Trimming a hazard logic tree with a new model-order-reduction technique
The size of the logic tree within the Uniform California Earthquake Rupture Forecast Version 3, Time-Dependent (UCERF3-TD) model can challenge risk analyses of large portfolios. An insurer or catastrophe risk modeler concerned with losses to a California portfolio might have to evaluate a portfolio 57,600 times to estimate risk in light of the hazard possibility space. Which branches of the logic tree matter most, and which can one ignore? We employed two model-order-reduction techniques to simplify the model. We sought a subset of parameters that must vary, and the specific fixed values for the remaining parameters, to produce approximately the same loss distribution as the original model. The techniques are (1) a tornado-diagram approach we employed previously for UCERF2, and (2) an apparently novel probabilistic sensitivity approach that seems better suited to functions of nominal random variables. The new approach produces a reduced-order model with only 60 of the original 57,600 leaves. One can use the results to reduce computational effort in loss analyses by orders of magnitude.
Citation Information
Publication Year | 2017 |
---|---|
Title | Trimming a hazard logic tree with a new model-order-reduction technique |
DOI | 10.1193/092616EQS158M |
Authors | Keith Porter, Edward H. Field, Kevin R. Milner |
Publication Type | Article |
Publication Subtype | Journal Article |
Series Title | Earthquake Spectra |
Index ID | 70189248 |
Record Source | USGS Publications Warehouse |
USGS Organization | Geologic Hazards Science Center |