{"id":{"repo_id":"strathclyde","oai_identifier":"oai:strathclyde:d791sg86d"},"canonical_url":"https://search.ndltd.org/etd/strathclyde/oai:strathclyde:d791sg86d","repository":{"repo_id":"strathclyde","name":"University of Strathclyde","base_url":"https://stax.strath.ac.uk/catalog/oai"},"display":{"title":"Improving earthquake risk assessment by using richer hazard inputs and better damage outputs","abstract":"Probabilistic seismic risk assessments are vital tools in understanding and estimating the losses, to people and infrastructure, that earthquakes can cause. There are two main approaches to earthquake risk assessment: unconditional, where risk is calculated directly from observations of a system, and conditional, where risk is estimated through a series of probabilistic steps. Although unconditional methods are considered more robust, they are computationally expensive and so rarely used. In contrast, there have been considerable developments in conditional risk modelling techniques in recent years, with the growing popularity of machine learning spurring innovation.This thesis evaluates potential improvements to both hazard and demand modelling techniques (two stages of the conditional approach) and explores how novel methods and machine learning opportunities can enhance the accuracy of risk estimates. Firstly, the ability of different ground-motion models to improve the accuracy of seismic hazard and risk estimates is investigated. Secondly, the combined selection of seismic intensity measure and demand model is evaluated to find the optimal combination for estimating component level damage assessment. Finally, modelling the covariance of seismic demand estimates is explored, with impacts assessed through loss estimates on a case study structure. The analysis is performed using hypothetical scenarios and stochastic ground motion models, allowing all results to be validated against risk estimates from the unconditional approach.The findings of this thesis help to demonstrate the potential for machine learning to improve earthquake risk assessment practice, whilst also highlighting that traditional practices can often still out-perform the state-of-the-art. Improving the way in which earthquake risks are modelled helps to better understand, prepare for, and protect against, these risks, ultimately resulting in safer communities and structures.","abstract_html":"Probabilistic seismic risk assessments are vital tools in understanding and estimating the losses, to people and infrastructure, that earthquakes can cause. There are two main approaches to earthquake risk assessment: unconditional, where risk is calculated directly from observations of a system, and conditional, where risk is estimated through a series of probabilistic steps. Although unconditional methods are considered more robust, they are computationally expensive and so rarely used. In contrast, there have been considerable developments in conditional risk modelling techniques in recent years, with the growing popularity of machine learning spurring innovation.This thesis evaluates potential improvements to both hazard and demand modelling techniques (two stages of the conditional approach) and explores how novel methods and machine learning opportunities can enhance the accuracy of risk estimates. Firstly, the ability of different ground-motion models to improve the accuracy of seismic hazard and risk estimates is investigated. Secondly, the combined selection of seismic intensity measure and demand model is evaluated to find the optimal combination for estimating component level damage assessment. Finally, modelling the covariance of seismic demand estimates is explored, with impacts assessed through loss estimates on a case study structure. The analysis is performed using hypothetical scenarios and stochastic ground motion models, allowing all results to be validated against risk estimates from the unconditional approach.The findings of this thesis help to demonstrate the potential for machine learning to improve earthquake risk assessment practice, whilst also highlighting that traditional practices can often still out-perform the state-of-the-art. Improving the way in which earthquake risks are modelled helps to better understand, prepare for, and protect against, these risks, ultimately resulting in safer communities and structures.","abstract_has_math":false,"creators":["Rudman, Archie"],"institution":"University of Strathclyde","degree_name":"phd","degree_level":"doctoral-pg","degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":["Douglas, John","Tubaldi, Enrico"],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-19T16:18:06Z","subjects":[],"languages":[],"rights":[],"rights_urls":["https://stax.strath.ac.uk/thesis_copyright_statement/"],"identifier_entries":[{"key":"dc:identifier.doi","label":"DOI","values":["10.48730/3sh7-hf04"],"render_values":[{"text":"10.48730/3sh7-hf04","href":"https://doi.org/10.48730/3sh7-hf04","code":true}]},{"key":"dc:identifier","label":"Identifier","values":["T18015"],"render_values":[{"text":"T18015","href":null,"code":true}]}],"person_identifier_entries":[{"key":"dc:creator.authoridentifier","label":"Author Identifier","values":["202178397"],"render_values":[{"text":"202178397","href":null,"code":true}]}]},"links":{"outbound_url":"https://stax.strath.ac.uk/concern/theses/d791sg86d","outbound_label":"Repository