1 Resolving the Source Parameters of the Parkfield Earthquake By Multiple Inversions of Different Data Sets

Inverting data to infer the kinematic source parameters of faulting has many inherent difficulties including incomplete knowledge of the structure of the medium, insufficient data in both quantity and quality and nonlinear relationships between the data and the parameters of the model. Consequently most models of faulting are poorly resolved. Because the Parkfield segment of the San Andreas Fault was the location of an on-going earthquake prediction experiment, the 40 km segment was closely monitored when a MW6.0 earthquake occurred on 28 September 2004. By subdividing the 43 near-source strong motion accelerometers into 12 overlapping subsets of 20 stations or more, we invert for the kinematic parameters of the faulting. This procedure allows us to determine the mean and variance of the kinematic parameters as if the same earthquake were recorded by 12 different distributions of stations. The mean value of each parameter is compared to the parameter found by inverting all 43 stations. Because the relationship between some of the parameters and the data is nonlinear, we also invert all the data using different initial conditions. This leads to a distribution of parameters with a mean and a standard deviation that provide insight into how well the parameters are resolved on the fault