UWSpace is currently experiencing technical difficulties resulting from its recent migration to a new version of its software. These technical issues are not affecting the submission and browse features of the site. UWaterloo community members may continue submitting items to UWSpace. We apologize for the inconvenience, and are actively working to resolve these technical issues.
 

Feature identification in time series data sets

Thumbnail Image

Files

1-s2.0-S2405844018365782-main.pdf (946.19 KB)

Date

2019-05

Authors

Shaw, Justin
Stastna, Marek
Coutino, Aaron
Walter, Ryan K.
Reinhardt, Eduard

Journal Title

Journal ISSN

Volume Title

Publisher

Elsevier

Abstract

We present a computationally inexpensive, flexible feature identification method which uses a comparison of time series to identify a rank-ordered set of features in geophysically-sourced data sets. Many physical phenomena perturb multiple physical variables nearly simultaneously, and so features are identified as time periods in which there are local maxima of absolute deviation in all time series. Unlike other available methods, this method allows the analyst to tune the method using their knowledge of the physical context. The method is applied to a data set from a moored array of instruments deployed in the coastal environment of Monterey Bay, California, and a data set from sensors placed within the submerged Yax Chen Cave System in Tulum, Quintana Roo, Mexico. These example data sets demonstrate that the method allows for the automated identification of features which are worthy of further study.

Description

Keywords

time series analysis, event detection, feature identification, geophysics, oceanography, atmospheric science, environmental science, geology, hydrology

LC Keywords

Citation

URI

https://doi.org/10.1016/j.heliyon.2019.e01708
 http://hdl.handle.net/10012/15621

Collections

Waterloo Research
Applied Mathematics