One of the main challenges in nuclear test ban monitoring is to detect and locate small seismic events, in situations where no local seismic network is available. This means that in a nuclear test ban monitoring context, the small events of interest may be located far from the nearest monitoring station, and such events may well be detected only by one sensitive array, at a considerable distance from the source, maybe several hundred kilometers or more. This is the main reason why we are focusing on optimizing one-array detection and location capabilities.
We have recently made significant progress in this regard, by developing systems for automated one-array detection and location of seismic events from selected mining areas. For example, an experimental on-line detection and location system, using the ARCES array, has been implemented for some of the mines in the Kola Peninsula, and the automatic process has been compared to the regular analyst reported bulletin. It turns out that the automated process, with appropriate calibration, can match or exceed the performance of the analyst in terms of location precision, even when the analyst has access to network data. The main reasons for this performance are the application of optimized, fixed frequency band filters together with careful application of automatic autoregressive onset estimation techniques.
The main results are illustrated in the panel below, showing three maps in the same scale.
Single-array processing.
The top map shows the location of several mines in northern Sweden and the Kola Peninsula in relation to the ARCES array. The middle map shows the results from standard processing of a number of mining explosions, where the explosions from different mines are color-coded. The bottom map shows results after applying the optimized method described above. There is an order-of-magnitude improvement in location accuracy.
|
 |
|
|
|
