Reducing Nonuniqueness in Finite Source Inversion Using Rotational Ground Motions
Friday, 19 December 2014
We assess the potential of rotational ground motion recordings to reduce nonuniqueness in kinematic source inversions, with emphasis on the required measurement accuracy of currently developed rotation sensors. Our analysis is based on synthetic Bayesian finite source inversions that avoid linearizations and provide a comprehensive quantification of uncertainties and trade-offs. Using the fault and receiver geometry of the Tottori 2000 earthquake as a testbed, we perform inversions for two scenarios:
In scenario I, we use translational velocity recordings only. In scenario II, we randomly replace half of the velocity recordings by rotation recordings, thus keeping the total amount of data constant. To quantify the noise-dependent impact of rotation recordings, we perform a sequence of inversions with varying noise levels of rotations relative to translations.
Our results indicate that the incorporation of rotational ground motion recordings can significantly reduce nonuniqueness in finite source inversions, provided that measurement uncertainties are similar to or below the uncertainties of translational velocity recordings. When this condition is met, rupture velocity and rise time benefit most from rotation data. The trade-offs between both parameters are then strongly reduced, and the information gain nearly doubles. This suggests that rotational ground motion recordings may improve secondary inferences that rely on accurate information about rise time and rupture velocity. These include frictional properties of the fault, radiation directivity and ground motion in general.