Generation and validation of broadband synthetic P waves in semistochastic models of large earthquakes

We present an approach for generating stochastic scenario rupture models and semistochastic broadband seismic waveforms that include validated P waves, an important feature for application to early warning systems testing. There are few observations of large magnitude earthquakes available for development and refinement of early warning procedures; thus, simulated data are a valuable supplement. We demonstrate the advantage of using the Karhunen-Loeve expansion method for generating stochastic scenario rupture models, as it allows the user to build in desired spatial qualities, such as a slip inversion, as a mean background slip model. For waveform computation, we employ a deterministic approach at low frequencies (<1 Hz) and a semistochastic approach at high frequencies (>1 Hz). Our approach follows Graves and Pitarka (2010) and extends to model P waves. We present the first validation of semistochastic broadband P waves, comparing our waveforms against observations of the 2014 Mw 8.1 Iquique, Chile, earthquake in the time domain and across frequencies of interest. We then consider the P waves in greater detail, using a set of synthetic waveforms generated for scenario ruptures in the Cascadia subduction zone. We confirm that the time-dependent synthetic P-wave amplitude growth is consistent with previous analyses and demonstrate how the data could be used to simulate earthquake early warning procedures.