Background Noise Impacts Harbor Porpoise Detections in Passive Acoustic Monitoring
Passive acoustic monitoring (PAM) is instrumental to monitor marine mammals, with many applications, including impact of noise. Noise, however, also affects the ability of PAM to detect sounds of interest. The performance of a PAM algorithm was tested by feeding it recordings of ship noise merged wi...
| Main Authors: | , , |
|---|---|
| Other Authors: | , , , |
| Format: | Article in Journal/Newspaper |
| Language: | English |
| Published: |
Springer
2023
|
| Subjects: | |
| Online Access: | https://portal.findresearcher.sdu.dk/da/publications/148e5ec4-8116-4b40-a110-5daf192fc33b https://doi.org/10.1007/978-3-031-10417-6_33-1 |
| Summary: | Passive acoustic monitoring (PAM) is instrumental to monitor marine mammals, with many applications, including impact of noise. Noise, however, also affects the ability of PAM to detect sounds of interest. The performance of a PAM algorithm was tested by feeding it recordings of ship noise merged with harbor porpoise (Phocoena phocoena) clicks. Three groups of porpoise clicks (“click trains”) were inserted every 10 sec in files containing ship noise, one at a time (n = 1180 min). The recall (probability of detecting porpoises in 10-sec bins) was modeled using a generalized linear model with a binomial distribution and one explanatory variable: noise in one of four frequency bands (<20 kHz, 20–80 kHz, 80–150 kHz, and the decidecade centered on 16 kHz). The best model had the 20–80 kHz band as the explanatory variable, but noise in all bands masked detections when exceeding 90 dB re. 1μPa, becoming a cofounding factor. Therefore, decreased detections at these noise levels cannot be attributed unequivocally to effects of noise on the behavior of porpoises but may be due to masking of the PAM instrument. However, PAM can be used, as long as the noise levels above 20 kHz are monitored and accounted for. |
|---|