← Back to Sessions

Retrospective Evaluation of Weekly UCERF3-ETAS forecasts of California Seismicity

Session: Constructing and Testing Regional and Global Earthquake Forecasts III

Type: Oral

Date: 4/22/2021

Presentation Time: 10:45 AM Pacific

Description: 

Earthquakes are known to exhibit clustering behavior on short time scales causing seismicity rates to vary with time, which implies that seismic hazard can increase immediately following large events. To model this behavior, Uniform California Earthquake Rupture Forecast v.3 with epidemic-type aftershock sequences (UCERF3-ETAS) implements a model for time-dependent seismicity in California that includes the clustering observed in earthquake sequences. UCERF3-ETAS generates synthetic earthquake catalogs whose events retain the spatiotemporal dependencies prescribed by the model, and represents a candidate model for operational earthquake forecasting. We compute UCERF3-ETAS forecasts from 01 January 1986 until 01 January 2020 in disjoint weekly intervals to provide a time-dependent estimate of the statewide seismicity for earthquakes with M > 2.5. Using newly developed consistency tests for synthetic catalogs, we evaluate UCERF3-ETAS, retrospectively, to better understand the uncertainties present in the model. The consistency tests target the seismicity rate, magnitude distribution, and spatial distributions of the forecast. In particular, we evaluate competing modeling assumptions including elastic rebound and the variability in aftershock productivity. To first order we find that UCERF3-ETAS captures the spatiotemporal behavior of seismicity. Quantifying the differences between forecasts and observations is crucial for improving our forecasting models by testing their assumptions, and understanding their uncertainties.

Presenting Author: William Savran

Student Presenter: No

Authors