Comparison of Ground Motion Prediction Equations (GMPE) for Chile and Canada With Recent Chilean Megathust Earthquakes

The ground shaking assessment allows quantifying the hazards associated with the occurrence of earthquakes. Chile and western Canada are two areas that have experienced, and are susceptible to imminent large crustal, in-slab and megathrust earthquakes that can affect the population significantly. In this context, we compare the current GMPEs used in the 2015 National Building Code of Canada and the most recent GMPEs calculated for Chile, with observed accelerations generated by four recent Chilean megathrust earthquakes (MW ≥ 7.7) that have occurred during the past decade, which is essential to quantify how well current models predict observations of major events.We collected the 3-component waveform data of more than 90 stations from the Centro Sismologico Nacional and the Universidad de Chile, and processed them by removing the trend and applying a band-pass filter. Then, for each station, we obtained the Peak Ground Acceleration (PGA), and by using a damped response spectra, we calculated the Pseudo Spectral Acceleration (PSA). Finally, we compared those observations with the most recent Chilean and Canadian GMPEs. Given the lack of geotechnical information for most of the Chilean stations, we also used a new method to obtain the VS30 by inverting the H/V ratios using a trans-dimensional Bayesian inversion, which allows us to improve the correction of observations according to soil conditions.As expected, our results show a good fit between observations and the Chilean GMPEs, but we observe that although the shape of the Canadian GMPEs is coherent with the distribution of observations, in general they under predict the observations for PGA and PSA at shorter periods for most of the considered earthquakes. An example of this can be seen in the attached figure for the case of the 2014 Iquique earthquake.These results present important implications related to the hazards associated to large earthquakes, especially for western Canada, where the probability of a large (MW 9.0) subduction earthquake is about 10-15% in the next 50 years. We discuss the reasons for the observed differences for the Canadian GMPEs, where one of the most relevant factors would be the different dataset used for their calculation.