Dr Douglas Gillespie:
Dr Douglas Gillespie
Scottish Oceans Institute
University of St Andrews
tel: 01334 462663
Sea Mammal Research Unit
School of Biology
Scottish Oceans Institute
edit dg50 details
Passive acoustic monitoring (PAM) is an effective way of detecting many species of cetacean and has an important role in abundance surveys and in detecting cetaceans in the vicinity of certain human activities which may cause harm, such as seismic surveys, military sonar exercises and even shipping.
Passive acoustic detection of beaked whales
Beaked whales are one of the least known marine mammal species due to their offshore habitat and deep diving behaviour. We are investigating the use of passive acoustics as a means of detecting beaked whales using towed hydrophones close to the surface. Several beaked whale species are known to produce narrow band high frequency clicks during deep foraging dives. Many beaked whale species have not yet been recorded.
The four main focuses of our research are
- To assess how efficiently beaked whale can be detected.
- To develop software which can automatically detect beaked whale clicks and tell them apart from other species.
- To try to record the sounds of previously unrecorded beaked whale species.
- To test the effectiveness of passive acoustic monitoring in developing habitat use models for beaked whales.
PAMGUARD is open source software for the detection and localisation of marine mammal vocalisations. It is optimised for real time use in the field and has applications both in abundance survey and in mitigation monitoring. I manage the PAMGUARD project and wrote both the core structure of the PAMGUARD and many of the detection, localisation and mapping modules within the software.
Back in the 1980’s available computers were not powerful enough to process acoustic data in real time. However, with the increased power available since the mid 90’s, it is now possible to develop software that will detect and classify sounds in real time on affordable PC’s.
As computers become ever more powerful, we have been able to develop more sophisticated detectors for more and more species, increasing the range of frequencies we can work at and the number of channels of data that can be processed. Now that we no longer hunger for more processing power, the trend in affordable computing has been for smaller and lower power devices. Indeed, most of us carry a mobile phone containing a processor that is more than capable of carrying out serious amounts of real-time data processing. Much of my current research therefore involves the development of detection systems that can run on low power devices mounted on moored buoys and autonomous vehicles such as submarine gliders. As well as the challenge of making useful detections on a limited power budget, we are also addressing the problem of how to interpret this type of data: for instance, if I hear 10,000 echolocation clicks from my glider, how many animals are there ?