| Autor: |
Zakarauskas, Pierre, Parfitt, Carlton J., Thorleifson, Jon M. |
| Zdroj: |
Journal of the Acoustical Society of America; 1991, Vol. 90 Issue 1, p470-474, 5p |
| Abstrakt: |
A new nonparametric transient detecting and extracting algorithm is described that was used to extract transients automatically from ambient noise time series. The advantage of this peak-thresholding algorithm over the more usual power-thresholding algorithm is that the beginning and end of transients can be found precisely. This new algorithm was applied to three 5-min-long ambient noise samples taken under the Arctic pack ice during the month of April. The samples cover a wide range of ambient noise conditions. One was associated with thermal ice cracking, another one to distant lead formation events, while the third had extremely low ambient noise levels. The data were bandpass filtered in order to study the transient content at different frequencies and improve detection performances. It was found that the relative duration and power of transients within the time series varied greatly with frequency. During noise periods, transients accounted for only 5%-25% of the power around 10 Hz, and 45%-80% of the power between 100 and 1000 Hz. On the other hand, during quiet periods, transients accounted for 35%-45% of the power around 10 Hz, and 80%-95% of the power between 100 and 1000 Hz. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
