A systematic protocol to evaluate earthquake vulnerability models
Project ID: 2531bc1602
(You will need this ID for your application)
Research Theme: Engineering
Research Area(s):
Built environment
Infrastructure and urban systems
Structural engineering
UCL Lead department: Risk and Disaster Reduction
Lead Supervisor: Roberto Gentile
Partner Organisation: Willis Limited
Stipend enhancement: £ 1,000
Project Summary:
In 2023, earthquakes alone were responsible for ~63,000 deaths (~85% of all disaster fatalities) and likely inflicted tens of billions of US dollars in damages globally. Accurate models to estimate future losses are fundamental towards effective risk mitigation. For an earthquake-prone area, risk models first characterise exposure (e.g., locate buildings and assign properties such as material and height). Hazard models quantify the probability of achieving/exceeding different values of intensity measures (e.g., ground accelerations). Vulnerability models are then adopted to quantify the consequences (e.g., economic loss, human displacement) of earthquake of different intensities on physical assets. Each module must be evaluated to guarantee sufficiently accurate risk estimates that can support decision making (e.g., asset retrofit, early warning, insurance pricing).
Exposure models are normally validated and improved increasing the amount of asset data (e.g., census, cadastral, satellite images). Hazard models are evaluated reproducing selected past earthquake scenarios and using ground motion station data -abundant and public- as a benchmark. However, there are only a few studies involving the evaluation of vulnerability models, and the scientific literature does not provide an agreed protocol to do so.
With a focus on building assets, this PhD project will tackle the above research gaps by developing:
- A protocol to benchmark vulnerability models using damage data observed in past earthquakes – when such data is available.
- An evaluation protocol based on systematically comparing a candidate vulnerability model to alternatives in the literature – when observed damage data is not available.
- A review of earthquake vulnerability models to enable the above.
The ideal candidate for this project is familiar with probabilistic risk assessment and has minimum coding/programming skills. Experience in the catastrophe risk modelling and/or insurance sectors is welcome but not mandatory. Upon completion, the candidate will become a leader in earthquake risk modelling and catastrophe risk modelling.