Organization: ImageCat & GFDRR
- Ron Eguchi, ImageCat
- Alanna Simpson, GFDRR
- Ron Eguchi, ImageCat (chair)
- Kelvin Berryman, GNS Science
- David Lallemant, Stanford University
- Keiko Saito, GFDRR
- John Bevington, ImageCat
- Fumio Yamazaki, Graduate School of Engineering Chiba University, Japan
Risk models are built on the best available but often less than ideal data which includes our understanding of the physical expression of hazards, engineering design and construction, etc. and many other factors. It is only when a disaster occurs that we can retrospectively assess how well our risk models performed in predicting the extent and magnitude of disaster impacts. In many cases, our models surprise us with their accuracy, but more often, the scale of the disaster is over or under estimated. Post-disaster forensics offer us a critical path for determining why our risk models fail; however, is this information being effectively utilized to improve our risk models?
This interactive session highlighted cases where risk models were effective in approximating the extent of a disaster damage and where they fell short. Panelists examined the reasons for model efficacy.
By Ron Eguchi, ImageCat