NEAR-FIELD SEISMIC LOCALIZATION USING DELTA T MAPPING

Abstract

Near-field seismic localization has important significance and wide applications in the real world, for instance, locating explosions or tracking traffic movements. The traditional methods designed for far-field scenarios are limited here due to the unknown velocity structures and high accuracy demands. This paper, for the first time, applied the Delta T Mapping (DTM) technique from acoustic emission detection in near-field seismic localization. DTM first needs to construct a mapping of the difference of first arrivals on which the following locating is based. There are two approaches for establishing such model: (1) grid search method: using linear scattered point interpolation to obtain new DTM of higher resolution; (2) statistical locating method: using Gaussian Process Regression to build the mapping from Delta T to positions. The experiment was conducted in an area of 140 m×90 m in the suburb of Beijing. The locating error was 0.5∼5.1 m. The results showed that DTM is reliable, highly accurate and suitable for real-time use for near-field seismic localization. The cost of learning DTM and analyzing data could be further decreased while obtaining highly accurate DTM by switching the sources and receivers. Furthermore, combining source-scan algorithm has certain potential to locate multi-sources.