Skip to main content
Article | WTW Research Network Newsletter

An integrated end-to-end approach to time-dependent earthquake risk assessment

By Salvatore Iacoletti and James Dalziel | May 11, 2023

Do conventional earthquake risk models only considering mainshock events and using a time-independent earthquake rupture forecast neglect long-term repeated occurrences of large earthquakes?
Risk & Analytics|Marine|Willis Research Network
Disaster Response Center|Insurer Solutions

This year marks the end of a WTW Research Network partnership with Salvatore Iacoletti at UCL, who has been working on time dependant earthquake risk assessment. To celebrate his project and achievements, we wanted to highlight some of the work that has been done by Salvatore, as well as offer some background on his past and future work.

After two years working as a seismic hazard specialist in an engineering consulting firm, Salvatore was motivated to continue his studies through a PhD degree at UCL and deepen his understanding of earthquake catastrophe risk modelling. This educational path and his past industry experience have given Salvatore unique skills to handle challenging seismic hazard quantification and risk analysis applications. Salvatore’s research has already resulted in four journal papers published in highly regarded international journals and some conference papers, one of which was awarded the Best Student Price at ICOSSAR 2021-22.

Project Motivation

Conventional earthquake risk models only consider mainshock events and use a time-independent earthquake rupture forecast to describe the occurrence of mainshocks. This approach neglects long-term repeated occurrences of large earthquakes on specific fault segments, the interaction of adjacent faults, the short-term spatial and temporal clustering of aftershocks, and the accumulation of damage in engineered assets from multiple sequential events.

The WTW Research Network (WRN) and University College London (UCL), UK, are currently tackling these shortcomings, providing scientifically proven and industry-oriented solutions for the (re)insurance and earthquake risk-modelling industries. The WRN’s sponsorship of the PhD work was particularly appealing for Salvatore because of the collaboration between academia and industry and the extra challenge of producing quality research that can be practically implemented. His PhD efforts over the past years have directly addressed these challenges. While at UCL, he has leveraged numerous opportunities to expand his background in fundamental risk analysis tools, taking courses in scientific computing, machine learning, Bayesian data analysis, and advanced structural engineering.

The Proposed Framework

Salvatore Iacoletti, Dr. Gemma Cremen and Prof. Carmine Galasso at UCL developed a unified earthquake risk modelling framework (Figure 1) that incorporates the latest scientific advancements in earthquake interaction, aftershock modelling, and time-dependent vulnerability assessment.

Schematic representation of the unified earthquake risk modelling framework
Figure 1 shows a unified earthquake risk modelling framework

The hazard component of the framework facilitates the modelling of realistic earthquake ruptures that reflect the behaviour of recent damaging events such as the 2016 Mw 6.5 Norcia in central Italy or the 2016 New Zealand Mw 7.8 Kaikōura earthquakes. Long-term time-dependent behaviour of earthquakes and interaction between adjacent faults are built-in features of the hazard module, promising more accurate hazard and risk estimates (Iacoletti et al. 2021). The proposed framework also includes a versatile and flexible methodology to incorporate aftershocks within the generated stochastic event set (Iacoletti et al. 2022a). Case studies focusing on Wellington in New Zealand, central Italy (Iacoletti et al. 2022b), and the Nankai subduction zone in South Japan (Iacoletti et al. 2022c) have been used to test the hazard module’s capabilities, limitations, and sensitivities.

Aftershocks can be more damaging to engineered assets than the corresponding mainshock, partly due to increased asset vulnerability caused by physical damage accumulation during an earthquake sequence (e.g., 2010 Canterbury sequence, New Zealand). Iacoletti et al. (2023, under review) developed a suite of more realistic state-dependent seismic fragility and vulnerability models for a wide range of building taxonomies, leveraging state-of-the-art methods to account for dynamic damage accumulation in structures due to multiple earthquake events (i.e., ground-motion sequences). The vulnerability component of the framework uses damage-dependent vulnerability models to compute economic losses accounting for the damaging effect of multiple ground-motion records.

Latest and ongoing developments

Case-study applications of the unified earthquake risk modelling framework reveal that the inclusion of time dependencies can notably alter economic loss estimations. Example aggregate exceedance probability curves are shown in Figure 2. UCL and WTW are currently conducting detailed sensitivity analyses of economic losses to quantify the framework’s most important contributors to loss variability and develop a set of guidelines for advancing the simplified seismic-hazard approaches currently used in the (re)insurance industry.

Example aggregate exceedance probability curves for a case study in Central Italy. The blue line denotes losses obtained with the conventional risk modelling approach that neglects time dependencies. The red lines denote those obtained from the proposed framework using various levels of time dependencies in the calculations. The solid red line only includes long-term time dependency and interaction between adjacent faults; in contrast, and the dotted red line also includes aftershocks (i.e. short-term time dependency).
Figure 2 shows an example aggregate exceedance probability curves for a case study in central Italy

The unified earthquake risk modelling framework results from continuous discussions and collaborations between UCL researchers and WTW practitioners, who are committed to providing state-of-the-art solutions for clients in the (re)insurance and earthquake risk-modelling industries.

Professor Carmine Galasso, Salvatore’s supervisor during the project, had to say of his candidate; “He always listens carefully and learns from colleagues at WRN involved in the supervisory team. He then develops the ideas and takes them forward, writing superb research plans and carrying the research out at the highest academic standards. Salvatore is also a gifted teacher. As a co-instructor/demonstrator for several classes at UCL, he has an excellent teaching manner, is patient with our students, and (most importantly) inspires them.”

References

Iacoletti, S., Cremen, G., Galasso, C., 2021. Advancements in multi-rupture time-dependent seismic hazard modeling, including fault interaction. Earth-Science Reviews 103650. https://doi.org/10.1016/j.earscirev.2021.103650

Iacoletti, S., Cremen, G., Galasso, C., 2022a. Validation of the Epidemic‐Type Aftershock Sequence (ETAS) Models for Simulation‐Based Seismic Hazard Assessments. Seismological Research Letters. https://doi.org/10.1785/0220210134

Iacoletti, S., Cremen, G., Galasso, C., 2022b. Integrating Long and Short-Term Time Dependencies in Simulation-Based Seismic Hazard Assessments. Earth and Space Science 9, e2022EA002253. https://doi.org/10.1029/2022EA002253

Iacoletti, S., Cremen, G., Tomassetti, U., Galasso, C., 2022c. Unsegmented long-term time-dependent modeling of the Nankai subduction zone (Japan), in: Proceedings of the 12th U.S. National Conference on Earthquake Engineering. Earthquake Engineering Research Institute, Salt Lake City, UT.

Iacoletti, S., Cremen, G., Galasso, C., 2023. Modelling damage accumulation during ground-motion sequences for portfolio seismic loss assessments. (under review)

Authors


Head of Earth Risks Research
email Email

Contact us