Speaker
Description
Since its establishment in 1997, the CTBT IDC has been receiving data continuously from an increasing number of hydroacoustic stations. These stations are at present composed of 11 hydrophone triplets in the ocean and 5 T-stations (seismometers) on land monitoring for nuclear explosions in the oceans. This presentation summarizes advancements in the IDC automatic processing system from the early deployment of the system receiving data from a couple of hydroacoustic stations until today’s hydroacoustic network. The advancement in IDC processing of hydrophone data has been obtained in underwater signal travel time modelling, improved rule-based signal classification, estimate of underwater signal detection and direction of arrival, introduction of early artificial intelligence for signal classification, and global network processing based on an advanced Bayesian framework (NET-VISA). These advances have improved the automatic signal classification, detection and event localization and therefore importantly reduced the amount of human interaction in the data processing. Further improvements of the processing are envisioned that includes detailed oceanographic models and databases, efficient complex signal propagation models, the Progressive Multi-Channel Correlation (PMCC) algorithm and the rapid development of machine learning algorithms. These future elements are proposed in an updated IDC automatic data processing paradigm.
Promotional text
Advances in oceanographic and hydroacoustic physics-based models, data processing algorithms and high-performance computing facilities provide opportunities to reduce uncertainties in monitoring and verification of possible nuclear explosions in the oceans
