Towards real-time integration of accelerometers and GPS

Abstract:

Automatic earthquake processing and early warning systems typically use broadband velocity data. In spite of its limitations, the use of velocity data has been widely adopted in seismology because displacement time series are just one integration away, however, the integration process is not exempt from mathematical complexities.
Because broadband instruments have limited dynamic range and operate in an inertial reference frame, it's not possible to recover the full static displacements or very long periods for large earthquakes due to unmodeled rotational motions. Additionally, broadband records saturate in the presence of large local events.
Here we present an implementation of real-time Kalman filtering at the Chilean National Seismological Center (CSN), which optimally combines GPS and strong motion data to produce robust corrected velocity and displacement traces. The Kalman filter method takes advantage of the individual strengths of each data type while minimizing their weaknesses. The calculated traces recover static offsets and very long periods while retaining the same sampling rate as acceleration data.
These very-broadband velocities and displacements can be used for a number of real-time and rapid seismological applications such as better understanding the earthquake rupture process and for improving early warning systems.