities.
Every region demands its own solution, but this does not mean that the findings from disparate regions cannot form integrated solutions. Indeed, consistencies between models are emerging.
The question is haw far the stakeholders will want to take the issue.
In the meantime, as insured losses continue to escalate, the insurance industry must continue develop proactive, not reactive solutions to manage the risk.
