Having catastrophe model developers, evaluators, users and regulators in the same room can be a great opportunity to tackle key challenges in assessing flood risk. The link between industry and research is not always one that every insurance company can make, which is why Willis Re recently convened a meeting of these professionals, known as the Willis Re Flood Club. While the discussion centred on harnessing new data sources, current best practices and professional expertise and partnerships needed to advance research and risk management, the focus was on Earth observation data and the impact of climate change.
1. How advances in Earth observations help enhance our view of flood risk
Can current satellites observe flood depths? How can we work out how fast a flood spreads across a river network and is it possible to capture “flash floods,” which can develop quickly and with little warning? Present data and processing techniques may not be able to answer all of these questions, but methods are evolving to derive the necessary information for better risk management:
- Higher resolution imagery to inform the disaggregation of aggregate exposure data, especially in remote regions, to improve modeling of exposed assets
- Increasing coverage and repeat times to allow capturing of reliable flood extents, using both optical and radar technology, for event response and creating historical flood footprints for model development, evaluation and also for portfolio stress-testing
There are a number of challenges, however. International effort is required to use available technology for highly detailed elevation models (DEMs) to create and maintain global coverage for the needs of risk management and emergency responders. And accessibility remains challenging for commercial applications compared to supporting research, governmental agencies and emergency responders.
Although pricing is becoming more competitive, we are still seeing a low take-up of available solutions due to high costs. Shared services have been a success story, as exemplified by the Lloyds market working with McKenzie Intelligent Services (MIS) to deliver real-time, imagery-derived intelligence across Lloyds syndicates during the recent hurricane seasons and wildfires in the US.
Remaining data limitations can be overcome by combining a variety of data sources, such as ground truth data for benchmarking, crowd sourcing and citizen science projects, as well as combining data from different sensors such as optical and radar.
2. The impact of climate change on flood risk and catastrophe modeling
The key question on how to incorporate climate change impact on flood models is whether to use statistical or physics-based approaches.
Climate change will probably increase flood risk, however quantification is difficult and swamped by uncertainty from climate model parameters and natural behavior, including cyclical climate patterns. As the Willis Research Network (WRN) Senior Academic, Professor Chris Kilsby of Newcastle University told the Flood Club, “uncertainty is the rule, not the exception”. He also pointed out that rare events, defined by relatively short catalogues, mean that stationarity cannot be completely dismissed.
Catastrophe models can help to quantify the impact of climate change, combining changes in exposure and vulnerability with changes in hazards associated with climate change. “Simple perturbations of current models seem more useful than elaborate event sets when it comes to incorporating climate change in flood models,” said Stefan Eppert of KatRisk. However, it is important that the perturbations are consistent with the narrative around possible future climates.
The re/insurance market is ready to embrace academic findings and new technology for risk management. It is clear that there is a surge of interest in climate risk in the financial sector, and in catastrophe modeling. There is a need to create representative scenarios to capture the climate change impact on flood risk from governments and academia, which means catastrophe modellers could readily implement those in flood models. Given the short period of interest in the re/insurance industry which largely operates on 12-month renewal cycles, whether new tools designed to capture the climate change impact on flood risk are fit for purpose and applicable to the insurance process, is still a topic for further discussion.
Nalan Cabi is a Lead Flood Specialist with Willis Re.
Tina Thomson is Head of Catastrophe Analytics for EMEA (West and South) at Willis Re and a fellow of the Royal Geographical Society.
Geoffrey Saville is a member of Willis Towers Watson’s Analytics Technology Team and Willis Research Network’s Atmospheric Hub Leader.