2.5. Data Limitations
2.5.1. Snapshot Data
- Aerial survey data represent a snapshot of marine mammal distribution and densities within a given survey period and may not fully capture the natural variability of marine mammal distribution or densities over time. Changes in sightings rates may be influenced by environmental conditions; however, due to the short time frames (single day) of data collection, this has not been possible to analyse. Therefore, whilst differences in sightings rates between months may be due to seasonal changes, environmental conditions also have the potential to influence these results. However, for the Marine Mammal Technical Report the aerial survey data were interpreted in the context of historic survey data collected for other offshore wind farms in the region, and in the context of historic published information available for this region, therefore providing a robust baseline (refer to volume 3, appendix 10.2).
2.5.3. Bias
- Availability bias, i.e., an estimator of the probability that an animal is available for detection (i.e. visible) at any randomly chosen time, is used as multiplier to account for the period of time that each species may be available for detection. In the case of DAS, the time when an animal is available for detection is during the period that an animal is on the sea surface or just below the surface.
- Availability bias is likely to be influenced by extrinsic factors that combine to produce a situation that is unique to each survey: factors such as light conditions, water clarity (turbidity) and animal behaviour can influence whether an animal will be detected, particularly those beneath the water surface. In most cases, animals under the sea surface were noted and identified from digital images (refer to section 3.1.3). The depth at which reliable interpretation of images is assured will therefore rely considerably on the factors mentioned and for this reason availability bias may differ from month to month.
- Estimates of availability bias during aerial surveys are often based on studies looking at diving behaviour of a species because diving animals are inherently less likely to be visible. The results of these studies provide a correction factor for the proportion of time that animals are under the sea surface and therefore not available for detection. For the purpose of this assessment, correction factors were derived from studies in both the North Sea and other regions (e.g. harbour porpoise diving behaviour in the Baltic and North Seas (Teilmann et al., 2013); white-beaked dolphin diving behaviour in Iceland (Rasmussen et al., 2013) and grey seal diving behaviour in the North Sea (Thompson et al., 1991; Ørsted, 2018)) (refer to sections 3.5.1, 3.5.2 and 3.5.3, respectively). The caveat here is that species correction factors are unlikely to be a true representation of availability bias from one region to another, or from one month to the next, due to the potential spatial and temporal differences in environmental conditions. However, a precautionary approach was taken by reviewing the literature to compare correction factors from different studies and different months and then applying a conservative estimate (refer to species accounts in section 3.5).
- Perception bias, i.e. where an animal is available for detection, but the detection is missed, is less of a limiting factor during DAS compared to visual boat based surveys since the high definition video utilised during DAS captures all animals on the sea surface, or just under the sea surface, and the detection is not influenced by the ability of an observer to detect an animal. In addition, during data processing, a 20% subsample of the data were quality assured by HiDef to ensure that images were not overlooked, and therefore the potential for perception bias is negligible (refer to section 2.3).
- Similarly, a response bias, i.e. where an animal may respond to the presence of the surveying platform by either moving towards or away from the platform, is considered to be less of a limiting factor for aerial surveys compared to boat based surveys, due to the height of observation (in this case being approximately 550 m ASL). Therefore, the potential for response bias associated with the DAS is considered negligible.
2.5.4. Species Identification
- Animals were identified first to a species group (e.g. seals) and then to species level where possible (for example grey seal or harbour seal). For seals, the identification to species level is more difficult as it is not always possible to distinguish between species in cases where an individual is submerged. A subsample of data was subject to an external QA process by a third-party marine mammal expert to ensure agreement in identification.
- Where a full species identification could not be made, rather than discarding data it is sometimes possible for animals identified to higher taxonomic groups (e.g. seals) to be assigned to a species based on the proportion of the key species identified within the aerial survey area. However, it was considered that this approach may introduce unquantifiable bias (NatureScot, 2023), and as such individuals which were not identifiable to species level were not included in the design based analyses. Instead, only animals for which it was possible to assign species-level identification were included for analysis, while animals not identified to species level were included only in a high-level summary (refer to sections 3.1, 3.2 and 3.3).