Methods for tracking fine scale underwater movements of marine mammals around marine tidal devices

4 Impact detection

It is unlikely that the technologies discussed above will be able to distinguish between an impact and a near miss. Yet this information is essential in assessing whether the observed movement behaviour resulted in successfully avoiding an impact. There is currently no proven way to directly detect the impact of a rotating blade against a marine mammal. However we consider here two approaches: mechanical sensing and video surveillance.

4.1 Mechanical sensing

Turbine blades are usually fitted with a variety of stress and acceleration sensors that are used to monitor the engineering performance of the device. Whilst is has been suggested that these sensors should detect a marine mammal strike, there has been no empirical test of this claim. If mathematical modelling suggests this is at all feasible, we suggest that this claim should be tested. Such a test could be based on a dead carcasses of similar appropriate mass ( e.g. a pig) being presented into a rotating turbine. Such tests would determine the false-positive and false-negative detection rates. If successful, automatic (near real time) algorithms could be developed and used to trigger detailed examination of data collected from other sources at the same time. If proven, such a mechanical based system would be easy to deploy on large numbers of operational turbine, at relatively small cost.

4.2 Video surveillance

A video recording system can be deployed on a seal, or operated remotely underwater in the vicinity of a turbine.

4.2.1 Animal-borne video surveillance

Animal-borne video surveillance (for example, Davis, Hagey & Horning 2004) are widely used to infer prey types. However the large amounts of video data collected (albeit the operating duration is only a few days) precludes data relay (Hooker et al. 2008). Thus the device must be physically recovered to download data, either by recapturing an instrumented seal or by recovering a remotely detached device.

The expense of the 'seal-ruggedized' camera systems and the low probability of an instrumented individual seal both interacting with a turbine (within the limited operating duration) and of successfully recovering the device preclude this approach as a practical option for UK grey and harbour seals. Furthermore it would be difficult to accurately reconstruct a track from such video data with few, if any, static geo-referenced features in the images.

4.2.2 Static video surveillance

Remote cameras have been used occasionally to monitor the local underwater activity of marine mammals (for example Simila & Ugarte 1993; Herzing 1996). However two issues limit their ability to observe marine mammal interactions at turbines. First, underwater visibility is usually too poor to allow sufficient range (and field of view) to monitor a complete turbine device. Visibility may be further reduced by the build-up of bio-fouling on the lens port. Second, video surveillance would require an artificial light source at night, a time when close interactions may be different. Artificial visible light at night would preclude the ability to investigate true night-time interactions. Infra-red ( IR) light, whilst invisible to marine mammals, is very rapidly absorbed in sea water. However the recent availability of high power IR light emitting diodes ( LEDs) and ultra-lowlight camera systems ¹⁴ may make warrant a practical re-investigation of the suitability of underwater video surveillance systems.

Notwithstanding these limitations, traditional video surveillance may have the potential for detecting impacts when the environmental conditions (daylight, good visibility and lack of bio fouling) allow. During these environmental windows it could be used to help interpret the fate of animals detected in the vicinity of the turbines using the other technologies. Detection events from these other technologies could be used to trigger the detailed examination of video sequences. Experience at EMEC suggests that this approach may have potential.