Automated Data Processing (ADP) Research and Development

Monitoring a comprehensive test ban treaty (CTBT) will require screening tens of thousands of seismic events each year. Reliable automated data analysis will be essential in keeping up with the continuous stream of events that a global monitoring network will detect. We are developing automated event location and identification algorithms by looking at the gaps and weaknesses in conventional ADP systems and by taking advantage of modem computational paradigms. Our research focus is on three areas: developing robust algorithms for signal feature extraction, integrating the analysis of critical measurements, and exploiting joint estimation techniques such as using data from acoustic, hydroacoustic, and seismic sensors. We identify several important problems for research and development; e.g., event location with approximate velocity models and event identification in the presence of outliers. We are employing both linear and nonlinear methods and advanced signal transform techniques to solve these event monitoring problems. Our goal is to increase event-interpretation throughput by employing the power and efficiency of modem computational techniques, and to improve the reliability of automated analysis by reducing the rates of false alarms and missed detections.