record","outbound_source":"dc:identifier.uri","record_url":"https://stax.strath.ac.uk/concern/theses/d791sg86d","record_label":"Repository record","record_kind":"record","record_source":"dc:identifier.uri","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://stax.strath.ac.uk/catalog/oai?verb=GetRecord&metadataPrefix=uketd_dc&identifier=oai%3Astrathclyde%3Ad791sg86d","prefix":"uketd_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Douglas, John","Tubaldi, Enrico"]},{"key":"dc:creator","label":"Author","values":["Rudman, Archie"]},{"key":"dc:creator.authoridentifier","label":"Author Identifier","values":["202178397"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.issued","label":"Date","values":["2026"]},{"key":"dc:publisher.department","label":"Dc Publisher Department","values":["Department of Civil and Environmental Engineering"]},{"key":"dc:publisher.institution","label":"Dc Publisher Institution","values":["University of Strathclyde"]},{"key":"dc:relation.uri","label":"Dc Relation URI","values":["https://stax.strath.ac.uk/downloads/0c483k13c"]},{"key":"dc:type.qualificationlevel","label":"Dc Type Qualificationlevel","values":["doctoral-pg"]},{"key":"dc:type.qualificationname","label":"Dc Type Qualificationname","values":["phd"]},{"key":"dc:type.uri","label":"Dc Type URI","values":["http://purl.org/coar/resource_type/c_db06"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:rights.uri","label":"Rights URI","values":["https://stax.strath.ac.uk/thesis_copyright_statement/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["T18015"]},{"key":"dc:identifier.doi","label":"DOI","values":["10.48730/3sh7-hf04"]},{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://stax.strath.ac.uk/concern/theses/d791sg86d"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Probabilistic seismic risk assessments are vital tools in understanding and estimating the losses, to people and infrastructure, that earthquakes can cause. There are two main approaches to earthquake risk assessment: unconditional, where risk is calculated directly from observations of a system, and conditional, where risk is estimated through a series of probabilistic steps. Although unconditional methods are considered more robust, they are computationally expensive and so rarely used. In contrast, there have been considerable developments in conditional risk modelling techniques in recent years, with the growing popularity of machine learning spurring innovation.This thesis evaluates potential improvements to both hazard and demand modelling techniques (two stages of the conditional approach) and explores how novel methods and machine learning opportunities can enhance the accuracy of risk estimates. Firstly, the ability of different ground-motion models to improve the accuracy of seismic hazard and risk estimates is investigated. Secondly, the combined selection of seismic intensity measure and demand model is evaluated to find the optimal combination for estimating component level damage assessment. Finally, modelling the covariance of seismic demand estimates is explored, with impacts assessed through loss estimates on a case study structure. The analysis is performed using hypothetical scenarios and stochastic ground motion models, allowing all results to be validated against risk estimates from the unconditional approach.The findings of this thesis help to demonstrate the potential for machine learning to improve earthquake risk assessment practice, whilst also highlighting that traditional practices can often still out-perform the state-of-the-art. Improving the way in which earthquake risks are modelled helps to better understand, prepare for, and protect against, these risks, ultimately resulting in safer communities and structures."]},{"key":"dc:description.abstract","label":"Abstract","values":["Probabilistic seismic risk assessments are vital tools in understanding and estimating the losses, to people and infrastructure, that earthquakes can cause. There are two main approaches to earthquake risk assessment: unconditional, where risk is calculated directly from observations of a system, and conditional, where risk is estimated through a series of probabilistic steps. Although unconditional methods are considered more robust, they are computationally expensive and so rarely used. In contrast, there have been considerable developments in conditional risk modelling techniques in recent years, with the growing popularity of machine learning spurring innovation.This thesis evaluates potential improvements to both hazard and demand modelling techniques (two stages of the conditional approach) and explores how novel methods and machine learning opportunities can enhance the accuracy of risk estimates. Firstly, the ability of different ground-motion models to improve the accuracy of seismic hazard and risk estimates is investigated. Secondly, the combined selection of seismic intensity measure and demand model is evaluated to find the optimal combination for estimating component level damage assessment. Finally, modelling the covariance of seismic demand estimates is explored, with impacts assessed through loss estimates on a case study structure. The analysis is performed using hypothetical scenarios and stochastic ground motion models, allowing all results to be validated against risk estimates from the unconditional approach.The findings of this thesis help to demonstrate the potential for machine learning to improve earthquake risk assessment practice, whilst also highlighting that traditional practices can often still out-perform the state-of-the-art. Improving the way in which earthquake risks are modelled helps to better understand, prepare for, and protect against, these risks, ultimately resulting in safer communities and structures."]