Simulation and Data Lab Statistical & Engineering Seismology
Major Competencies
The Statistical & Engineering Seismology Lab is a research team at the University of Iceland working at the interface of statistical seismology and engineering seismology. We develop transparent, testable, and uncertainty-aware methods that connect near-real-time seismicity information to short-term forecasts and time-dependent hazard and risk metrics, with particular emphasis on Iceland’s tectonic and volcano-tectonic environments. Our work bridges statistical modeling, physical understanding of seismic processes, and operational needs to support rigorous communication of hazard and uncertainty.
High-Performance Computing for Real-Time Earthquake Forecasting
We build HPC-enabled pipelines for robust real-time model updating, ensemble forecasting, large-scale spatiotemporal simulation, and uncertainty propagation. This includes operational workflow design (automation,reproducibility, checkpointing) to keep forecasting systems reliable over longtime horizons.
Physics-Informed Seismology & Ground-Shaking Simulation
We integrate physical understanding of seismic processes into operational products, including physics-based ground-shaking simulation, site effects, and hybrid approaches that incorporate crustal deformation / stress-change information alongside statistical models, especially important in volcano-tectonic settings.
Large Data, Bayesian & Machine Learning Methods for Uncertainty-Aware Modeling
We develop and test Bayesian and ML-assisted models that areexplicitly uncertainty-aware (posterior predictive simulation, calibrated probabilistic outputs), scalable to large earthquake catalogs, and multi-sensor data streams and suitable for transparent prospective testing.
Lab news and events
OEFIS— Operational Earthquake Forecasting System for Southwest Iceland
OEFIS is our flagship effort, funded by the Icelandic Research Fund (RANNÍS), developing an advanced earthquake forecasting model for South Iceland Seismic Zone and Reykjanes Peninsula, where seismicity directly affects populated areas and critical infrastructure. Our work spans the full chain from seismicity, forecasting, to hazard and risk metrics, with an emphasis on computational methods and reproducibility.
What we build and test:
- Spatiotemporal ETAS and ensemble forecasting models for tectonic and volcano-tectonic earthquake sequences
- HPC-enabled pipelines for operational reliability, rapid and large-scale simulation, robust uncertainty propagation and scalable testing
- Machine Learning and Bayesian inference for uncertainty-aware modeling and robust updating
- Hybrid forecasting that combines physics-based inputs (e.g., deformation/stress proxies) with statistical ETAS models
- Time-dependent probabilistic seismic hazard assessment suitable for operational communication
- Model evaluation & benchmarking aligned with CSEP testing practices (Collaboratory for the Study of Earthquake Predictability, https://cseptesting.org/)
Our ultimate objective is to develop a policy-neutral evidence-based seismicity forecasting system that remains scientifically rigorous, explicitly quantifies uncertainty, and is operationally sustainable over long time horizons.
People
Dr. Atefe Darzi
Dr. Benedikt Halldórsson
Dr. Birgir Hrafnkelsson
Pariya Yavarirad
Projects & Cooperations
All IHPC Projects
Selected Publications
· Darzi, A., Halldórsson, B., Cotton, F., & Hrafnkelsson, B. (2024). Feasibility of short-term seismic hazard forecasting for aftershock sequences in southwest Iceland. In World Conference on Earthquake Engineering proceedings.
· Darzi, A., Halldorsson, B., Hrafnkelsson, B., Ebrahimian, H., Jalayer, F., & Vogfjörð, K. S. (2023). Calibration of a Bayesian spatio-temporal ETAS model to the June 2000 South Iceland seismic sequence. Geophysical Journal International, 232(2), 1236–1258. https://doi.org/10.1093/gji/ggac387
· Darzi, A., Hrafnkelsson, B., & Halldorsson,B. (2023). Bayesian Modelling in Engineering Seismology: Spatial Earthquake Magnitude Model. In Statistical Modeling Using Bayesian Latent Gaussian Models (pp. 171–192). Springer. https://doi.org/10.1007/978-3-031-39791-2_5
· Darzi, A., Halldorsson, B., Hrafnkelsson, B.,& Vogfjörð, K. S. (2022). Short-term Bayesian ETAS spatiotemporal forecasting of the Ölfus 2008 earthquake sequence in Iceland. Tectonophysics,839, 229522. https://doi.org/10.1016/j.tecto.2022.229522