Automated Classification of Ground Motion Record Quality Using Machine Learing
Motivation Densification of strong-motion station networks, their increased sensitivity, and the desire to use smaller magnitude data, is leading to exponentially-increasing ground motion datasets. Despite the improving reliability of seismic instrumentation, recorded ground motions are not of uniform quality, and the exponentially-increasing dataset sizes require automated quality assessment in order to be scalable. Here we propose a two-layer neural network that takes key ground motion metrics as inputs to automatically determine the quality of the records.
- Documents and Posters