Assessing the predictive accuracy of earthquake strike angle estimates using nonparametric Hawkes processes

Earthquake focal mechanism estimates have been posited to have predictive value for forecasting future seismicity. In particular, for strike‐slip earthquakes, aftershocks should occur roughly along the estimated mainshock strike. However, the errors in such estimated strike angles are considerable. We compare the degree to which estimated strike angles forecast the direction of future seismicity around a given earthquake with that of uniformly distributed angles and with strike angles estimated based on previous seismicity. The fit of nonparametrically estimated Hawkes models using the estimated strike angle that best fits the post‐mainshock set of events for each mainshock is compared with that of corresponding models that exclude these estimates. Strike angle estimates are shown to have marginal predictive value for forecasting the direction of future seismicity but no more than the better fitting of a uniformly distributed angle and its complement.