1 / 23

European wind storms and reinsurance loss: New estimates of the risk

European wind storms and reinsurance loss: New estimates of the risk. Paul Della-Marta, Mark Liniger, Christof Appenzeller, David Bresch, Pamela Köllner-Heck, Veruska Muccione. ACRE Meeting, Zurich, 24.06.2008. Outline. Overview of wind storm risk assessment

alarice
Download Presentation

European wind storms and reinsurance loss: New estimates of the risk

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. European wind storms and reinsurance loss: New estimates of the risk Paul Della-Marta, Mark Liniger, Christof Appenzeller, David Bresch, Pamela Köllner-Heck, Veruska Muccione ACRE Meeting, Zurich, 24.06.2008

  2. Outline • Overview of wind storm risk assessment • European wind storm climate from 1880 (EMULATE) • The PreWiStoR project • Improved estimates of European wind storm climate • Improved estimates of wind storm loss • Conclusions • How can ACRE improve wind storm risk assessment?

  3. Wind Storms in Europe: What is the risk?

  4. How can we estimate the Risk? : Data In-Situ Model

  5. Data example: Daria, 26.01.1990 • Maximum wind speed at each gridpoint over the duration of the storm m/s SwissRe: • MSLP • In-situ wind • Dynamics ERA-40 • Geostr. wind @850hPa • Assimilated obs. *not wind • Dynamics

  6. How can we estimate the Risk? • Extreme Value Theory • Method of Estimation • e.g. Maximum likelihood, L-Moments • Quantification of Uncertainty • Model parameters, return periods • Model diagnostics • Goodness-of-fit • Maximal use of information • Peak Over Threshold instead of Block Maxima See Coles (2001)

  7. The wind climate of Europe 1880-2003 • Geostrophic Wind Speed derived from daily MSLP EMULATE (Ansell et al 2006) (over land only, ONDJFMA)

  8. Storm Selection Method Scalar wind statistic 95% threshold Winter 1999/2000

  9. The storm climate of Europe 1880-2001 • Derived from daily MSLP EMULATE data 1957-2002 1880-2002

  10. Change in return period and confidence using EMULATE data • Land areas only 1957-2002 1880-2002

  11. PreWiStoR: Prediction of winter Wind Storm Risk • Problem: Observed records of wind storms are not long enough • Solution: ~150 storms based on observations. • Use probabilistic modelling to generate synthetic storms based on perturbed statistics • Calculate losses • New approach to use ENSEMBLE prediction systems (seasonal to decadal, s2d) • Replace statistical perturbation with physics • Utilise around ~500 seasons of s2d data • Obtain a better estimate of wind storm risk and losses See van den Brink et al. IJC (2005)

  12. The concept of the PreWiStoR project • Let Lorenz attractor represent the possible trajectories of extreme wind related weather in the current climate • The envelope (climate) of trajectories of the system observed in ERA-40 (45years) • The climate of trajectories sampled in s2d data (300+ years) Source: Wikipedia

  13. PreWiStoR: Data • Seasonal to decadal (s2d) climate prediction models • Using the seasonal forecasting model of the ECMWF • A coupled ocean-atmosphere Global Circulation Model • 6-7 month forecast, T159, 26 years hindcast 11 (41) member ensemble • First month removed to ensure independence • Separate ocean analysis system to initiate the seasonal forecasts • ENSEMBLE prediction system: Model is run many times  Initial conditions are perturbed  Probabilistic Forecasts

  14. Comparison of wind storm climatologies from s2d • Wind storm climatologies are different in magnitude and shape and frequency • All s2d models seem to have a less negative shape than ERA-40

  15. Percentile calibration and sampling 45 year periods • SYS3 has a more extreme wind climate than ERA-40 • SYS2 has a similar extreme wind climate to ERA-40 • After application of calibration • Which one is closer to the true climate? • Can we tell which one is correct?

  16. Sampling experiments with s2d climatologies • Randomly order the 315 years of s2d winter seasons • Plot the estimated shape parameter as a function of the number of seasons used in the calc. • Repeat many times to simulate chaotic inter-annual variability Shape (& Scale) parameter has not converged at most (~70%) 45 year length climatologies  ERA-40 underestimates European storm climate See Vannitsem Tellus (2007)

  17. Sampling experiments with s2d climatologies • Randomly order the 315 years of s2d winter seasons • Plot the estimated shape parameter as a function of the number of seasons used in the calc. • Repeat many times to simulate chaotic inter-annual variability Shape (& Scale) parameter has not converged at most (~70%) 45 year length climatologies  ERA-40 underestimates European storm climate See Vannitsem Tellus (2007)

  18. Swiss Re Wind Storm Loss Model(catXos) • Vulnerability curve shows a cubic relation which is capped • Portfolio value distribution is inhomogeous

  19. Range of loss uncertainty due to sampling 45 year periods • Swiss Re estimates of expected loss-frequency fit within the range of sampling uncertainty • SYS3 shows the added value of using a longer dataset

  20. Conclusions • Using EMULATE data we can improve our estimates of wind storm risk longer climatology provides greater confidence • EMULATE data represents a 24hr mean MSLP  not ideal for wind storms  Geostrophic approximations • Ensemble dynamical forecasts also improve storm risk estimates Evidence that ERA-40 underestimates the risk • Dynamical models have biases and other deficientcies • Use of s2d data has replaced statistical perturbation of storms (SwissRE) with dynamical perturbations (s2d) • Sampling experiments lead to greater insight to risk of storms and loss uncertainty

  21. How can ACRE improve wind storm risk assessment? • Longer reanalyses  sample more of the dynamics • Longer ensemble based reanalyses  sample more of the dynamics + account for observation errors • Longer reanalyses can be used to provide a greater range of initial conditions • S2d data only have a hindcast of ~ 30 years  limited sample of initial conditions • But to be most useful we need: • High temporal resolution output variables and/or integrated quantities e.g. Max. Wind Gust • No just event-based but a full climatology else frequencies are hard to define

  22. A bivariate extreme value peak over threshold model for wind storm intensity and loss • Using the methodology in Coles (2001) and the evd R -package • Fitted to ERA40 wind storm Q95 and the transformed %TIV • Could be used to define the vulnerability with real loss data

More Related