Project Abstract
|
Two approaches were considered for reducing the computational burden of UCERF3 in CyberShake. In the first, a crafted downsampling of the UCERF3 rupture set is used. These are all UCERF3 ruptures, but the subset more efficiently spans the model space, and thus might have been found by the Monte Carlo inversion. A set of about 33,000 ruptures fit the UCERF3 data constraints at least as well as the original full set. This set is larger than was used in UCERF2, but still small enough that it could be used in CyberShake. In a second approach, a point deaggregation method of UCERF3 was developed, and published with partial support from this award (Biasi and Anderson, 2016; SCEC Contribution 7098). The full rupture set is deaggregated by subsection, then summarized as subsection MFDs. With GMPE relations, the hazard is computed directly from the subsection MFDs, and summed. This method might be applied in CyberShake at any individual point, but problems arose that suggest that more work will be required before the deaggregation approach can be recommended for CyberShake.
|