What the exercise of the SPICE source inversion validation BlindTest 1 did not tell you

Abstract

SUMMARY Uncertainties of earthquake finite‐fault inversions based upon strong motion data are investigated using the source inversion validation BlindTest 1 exercise of the SPICE (Seismic Wave Propagation and Imaging in Complex Media: A European Network) project, motivated by previous counterintuitive results. The distribution of slip and the shapes of asymmetric slip rate functions are simultaneously inverted by matching 10 or 33 broad‐band three‐component velocity waveforms within the period ranging from 0.02 to 2 Hz, using a finite‐fault method that carries out the waveform inversion in the wavelet domain. The effects of subfault size, data noise and the number of stations have been explored. Our results suggest: (1) Although there are inevitable discrepancies between the inverted model and the target model because of ignoring the spatial slip variations within individual subfaults, the fault slip and rise‐time distributions can be well constrained even with the data including large Gaussian noise. (2) It is crucial for source studies to develop new inversion schemes that can properly honour the frequency‐ and time‐dependent energy distribution of seismic radiation and data noise. For instance, inversions using the variance reduction function of velocity waveforms as the objective function have low sensitivities to the total seismic moment and peak slip. (3) Although the relative value of the objective function is guided in the inversion, the absolute value of the objective function cannot be used to evaluate the quality of an inverted model. (4) Because the source inversion is based on surface observations, the spatiotemporal resolution of source inversion is affected not only by the data quality but also by the earthquake itself. For vertical strike‐slip faults, the along‐strike resolution is better than that along the downdip direction.

Uncertainty in Earthquake Source Imaging Due to Variations in Source Time Function and Earth Structure

“…We also test the effects on the inversion when using different STFs, an isosceles triangle and two regularized Yoffe functions with different acceleration times. The fits to the data of the synthetics produced by these rupture models give a variance reduction around 94%, lower than reported by Shao and Ji (2012), who reported a variance reduction of about 99% for the initial blind test of the source inversion validation exercise. The reason for this difference is that the true model in our case is based on a dynamic rupture simulation, whose temporal properties are difficult to recover in detail by a kinematic source inversion (Konca et al, 2013); the initial blind test featured a simple kinematic rupture as reference rupture model.…”

“…13 in which n is the number of points on the fault, R j and I j represent the reference and inverted rupture models, respectively, at one point j on the fault (e.g., Graves and Wald, 2001;Shao and Ji, 2012). The value of ρ could vary from 0 (no correlation) to 1 (best correlation).…”

Resolution of Rise Time in Earthquake Slip Inversions: Effect of Station Spacing and Rupture Velocity

“…This parametric approach leads to nonlinear formulations of the inverse problem (Ji et al, 2002;Liu and Archuleta, 2004). Although care is sometimes taken to choose slip-time functions that are similar to those of dynamic rupture modeling, the assumption of the same slip-time function on all fault patches is highly simplified (Shao and Ji, 2012). Formulated in the wavelet domain (Ji et al, 2002) or time domain (Hartzell and Heaton, 1983), such methods rely on global optimization methods such as simulated annealing (Sen and Stoffa, 1991) or genetic algorithms (Sambridge and Drijkoningen, 1992).…”

Analysing earthquake slip models with the spatial prediction comparison test

“…However, there are several impeding issues regarding the reliability of the inverted models arising from the ill-posed nature of the inverse problem, limited and non-uniform data coverage, differences in selection and processing of the available data, incompletely known Earth structure, and variations in a priori assumptions on the faultgeometry (Beresnev 2003;Mai et al 2007;Shao & Ji 2012). Hence, earthquake source inversions come with considerable uncertainty, which however is only rarely investigated in as much detail as by Hartzell et al (1991Hartzell et al ( , 2007, Custodio et al (2005), Monelli & Mai (2008), Monelli et al 2009 andRazafindrakoto &Mai (2014).…”