Overview
Introduction
Earthquakes remain among the most damaging natural hazards, with more than one million deaths and nearly a trillion USD in economic losses recorded over the past century. A substantial share of these impacts stems from inadequate performance of buildings not designed to satisfy seismic provisions. This reality makes reliable vulnerability functions essential for producing meaningful loss estimates in seismic risk assessments, thereby shaping how risk drivers are identified.
This documentation describes the GEM Global Seismic Vulnerability Model, a comprehensive database of fragility and vulnerability functions developed by the Global Earthquake Model (GEM) Foundation for seismic risk assessment. The model advances large-scale seismic risk modelling through four key pillars:
Expanded building class coverage: Over 1,000 building classes capturing a wide range of primary vulnerability modifiers such as: primary construction material, lateral load-resisting system, seismic design code and earthquake-resistant design levels and number of storeys
Improved numerical modelling: Efficient multi-degree-of-freedom (MDOF) stick-and-mass models to simulate seismic response at scale for explicit treatment of storey- and floor-level demands
Extensive ground-motion records: A comprehensive suite spanning diverse tectonic environments (active shallow crust, stable continental, and subduction zones)
New damage and loss models: Fragility functions describing the probability of exceeding pre-defined damage states (DSs) and vulnerability functions characterising the expected loss - economic and human-centric - both conditioned on a ground-shaking intensity measure (IM) with explicit treatment of uncertainty
Building Classification
Buildings are categorised based on primary construction materials, lateral load-resisting systems, seismic design code, earthquake-resistant design levels, and height. Each building class is encoded using a unique taxonomy string following GEM’s taxonomy system.
For instance, CR/LFINF/CDL+ERL/H:3/RES identifies a three-storey (H:3) reinforced concrete (CR)
infilled moment-resisting frame (LFINF) with low code design (CDL), low earthquake-resistant design (ERL),
and residential (RES) occupancy.
Attribute |
String |
Details |
|---|---|---|
Construction Material |
CR |
Reinforced Concrete |
E |
Earth Construction |
|
ME+MEO |
Metal Sheet Construction (Informal) |
|
MCF |
Confined Masonry |
|
MR |
Reinforced Masonry |
|
MUR |
Unreinforced Masonry |
|
MUR+ADO |
Adobe Construction |
|
MUR+CBH |
Unreinforced Masonry + Concrete Block |
|
MUR+CLBRH |
Unreinforced Masonry + Hollow Clay Brick |
|
MUR+CLBRS |
Unreinforced Masonry+ Solid Clay Brick |
|
MUR+STDRE |
Unreinforced Masonry + Dressed Stone |
|
MUR+STRUB |
Unreinforced Masonry + Rubble Stone |
|
S |
Steel |
|
SRC |
Concrete Composite with Steel Sections |
|
W |
Unknown Timber |
|
W+WHE |
Heavy or Mass Timber |
|
W+WLI |
Light Timber |
|
W+WBB |
Bamboo |
|
W+WWD |
Wattle and daub |
|
Lateral Load-Resisting System |
LDUAL |
Dual System |
LFINF |
Infilled Moment-Resisting Frame |
|
LFM |
Bare Moment Frame |
|
LPB |
Post and Beam/ Large Panel Building |
|
LWAL |
Load-Bearing Wall/ Shear Wall System |
|
LFBR |
Braced Frame |
|
Code Level |
CDN |
No Code |
CDL |
Low Code |
|
CDM |
Moderate Code |
|
CDH |
High Code |
|
CDS |
Severe Code |
|
Earthquake-Resistant Design |
ERN |
No Earthquake-Resistant Design |
ERL |
Low Earthquake-Resistant Design |
|
ERM |
Moderate Earthquake-Resistant Design |
|
ERH |
High Earthquake-Resistant Design |
|
ERS |
Severe Earthquake-Resistant Design |
|
Height |
H:n |
n is the number of storeys |
Methodology Overview
The methodology adopted to develop the global vulnerability database consists of:
Numerical Modelling of Idealised System: Calibration and modelling of stick-and-mass MDOF systems for over 1,000 building classes
Nonlinear Time-History Analysis: Modified cloud analysis (MCA)-based nonlinear time-history analysis using an extensive suite of unscaled ground motions (or with limited scaling)
Fragility Derivation: Fragility function fitting using probabilistic seismic demand modelling following MCA, characterising structural damage conditional on ground-shaking IMs
Vulnerability Function Assembly: Comprehensive vulnerability functions combining fragility with damage-to-loss ratios and integrating seismic demands (i.e., peak storey drifts and peak floor acceleration) with storey loss functions derived from component-based analysis
How Were the Models Developed?
A complete description of the vulnerability modelling workflow is available in the following sections of the model documentation. The models were developed using the OpenQuake Vulnerability Modeller’s Toolkit OQ-VMTK (Nafeh et al., 2026), a suite of open-access tools created by the vulnerability modellers at the GEM Foundation to provide earthquake engineers with a comprehensive platform for developing fragility and vulnerability models. A demonstration of the capabilities and functionalities of the VMTK is also provided in the OQ-VMTK ‘demos’ folder. A documentation of the OQ-VMTK modules and functionalities is available on: https://gemsciencetools.github.io/oq-vmtk/
Data Access and Recommended Use
All derived functions are released in an open-access repository. The vulnerability functions hosted on this repository are intended for (sub-)national and regional seismic risk assessments, and are designed to quantify loss metrics such as the average annual loss ratio. These functions should be applied only to building classes that are consistent with the taxonomy attributes adopted in their derivation, including construction material, lateral load-resisting system, height range, code level, and ductility. The functions are derived from representative MDOF stick models for each building class that capture the key properties governing structural response, such as material characteristics, structural system, and building height. The country-specific models hosted here are consistent with the master branch of the Global Exposure Model (Yepes-Estrada et al., 2023).
When local building stocks deviate from these archetypes (e.g., due to structural irregularities or region-specific construction practices) in ways that may significantly affect loss estimates, the preferred approach is to re-calibrate the vulnerability functions using local information via the OQ-VMTK (Nafeh et al., 2026), for example by updating capacity curves and/or ground-motion sets. Alternatively, well-justified adjustment factors may be applied to the existing vulnerability functions to account for local conditions.
References
Please cite this work as follows:
Nafeh, A.M.B., Aljawhari, K. & Silva, V. (2025), The 2025 Global Vulnerability Model of the GEM Foundation, GitHub. https://github.com/gem/global_vulnerability_model
Aljawhari, K., Nafeh, A.M.B. & Silva, V. A new global vulnerability model for regional seismic risk assessments: Part 1 – structural vulnerability. Bull Earthquake Eng (2026). https://doi.org/10.1007/s10518-026-02443-7
Nafeh, A.M.B., Aljawhari, K., & Silva, V. A new global vulnerability model for regional seismic risk assessments: Part 2 – nonstructural and building contents vulnerability. In Press