},{"key":"dc:title","label":"Title","values":["Improving earthquake risk assessment by using richer hazard inputs and better damage outputs"]}]}],"canonical_facts":{"dc:contributor.advisor":["Douglas, John","Tubaldi, Enrico"],"dc:creator":["Rudman, Archie"],"dc:creator.authoridentifier":["202178397"],"dc:date.issued":["2026"],"dc:description":["Probabilistic seismic risk assessments are vital tools in understanding and estimating the losses, to people and infrastructure, that earthquakes can cause. There are two main approaches to earthquake risk assessment: unconditional, where risk is calculated directly from observations of a system, and conditional, where risk is estimated through a series of probabilistic steps. Although unconditional methods are considered more robust, they are computationally expensive and so rarely used. In contrast, there have been considerable developments in conditional risk modelling techniques in recent years, with the growing popularity of machine learning spurring innovation.This thesis evaluates potential improvements to both hazard and demand modelling techniques (two stages of the conditional approach) and explores how novel methods and machine learning opportunities can enhance the accuracy of risk estimates. Firstly, the ability of different ground-motion models to improve the accuracy of seismic hazard and risk estimates is investigated. Secondly, the combined selection of seismic intensity measure and demand model is evaluated to find the optimal combination for estimating component level damage assessment. Finally, modelling the covariance of seismic demand estimates is explored, with impacts assessed through loss estimates on a case study structure. The analysis is performed using hypothetical scenarios and stochastic ground motion models, allowing all results to be validated against risk estimates from the unconditional approach.The findings of this thesis help to demonstrate the potential for machine learning to improve earthquake risk assessment practice, whilst also highlighting that traditional practices can often still out-perform the state-of-the-art. Improving the way in which earthquake risks are modelled helps to better understand, prepare for, and protect against, these risks, ultimately resulting in safer communities and structures."],"dc:description.abstract":["Probabilistic seismic risk assessments are vital tools in understanding and estimating the losses, to people and infrastructure, that earthquakes can cause. There are two main approaches to earthquake risk assessment: unconditional, where risk is calculated directly from observations of a system, and conditional, where risk is estimated through a series of probabilistic steps. Although unconditional methods are considered more robust, they are computationally expensive and so rarely used. In contrast, there have been considerable developments in conditional risk modelling techniques in recent years, with the growing popularity of machine learning spurring innovation.This thesis evaluates potential improvements to both hazard and demand modelling techniques (two stages of the conditional approach) and explores how novel methods and machine learning opportunities can enhance the accuracy of risk estimates. Firstly, the ability of different ground-motion models to improve the accuracy of seismic hazard and risk estimates is investigated. Secondly, the combined selection of seismic intensity measure and demand model is evaluated to find the optimal combination for estimating component level damage assessment. Finally, modelling the covariance of seismic demand estimates is explored, with impacts assessed through loss estimates on a case study structure. The analysis is performed using hypothetical scenarios and stochastic ground motion models, allowing all results to be validated against risk estimates from the unconditional approach.The findings of this thesis help to demonstrate the potential for machine learning to improve earthquake risk assessment practice, whilst also highlighting that traditional practices can often still out-perform the state-of-the-art. Improving the way in which earthquake risks are modelled helps to better understand, prepare for, and protect against, these risks, ultimately resulting in safer communities and structures."],"dc:identifier":["T18015"],"dc:identifier.doi":["10.48730/3sh7-hf04"],"dc:identifier.uri":["https://stax.strath.ac.uk/concern/theses/d791sg86d"],"dc:publisher.department":["Department of Civil and Environmental Engineering"],"dc:publisher.institution":["University of Strathclyde"],"dc:relation.uri":["https://stax.strath.ac.uk/downloads/0c483k13c"],"dc:rights.uri":["https://stax.strath.ac.uk/thesis_copyright_statement/"],"dc:title":["Improving earthquake risk assessment by using richer hazard inputs and better damage outputs"],"dc:type.qualificationlevel":["doctoral-pg"],"dc:type.qualificationname":["phd"],"dc:type.uri":["http://purl.org/coar/resource_type/c_db06"]},"updated_at":"2026-11-19T16:18:06Z"}