A High-Frequency Distance Metric in Ground-Motion Prediction Equations Based on Seismic Array Backprojections

Typical ground-motion prediction equations (GMPEs) measure source-to-site distances relative to the closest point on the rupture plane (R-rup). However, for megathrust earthquakes (M-w > 8), the oversimplification of the earthquake source characteristics in distance metrics results in significant bias. Recent studies suggest that the high-frequency (HF) and low-frequency (LF) energy tend to emanate from different portions of the megathrusts. This phenomenon motivates an alternative distance metric based on the array backprojection imaging technique that effectively captures regions releasing HF energy. Herein, we define an HF distance metric (R-hf) as the distance from the site to the high-frequency radiation zone. We study five M-w > 7.2 megathrust earthquakes in Japan and Chile and find that R-hf outperforms R-rup in predicting the ground shaking intensity between 0.5 and 4 Hz. We consider R-hf as a complementary measure to conventional GMPE distance metrics and a more accurate ground-motion predictor in many cases. Plain Language Summary Traditional distance metrics for GMPEs are frequency-independent, which perform reasonably well in predicting the shaking intensity of small earthquakes. However, the frequency dependence of source distances becomes more significant for large subduction-zone earthquakes. Our paper proposes a new frequency-dependent distance metric based on backprojection imaging technique. It improves the accuracy of GMPE prediction for the large megathrust earthquakes, especially for the high-frequency ground motion. We study five large subduction zone earthquakes in Japan and Chile: 2011 M-w 9.0 Tohoku, 2003 M-w 8.0 Tokachi, 2005 M-w 7.2 off-Miyagi earthquakes, 2014 M-w 8.1 Iquique earthquake, and 2015 M-w 8.3 Illapel earthquake. In these five earthquakes, our frequency-dependent distance metric outperforms traditional distance metrics.