SCEC Award Number 16115 View PDF
Proposal Category Individual Proposal (Integration and Theory)
Proposal Title Next Generation SDSU BBP Module Validation
Investigator(s)
Name Organization
Kim Olsen San Diego State University
Other Participants Ms. Rumi Takedatsu, Research Assistant at SDSU.
SCEC Priorities 6e, 6b, 6c SCEC Groups GMP, CME, GMSV
Report Due Date 03/15/2017 Date Report Submitted 06/05/2017
Project Abstract
We have validated the SDSU BBP module in terms of ground motion duration, intra-event variability, and PSA against NGA-West2 GMPEs. We obtained significantly improved 50-realization PSA GOF to NGA-West2 (Part A+B), by adjusting the shape and amplitude of the source-time function convolved with the high-frequency scattering functions in the SDSU BBP module. We find that the SDSU module produces median intra-event standard deviations in very good agreement with GMPEs, primarily due to the inherent variability built into the HF scattering functions. We validated the (HF) durations associated with the SDSU module synthetics, after refining the scattering parameters in the code. The current technique of estimating the goodness-of-fit (GOF) for durations (Anderson set of GOF metrics recently implemented on the BBP) aligns synthetic and recorded time series to the time of the 5% energy and cuts the synthetic time series when the data record ends. This technique can provide a biased GOF value when the data is truncated (i.e., due to too short of a recording window). We find that the synthetics should not be truncated to provide a fair comparison. We have eliminated data records that appear to be truncated and several bugs in the BBP GOF procedure associated with time interpolation and alignment fixed. 50-realization significant duration estimates from the updated SDSU module were generally improved of the WUS and Japanese events. We implemented multi-segment rupture and demonstrated for a 3-segment Landers approximation, with improved durations and little change in the PSA GOF compared to the 1-segment approximation.
Intellectual Merit SCEC aims to advance the modules implemented on the BBP beyond the original metric of PSA. In order to ensure that the BBP generates synthetics that agree with data it is important to validate other metrics, such as variability (e.g., intra-event standard deviation) and duration measures. Furthermore, the validation should be carried out with the most recent GMPEs (NGA-West2) as reference to be up to date.
Broader Impacts Immediate applications of the improved SDSU modele includes the SDSU Broadband Platform (BBP) method (currently sharing the Graves and Pitarka rupture generator). The module is also available for broadband computation on the CyberShake Platform. The project includes training of Rumi Takedatsu (female, researcher).
Exemplary Figure Comparison of intraevent standard deviations for 50 realizations of the Loma Prieta event (gray lines) with the mean shown in black, compared to 3 leading NGA-West GMPEs, at periods of 0.15s and 0.3s. The mean of the intra-event standard deviations from the 50 realizations compare very well with the (practically distance-independent) values from the GMPEs. Credit: R. Takedatsu.