← Back to Sessions

Sensitivity of Seismic Hazard Models to Catalog Magnitude Conversion Relations

Description: 

The 2023 U.S. National Seismic Hazard Model (NSHM) showed that hazard estimates can be extremely sensitive to changes in magnitude. Hazard is typically computed for earthquakes above a minimum magnitude (e.g., M_min=5), but smaller magnitude earthquakes are used to 1) develop the spatial kernel that forecasts where damaging earthquakes are likely to occur in the future, and 2) determine the Gutenberg-Richter b-value that is used to extrapolate the occurrence rate of the more common smaller events to the rate of M_min earthquakes. Changes in the magnitudes, whether due to revised estimates or changes in how various magnitude types are converted to a uniform moment magnitude, can therefore affect forecasts of both where and how often damaging shaking is expected to occur in the future.

Here we focus on the sensitivity of the gridded seismicity model to the catalog conversion equations in the eastern United States. In the 2023 NSHM, magnitudes of various types were converted to moment magnitudes using equations developed by the Central and Eastern United States Seismic Source Characterization for Nuclear Facilities, based on linear regressions made using data from a catalog containing events up through 2008. We update these regression equations using more recent earthquakes from 2000 through 2023. Also, rather than the least squares regression used previously, we adopt orthogonal regression, following Arabasz et al. (2016), which accounts for uncertainties in both dependent and independent regression variables (observed moment magnitude and alternative magnitude type, respectively). We compare the spatial distribution of annual M≥5 rates, which directly feed into the hazard computation, using 3 different models: 1) the previous 2023 NSHM conversions, 2) our updated conversions, and 3) no conversions at all. We find that the choice of conversions can lead to significant differences in the rate forecasts, which can greatly impact the seismic hazard model, particularly in regions with low seismicity rates like the eastern U.S. where the hazard is dominated by gridded seismicity rather than a fault model.

Session: Advancing Seismic Hazard Models [Poster]

Type: Poster

Date: 4/17/2025

Presentation Time: 08:00 AM (local time)

Presenting Author: Andrea

Student Presenter: No

Invited Presentation: 

Poster Number: 57

Authors