Researchers Evaluate Earthquake Forecast Models in California

A new study analyzes 51,625 seismic forecasts to determine the reliability of predictive modeling across the state.

Updated on Oct. 29, 2026 in Geology

Isometric editorial illustration of a sedimentary rock layer and a small seismic sensor, evoking geological predictive modeling research. — Researchers from the Collaboratory for the Study of Earthquake Predictability evaluated 27 forecasting models against data from 597 seismic events in California. AI Illustration. Upload story photo >

Live Poll

Do you trust current earthquake forecasting models to accurately predict seismic risks in your area?

Researchers recently published a study in Nature Communications evaluating 27 earthquake forecasting models based on activity in California between 2007 and 2018. The analysis compared these models against 597 recorded earthquakes of magnitude 3.95 or greater.

Why it matters

The study aims to identify which forecasting models can reliably describe patterns in seismic activity. This evaluation provides critical insight into the accuracy of predictive tools used to track earthquake frequency, magnitude, and location.

The study analyzed 51,625 forecasts against 597 earthquakes of at least magnitude 3.95. While 95% of models were consistent with the total number and magnitude of events, only about two-thirds correctly matched the specific locations of the quakes.

The players

University of Bristol

This public research university located in the United Kingdom led the publication of the findings.

Collaboratory for the Study of Earthquake Predictability

This research organization specializes in the prospective testing and evaluation of earthquake forecasting models.

The details

Conducted by the Collaboratory for the Study of Earthquake Predictability, the experiment compared model outputs with actual seismic data recorded within 24 hours of each forecast. Researchers assessed these models to determine their effectiveness in mapping geological patterns across California.

Testing for the forecasting models began in August 2007.
The data collection period for the study concluded in August 2018.
Findings were published in the journal Nature Communications in August 2026.
The University of Bristol released the study news on October 15, 2026.

The Big Picture

This study follows the established protocols for prospective model testing maintained by the Collaboratory for the Study of Earthquake Predictability. By validating these models, the research helps shift the discipline toward more rigorous, evidence-based standards for seismic forecasting.

The validation of these models could lead to more accurate seismic hazard assessments for the California public. Reliable forecasting helps scientists better understand the potential for future seismic events and their impact on regional infrastructure.

The takeaway

Reliable seismic forecasting depends on a model's ability to accurately project both the magnitude and the specific location of a potential earthquake. While most models perform well on broad activity metrics, local precision remains a significant challenge for scientists.