This report describes the method, output and learning from a trial of a high-definition video survey technique for census of seabirds and windfarm sites. The technique utilises a gyroscopically stabilised camera mount to eliminate aircraft vibration enabling the use of high magnification lenses. This in turn enables the helicopter to fly at a significant distance from the birds, while still maintaining sufficient image quality to identify them and assess certain aspects of their behaviour. The video is coupled with a GPS logging system that associates a geographical location with each video frame, enabling subsequent statistical analysis of the spatial distribution of birds. The general findings of the trial are as follows:
The system provides sufficient detail to enable reliable identification of birds in most cases.
Altitudes in excess of 1km are feasible; cloud-base would typically be the limiting factor.
Although this survey was not carried out over an existing windfarm, the altitude and image quality demonstrate that it directly applicable to windfarm survey.
Despite higher wind speeds than would be tolerated on a conventional survey, it was possible to conclude that all birds had been observed on most transects with a high degree of confidence.
To ensure recognition of smaller species, image width must be fixed at around 30m, which is substantially smaller than the zone in which it is assumed conventional surveys have a 100% detection rate.
Transects must therefore be significantly closer than is the case during conventional surveys.
Some difficulties were encountered in operating the video equipment in the manner required by the trial and an approach that minimises these effects has been identified.
Technical refinements to improve quality and reduce cost have been identified, which it may be appropriate to test in a second survey.
The use of video provides a permanent record of the survey, which confers the advantage of ‘auditability’ of survey data. It also enables subsequent re-use data for other purposes, such as monitoring of non-windfarm human impacts or marine mammal activity; indeed the video provides a complete record of all visible aspects of the condition of an area of water on a given day.
While the mode of operation of the equipment for survey purposes poses significant challenges not normally encountered in aerial filming, the potential of the technique is largely as anticipated. The trial successfully identified a range of birds from a variety of families and species, including auks, gulls, gannets, and shags. Despite the relatively small scope of the survey, these birds demonstrate the ability of the technique to cope with a range of bird sizes, plumages and behaviours. The GPS data enabled an approximate location to be allocated to each bird, revealing anticipated patterns in their spatial distribution.
The coverage rate of the technique during the survey was found to be approximately 20 sq. km per hour. A significant proportion of survey time was spent on manoeuvring due to the small size of the survey area, and it is anticipated that a very significant increase in coverage rate would be found for larger survey areas.
Several technical difficulties encountered during the survey reduced image quality when compared with the theoretical optimum. Nevertheless, bird identification was straightforward in most cases, even when the image quality was poor. This was enabled by the ability of an experienced observer to recognise a bird by size and wing motion. It was also found that consistency of appearance across multiple frames was the strongest visual clue for detecting sitting birds. It is therefore concluded that video has significant advantages over still images as a survey medium in this application.
Appropriate measures to eliminate or mitigate each of the technical difficulties encountered have been identified, and it is anticipated that future trials will show a significant increase in both image quality and statistical robustness of data analysis. In particular, logging camera orientation and providing visual aids to facilitate camera control will be a great advantage in future.
At this stage, while qualitative comparisons have been made, it is not possible to make a quantitative comparison in terms of accuracy and cost between the high definition video method and either the plane or ship based survey techniques. This cost/benefit analysis will be required in future to ensure adoption of the most appropriate survey techniques for census of seabirds at windfarm sites.
It is therefore recommended that a further trial be undertaken, to run alongside a conventional survey of a windfarm site. This would both enable a direct comparison of techniques to be made, and provide the opportunity to test technical refinements identified as a result of this survey.