Abstract
Following from an initial investigation into the use of High Definition HD video for aerial waterbird survey undertaken in August 2007, this paper presents the findings of a full scale trial of the technique. The trial was carried out at the Shell Flats area off Blackpool, and aimed to satisfy five objectives. The outcome of this trial is assessed against its objectives as follows:
Demonstrate the ability of the technique to reliably survey the more sensitive species of waterbird, such as common scoter or red throated diver.
A large number of common scoter were encountered. When observed from 600m, the birds showed no signs of awareness of the aircraft. Indeed, Blackpool air traffic control notified the survey aircraft of several other aircraft in the vicinity at a similar altitude, and even required that the survey be halted at one point to make way for a large jet. If aircraft at this altitude disturb birds then a revision of survey methodology at areas near large airports would be required. When flying at 210m, the birds were clearly aware of the aircraft, but the vast majority remained on the water at the point of encounter. Only a very small number of birds were observed in the air or taking flight. It appears that if the survey altitude is kept at 270m or more, Scoter will not take flight as a result of the survey aircraft. It is therefore recommended that 270m be adopted in future as the minimum survey altitude.
Compare the output of the technique with data from other surveys to provide insight into the performance of the technique.
Some comparison has been made with historic data from the 2004/5 surveys undertaken for BERR (formerly the DTI) and with four datasets generated by WWT between 2004 and 2007. These show very good agreement in terms of the spatial distribution of birds, with the same key feeding locations and seasonal behaviours identified across all data sets. Comparison of overall population reveals that our estimated population (7000 birds) is somewhat less than those from the WWT surveys (15000-16000). However, this is not seen as a reason for concern, partly as there are good reasons to expect the estimate to be lower in our case, and also because the way in which we have analysed the WWT data is likely to be generating an over-estimate of the population. It is proposed that consistency of the data, the quality of the video and the relative statistical simplicity of our method, coupled with its intrinsic advantages for surveying built windfarms justify its use along side existing techniques as a mainstream tool for environmental data gathering and analysis.
To undertake a trial on a statistically significant scale, to demonstrate the consistency of the technique and the usefulness of the data.
An area of 226 sq. km has been surveyed at a coverage density of 12%. A total of 823 birds were observed, 785 of which were ‘on transect’ with a further 38 observed between transects. This provides a suitably large observed population for statistical analysis. The format of the data permits straightforward population and distribution analysis by well-established statistical methods. A particular benefit is the ability to perform local rather than global statistical normalisation, which could potentially increase the power of local population variation analysis. While this is particularly relevant to species which form localised clusters around static features such as cockle beds, in principle the scale of analysis can be varied to suit the distribution and behaviour of any species. In particular, it will be relevant in the future to tune the scale of analysis to the scale of a given windfarm in order to assess the likelihood that a significant alteration in bird population and/or distribution has occurred. Investigation of the potential performance of such a technique will be explored further using methods comparable to those described in Ref. 8.
Demonstrate the feasibility of an aeroplane rather than a helicopter as the survey vehicle.
As expected, the aeroplane was well suited to the task, providing a more stable platform, a quieter more comfortable environment for the operators, using less fuel, and costing significantly less per hour (excluding international ferry flights). Unlike visual observer methods, a large number of twin engine aircraft are suitable for this form of surveying, and can be put into service with only minor interior modifications.
To test the technical refinements proposed as a result of the first trial.
The recommended techniques for control of focus, image size and contrast were implemented and found to result in a significant improvement in video quality overall. In particular setting of contrast for bright whites rather than midtones was found to be highly beneficial. Control of focus still presents some difficulties over flat seas, and focus would preferably be capable of being fixed in a dedicated survey aircraft. Similarly, zoom and camera attitude could be locked without an adverse impact on the survey quality, which would considerably reduce the burden on the operator. Indeed, this would probably be a necessity for a multi-camera system controlled by a single operator.
The target image width of 30m metres identified during the first trial was found to be pessimistic; the increased image quality resulting from the other technical refinements was sufficient to enable an increase of image width to 40m while still maintaining an overall improvement in quality. This adjustment confers a 30% increase in coverage rate.