Premium
Information theory measures for the engineering validation of ground‐motion simulations
Author(s) -
Tsioulou Alexandra,
Galasso Carmine
Publication year - 2018
Publication title -
earthquake engineering and structural dynamics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.218
H-Index - 127
eISSN - 1096-9845
pISSN - 0098-8847
DOI - 10.1002/eqe.3015
Subject(s) - probabilistic logic , computer science , measure (data warehouse) , earthquake engineering , ranking (information retrieval) , ground motion , uncertainty quantification , divergence (linguistics) , data mining , engineering , machine learning , artificial intelligence , structural engineering , linguistics , philosophy
Summary This short communication introduces a quantitative approach for the engineering validation of ground‐motion simulations based on information theory concepts and statistical hypothesis testing. Specifically, we use the Kullback‐Leibler divergence to measure the similarity of the probability distributions of recorded and simulated ground‐motion intensity measures (IMs). We demonstrate the application of the proposed validation approach to ground‐motion simulations computed by using a variety of methods, including Graves and Pitarka hybrid broadband, the deterministic composite source model, and a stochastic white noise finite‐fault model. Ground‐motion IMs, acting as proxies for the (nonlinear) seismic response of more complex engineered systems, are considered herein to validate the considered ground‐motion simulation methods. The list of considered IMs includes both spectral‐shape and duration‐related proxies, shown to be the optimal IMs in several probabilistic seismic demand models of different structural types, within the framework of performance‐based earthquake engineering. The proposed validation exercise (1) can highlight the similarities and differences between simulated and recorded ground motions for a given simulation method and/or (2) allow the ranking of the performance of alternative simulation methods. The similarities between records and simulations should provide confidence in using the simulation method for engineering applications, while the discrepancies should help in improving the tested method for the generation of synthetic records.