The 2019 Ridgecrest sequence. Adaptive slip imaging with variable rake. How well can we describe the slip details without loosing resolution?

In the last years, the great number of available satellite and GPS data allows to perform routinely, almost real-time computations of source parameters after the occurrence of moderate to big seismic sequences. In the last years we developed and implement a free to use algorithm, which easily performs a standard two steps linear non-linear inversion to retrieve fault geometries and slip distribution. The major novelty of our method with respect to the majority of the inversion approaches is the determination of a fault subdivision which keeps into account the real resolution power of the available geodetic datasets. Our code keeps the resolution power as high as possible, while maximizing the number of patches in the fault subdivision. The goodness of resolution is mathematically determined with the normalized Dirichlet Spread Function, which quantifies the differences between the obtained resolution and the ideal perfect resolution.

Furthermore, in the last implementation of the algorithm, we also let the rake free to vary, adding further information to the imaged slip pattern on main faults. As a consequence, the number of slip patches needed for a good data replication is further reduced.

We apply our approach to the July 2019 sequence in California, where a foreshock of 6.4 of magnitude preceded the mainshock of Mw 7.1 in a desert area near the town of Ridgecrest. The main events faults activated previously unmapped faults.

We show the dramatic reduction of slip details with depth due to the geometry of fault planes of the major events, as already emphasized previously for near vertical fault planes.

As a side analysis we check the interaction between the activated faults and the mapped surrounding structures, by computing the static variation of the static Coulomb stress.