Abstract
This report presents the results from analysis of subsea video footage gathered between October 2015 and March 2020 as part of Nova Innovation’s programme of environmental monitoring for the Shetland Tidal Array, Bluemull Sound. The use of turbine-mounted subsea cameras to monitor nearfield interactions with mobile species (fish, birds and marine mammals) commenced in October 2015, when the first turbine in the array was deployed. A second turbine was added in August 2016 followed by a third in August 2017. Subsea cameras have been used continuously throughout this period to monitor the nearfield environment around each of the turbines in the array and is ongoing. Each of the three turbines deployed in the Shetland Tidal Array up to March 2020 has three integrated cameras, as follows:
- One on top of the nacelle directed towards the blades.
- One on the side of the nacelle directed towards the blades.
- One on the underside of the nacelle directed towards the seabed.
Video recording is continuous, with footage retention triggered by a motion detector system. Retained footage includes a few seconds captured prior to the trigger event to facilitate identification of the cause. In addition to the recorded video, the camera motion detector system records a still image from the video, which provide a more rapid method of detecting potential “events” of interest. The total length of recordings varies from a few seconds up to 15 minutes. All triggered footage is retained and stored. To the end of March 2020, the total dataset of retained and stored video footage comprised nearly 1 million videos, representing almost 20,000 hours of footage. The storage footprint totalled more than 3 TB.
This report presents the results from examination and analysis of four subsets sampled from the full 20,000 hours dataset of retained footage spanning October 2015 to March 2020. Collectively, the subsets comprise 4,049 hours of video, representing over 20% of all retained video footage.
Subset 1 comprised footage captured only during daylight hours and when turbines were operating. Footage in this subset was also selected from within three months of camera maintenance, so that biofouling on cameras was minimal, ensuring images of high quality. This subset comprised 1,115 videos totalling 28 hours of footage.
Subset 2 comprised footage corresponding to the times of the highest surface counts of black guillemot (Cepphus grylle) and European shag (Phalacrocorax aristotelis) in the array area, derived from Nova’s Bluemull Sound vantage point survey data1 . This subset comprised 451 videos totalling 18 hours of footage.
Subset 3 comprised footage corresponding to times when sightings of marine mammals were recorded in the array area, derived from Nova’s Bluemull Sound vantage point survey data. This subset comprised 336 videos totalling 3 hours of footage.
Subset 4 comprised all retained footage covering the entire period March 2016 to January 2017, with some additional footage gathered from a short deployment of Nova’s first installed turbine in October 2015. This subset comprised 92,000 videos totalling 4000 hours of footage and 72,000 stills images derived from the motion detection capture system. This subset had previously been examined and summarised in report submitted to MS-LOT in 2017. Revisiting footage in this way, as the subsea video dataset grows, is a valuable way of optimising the value of the data to improve understanding for nearfield interactions between turbines and mobile species.
In the absence of automated data processing based on machine learning or algorithms to filter or categorise the subsea video, the four subsets were analysed manually. For each video clip, the source of the trigger was identified along with any occurrences of mobile marine species in the rest of the footage. The total number of occurrences of species of marine mammal, diving bird or fish in reviewed footage were quantified and species identified. Occupancy patterns and behaviour in relation to the operational state of the turbines was examined through the sampling strategy. All sampled footage was scrutinised for any evidence of nearfield encounters or contact between individual animals and turbine blades (collisions). The very low numbers of animals observed in the subsea video data meant that there were insufficient data to perform detailed analyses.
There were no instances of physical contact between marine species and the turbine blades in any of the four subsets of video footage and stills examined. All mobile species were identified with high confidence to species level. Saithe (Pollachius virens) was the most frequently observed and abundant species, with groups observed aggregating around the turbines, moving vertically up and down in the water column according to tidal flow. The only other mobile species in the sampled four subsets of footage were very infrequent observations of individual European shag (Phalacrocorax aristotelis), black guillemot (Cepphus grylle) and harbour seal (Phoca vitulina), with fewer than thirty occurrences of these species in total across the four subsets. These species were only observed at times when the turbines were not operating (i.e., blades were stationary). Jellyfish, such as Lion’s mane, Cyanea capillata, were occasionally observed in video either actively swimming at slack tide or being carried passively past the turbines during the ebb and flood.
On the basis of fine-scale occupancy patterns observed in the mobile species in the footage examined comprising the four sampled subsets, a very low risk of encounters with operational turbines in the Shetland Tidal Array is indicated. Diving birds and marine mammals were only observed in the nearfield environment of turbines in subsea footage at times when they were not operating (i.e., blades were stationary). No diving birds or mammals were observed in any of the footage when turbines were operating. Similarly, fish observed in footage were seen to generally drop to the seabed as current speed increases. When fish were observed at times when turbines were operating (rotating), there was evidence of some nearfield evasion in individuals.
The monitoring and the results presented in this report provide insights into interactions between mobile species and Nova’s turbines in the Shetland Tidal Array. This includes during periods when the turbines are rotating and generating, as well as periods of lower current speed when blades are stationary. The implications of the analysis presented in this report for refining understanding about collision risk for the Shetland Tidal Array are discussed. The contribution that the results make to the wider advancement of the knowledge base on the environmental effects of tidal energy and best practice in environmental assessment is also considered.