CSEP (Collaboratory for the Study of Earthquake Predictability)

CSEP (Collaboratory for the Study of Earthquake Predictability)
Name	Collaboratory for the Study of Earthquake Predictability
Abbreviation	CSEP
Formation	2004
Type	Research collaboratory
Headquarters	Southern California
Fields	Seismology, Statistical Seismology, Geophysics

CSEP (Collaboratory for the Study of Earthquake Predictability) is an international research collaboratory that coordinates prospective testing of earthquake forecasting models, linking observational networks, modeling groups, and evaluation centers to improve operational forecasting and hazard assessment. Founded to formalize hypothesis testing in applied seismology, it brings together institutions, observatories, and research programs to standardize model comparison, data formats, and statistical testing across regional and global scales. CSEP activities intersect with initiatives in seismic hazard mapping, earthquake early warning, and geodetic monitoring through coordinated experiments and open evaluation.

Overview

CSEP connects research groups at institutions such as California Institute of Technology, Massachusetts Institute of Technology, University of California, Berkeley, US Geological Survey, Swiss Seismological Service, University of Auckland, National Research Institute for Earth Science and Disaster Resilience, Tokyo Institute of Technology, University of Tokyo, NASA, European Space Agency, Imperial College London, University of Oxford, University of Cambridge, ETH Zurich, Columbia University, Lamont–Doherty Earth Observatory, Purdue University, University of Southern California, University of Washington, University of British Columbia, National Oceanic and Atmospheric Administration, Seismological Society of America, Geological Survey of Canada, Italian National Institute of Geophysics and Volcanology, Australian National University, University of California, Santa Cruz, Scripps Institution of Oceanography, Stanford University, Brown University, Lehigh University, University of Alaska Fairbanks, University of Colorado Boulder, Los Alamos National Laboratory, Lawrence Berkeley National Laboratory, University of Iceland, Seismological Society of Japan, Korea Institute of Geoscience and Mineral Resources, National Taiwan University, Peking University, Tsinghua University, Indian Institute of Technology Bombay, Universidad Nacional Autónoma de México, Pontifical Catholic University of Chile, University of Chile, Universidad de São Paulo, Federal University of Rio de Janeiro, Southwest Research Institute, NOAA Pacific Marine Environmental Laboratory, Japan Meteorological Agency, Geoscience Australia, British Geological Survey, Institut de Physique du Globe de Paris, Centre National de la Recherche Scientifique, Max Planck Society, German Research Centre for Geosciences, Kobe University, University of Barcelona, University of Lisbon, University of Oslo, University of Helsinki, University of Tokyo, University of Naples Federico II, University of Bologna, Seismological Center of China, Chinese Academy of Sciences].

History and Development

CSEP emerged from collaborations among research programs associated with events and organizations such as the 1994 Northridge earthquake, 1995 Kobe earthquake, 2004 Indian Ocean earthquake and tsunami, 1992 Landers earthquake, 1989 Loma Prieta earthquake, 1999 İzmit earthquake, 2008 Sichuan earthquake, 2011 Tōhoku earthquake and tsunami, Working Group on California Earthquake Probabilities, Collaboratory for Multi-scale Chemical Science, Southern California Earthquake Center, International Association of Seismology and Physics of the Earth’s Interior, and the Collaboratory for the Study of Earthquake Predictability founders from academic, governmental, and international observatories. Early development built on methodologies and debates exemplified by Utsu Omori, Gutenberg–Richter law, Omori's law, Reasenberg and Jones model, Charles F. Richter, Beno Gutenberg, Kiyoo Mogi, Roger Bilham, Lee Bollinger, Thomas H. Jordan, and institutional discussions at venues like American Geophysical Union meetings, European Geosciences Union assemblies, and workshops hosted by International Seismological Centre.

Scientific Objectives and Methodologies

CSEP’s objectives align with statistical testing and model comparison traditions embodied by frameworks from Neyman–Pearson lemma-inspired hypothesis testing, information criteria influenced by Akaike Information Criterion, and model-selection debates reminiscent of work at Royal Statistical Society and Institute of Mathematical Statistics symposia. Methodologies include prospective testing protocols, null hypotheses drawn from catalogs maintained by agencies like USGS National Earthquake Information Center, declustering approaches related to Gardner and Knopoff, rate-and-state physics influenced by Dieter Helmstaedt-style friction laws, and stochastic point-process modeling such as Epidemic-Type Aftershock Sequence models developed from ideas by Y. Ogata and others. CSEP employs formal evaluation metrics analogous to practices at Association for Computing Machinery, American Statistical Association, and statistical-skill-score traditions from World Meteorological Organization forecasting comparisons.

Forecasting Experiments and Projects

CSEP hosts regional and global experiments mirroring organizational efforts seen in Collaboratory for the Study of Earthquake Predictability-linked trials, including regional forecasting tests for regions like California, Italy, Japan, New Zealand, Turkey, Greece, Chile, Mexico, Iceland, China, Taiwan, Alaska, Nepal, and for global catalogs such as Global Centroid Moment Tensor Project compilations. Major projects include retrospective benchmarking that connected methods from Reasenberg and Jones and ETAS variants, time-dependent forecasting trials comparable to exercises at International Seismological Centre, long-term rate experiments resembling work by the Working Group on California Earthquake Probabilities, and short-term forecasting initiatives aligned with research by USGS Earthquake Hazards Program and academic groups at Caltech and UC Berkeley.

Results and Impact on Seismology

CSEP has clarified strengths and limitations of model classes in ways influential to agencies like USGS, Italian National Institute of Geophysics and Volcanology, Geoscience Australia, and policy discussions at United Nations Office for Disaster Risk Reduction, International Council for Science, and World Bank loss-estimation studies. Peer-reviewed syntheses published in venues including Bulletin of the Seismological Society of America, Journal of Geophysical Research, Geophysical Research Letters, Nature Communications, Science Advances, Proceedings of the National Academy of Sciences, and Earth and Planetary Science Letters have documented how prospective testing altered adoption of ETAS variants, time-independent hazard models, and ensemble approaches. Results influenced operational systems such as regional forecasting dashboards, early warning trials at ShakeAlert, seismic risk assessments for infrastructure projects overseen by Federal Emergency Management Agency, National Institute of Standards and Technology, and frameworks used by insurers like Munich Re and Swiss Re.

Organizational Structure and Partnerships

CSEP’s governance model resembles consortia frameworks used by entities like Southern California Earthquake Center, International Seismological Centre, Incorporated Research Institutions for Seismology, European Plate Observing System, and Global Seismographic Network, with steering committees composed of scientists from Caltech, USGS, SCEC, ETH Zurich, University of Tokyo, and partner observatories. Funding and partnership sources include national research agencies such as National Science Foundation, Japan Society for the Promotion of Science, European Research Council, Natural Sciences and Engineering Research Council of Canada, Australian Research Council, and intergovernmental collaborations involving UNESCO-supported programs.

Data, Software, and Evaluation Frameworks

CSEP maintains standardized data formats and software infrastructures inspired by projects like QuakeML, ObsPy, SeisComP, IRIS, ComCat, and catalog maintenance practices at Global Centroid Moment Tensor Project, International Seismological Centre, and USGS NEIC. Evaluation frameworks use open-source tools and statistical libraries analogous to those in R Project for Statistical Computing, Python Software Foundation ecosystems, MATLAB, and high-performance computing resources similar to clusters at National Center for Atmospheric Research. The collaboratory’s reproducible-testing ethos parallels open-science initiatives at Open Science Framework, Zenodo, and data-sharing practices promoted by Science Commons.

Category:Seismology