EGU2020-20647
https://doi.org/10.5194/egusphere-egu2020-20647
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Climate change, historical data and catastrophe modelling

Richard Dixon1,2, Sam Franklin3, Len Shaffrey2, and Debbie Clifford3
Richard Dixon et al.
  • 1CatInsight, London, United Kingdom (richard@catinsight.co.uk)
  • 2Department of Meteorology, University of Reading, Reading, United Kingdom
  • 3Institute for Environmental Analytics, Reading, United Kingdom

This presentation will discuss climate change in the context of catastrophe modelling and tail risk. Given that the catastrophe modelling industry typical only has short historical records that provide limited information as to whether hazard is non-stationary, what are the methods and datasets that may aid the catastrophe modelling community to better understand how and whether risk is changing temporally? 

The issues will be framed by using examples of output from a multi-year multi-ensemble 60km global climate simulation, where extra-tropical windstorm daily maximum gust data has been converted into yearly aggregate European insurance loss with the help of PERILS European industry exposure data. The data is used to show how reliance on single historical datasets can produce misleading trends in catastrophe losses - but also potentially point to underlying trends in risk that single historical datasets may not be able to detect.

How to cite: Dixon, R., Franklin, S., Shaffrey, L., and Clifford, D.: Climate change, historical data and catastrophe modelling, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-20647, https://doi.org/10.5194/egusphere-egu2020-20647, 2020

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 1 – uploaded on 29 Apr 2020
  • CC1: Comment on EGU2020-20647, Punit Bhola, 06 May 2020

    Dear Authors,

    Really nice presentation, I have only one question

    What is the rationale behind using two separate hydraulic models and do they have same model structure ? Introducing another model requires seperate calibration/validation and brings incosistency in the modelling results, especially for events that are both pluvial and fluvial. As far as I know MIKE FLOOD can model both fluvial-pluvial and has no restriction with run-time (provided GPU licence) so why not use only one.

    Best regards,

    Punit

    • CC2: Reply to CC1, Punit Bhola, 06 May 2020

      I am really sorry, please ignore the comment, it was not intended for another presentation. My mistake