Search site

Raising climate awareness and informing policy: flooding

Objectives: This mock case study used dummy data to provide an example of how UKCP09 might be used to raise awareness of the issue of climate change. It also shows how the data can inform policy and long-term strategic decision making in a business risk management context, around capital allocation and portfolio management. It does so by estimating future insured and economic losses from inland flooding at the river-basin catchment level.

How they used UKCP09 dummy data

1. This example provides an initial exploration of how UKCP09 outputs could be used in combination with the RMS UK Inland Flood Model to provide indicative estimates of the potential implications of climate change for future UK flood risk.

2. The methodology involves using the RMS Inland Flood Model to develop a simplified relationship between precipitation and insured/economic loss amount at a river-basin level today (for each season) (i.e. a conditional loss distribution), and then using this to extrapolate the losses to a future climate using UKCP09 precipitation data, to give a sense of the potential range of future losses from flooding in the UK.

3. To ensure that the precipitation information from the projections and the RMS Inland Flood Model are directly comparable, their respective baselines must be normalised, i.e. seasonal-mean and return period precipitation from UKCP09 would be compared to that from the RMS Inland Flood Model and any bias/differences between the baseline precipitation information removed by applying a simple scaling factor.

4. UKCP09 Probability level would be chosen from the probabilistic projections in such a manner as to explore the full shape of the output flood risk probability (or impact) function within the RMS Inland Flood Model. You can achieve this by calculating the future flood risk for sample probability levels of the UKCP09 PDFs and testing the sensitivity for each of the locations, time periods and emissions scenarios being considered.

5. For each of the chosen probability levels, a future loss estimate can be determined by:

· Exchanging today's precipitation anomalies at each return period from the RMS Inland Flood Model with the chosen UKCP09 normalised future precipitation anomalies (from step 2 above) and estimating based on these the implications for future losses using the conditional loss distribution from the RMS UK Inland Flood Model.

· Estimate a series of future average annual loss estimates and exceedance probability curves for each river-basin region, emissions scenario, season and time period.

Potential next steps

Simple adaptation experiments could be run at a river basin level to test the sensitivity of the estimated average annual losses to changes in flood defences, exposure or the vulnerability of buildings.

The use of the 11-member RCM information could also be explored.

Lessons learned

  • The nature of UKCP09 output means that the experiment must assume that the general spatial patterns of precipitation (and the correlation of flood events) will be unchanged, and only their frequency and severity are to be adjusted. This is because of the temporal average (30-year time periods) of UKCP09 outputs.
  • Grid level precipitation information could not be used for the flood risk estimates as the probability distributions for individual grid cells are independent (i.e. spatial correlations are not preserved). Consequently, this also limits the ability to model detailed adaptation scenarios.
  • Similarly, the UKCP09 Weather Generator outputs could not be used, as they are not spatially correlated precipitation estimates across grid squares.
  • Only the pre-prepared aggregations (at the river-basin level in this case) could be used because aggregation of 25 km grid squares cannot be undertaken by the user.
  • A clear understanding of the probabilities is essential.

Find out more

  • Contact details: Dr. Nicola Patmore, Risk Management Solutions Ltd.