1. Introduction

  1. Berwick Bank Wind Farm Limited (BBWFL) is a wholly owned subsidiary of SSE Renewables Limited and will hereafter be referred to as ‘the Applicant’. The Applicant is developing the Berwick Bank Wind Farm (hereafter referred to as ‘the Project’) located in the outer Forth and Tay region ( Figure 1.1   Open ▸ ).
  2. The Project is located adjacent to the consented Forth and Tay offshore wind farms (OWFs) consisting of Seagreen to the north, Inch Cape to the northwest and Neart na Gaoithe to the west ( Figure 1.1   Open ▸ ).
  3. The Project will, if consented, provide an estimated 4.1 GW of renewable energy. Given the anticipated operational life span of 35 years, the Project will make a critical contribution to Scotland’s renewable energy target of 11 GW of new offshore wind by 2030.
  4. Turbine capacity is predicted to be 14 MW to 24 MW per wind turbine generator (WTG), with the number of turbines on site to be 179 to 307. Importantly, the minimum lower blade tip height has been increased from 22 m to 37 m (LAT) as an engineering design measure to increase the air gap and reduce potential collision risk to seabirds. The effectiveness of this is demonstrated in Annex A. 

Figure 1.1:
Boundaries for all consented and proposed offshore wind farms currently within the Outer Firth of Forth.

Figure 1.1: Boundaries for all consented and proposed offshore wind farms currently within the Outer Firth of Forth.

 

2. Purpose of the report

  1. This Technical Report provides estimates of collision risk for eight seabird species (Section 3.3) in relation to the Project.
  2. Two approaches to collision risk modelling (CRM) were used:
  • Deterministic offshore Band CRM (Band, 2012); and
  • Stochastic CRM (sCRM) (Masden, 2015; McGregor et al., 2018).
  1. The deterministic Band model was used following the advice issued in the Scoping Opinion (4 February 2022) and provides the primary estimates for assessment of collision risk within the Proposed Development Array area, with these outputs used in the Population Viability Analyses (Technical Appendix 11:6: Ornithology Population Viability Analysis Technical Report). The sCRM approach, which takes account of the variability around input parameters, is used only for comparative purposes because the avoidance rates for use with this model (Bowgen and Cook, 2018), are not currently endorsed by NatureScot or Marine Scotland Science. This was agreed via the Ornithology Road Map process and follows the Scoping Opinion. 

3. Methods

3.1. Overview of methods

  1. The method for estimating the number of collisions can be simply expressed as:

where

FoT = Flux rate multiplied by the operational time of the wind farm,

Q2r = Proportion of flying birds at collision risk height

Probability of collision = the probability of a single bird colliding with a turbine assuming no avoidance behaviour

  1. Band (2012) and the sCRM make use of two model frameworks; a basic model which uses a straightforward calculation of Q2r from either site-based flight height estimates or generic flight height distributions, and an extended model which uses a flight height distribution but assumes that risk varies over the area of the turbine blades. As guidance, Band (2012) suggests that collision risk estimates get presented using three Options:
  • Option 1 – Basic model: Proportion of birds at collision height (calculated manually) based on site-based flight height data, which assumes a uniform distribution of risk over the extent of the rotor swept area.
  • Option 2 – Basic model: Proportion of birds at collision risk height (calculated automatically), based on a generic flight height distribution, also assuming a uniform distribution of risk over the rotor swept area.
  • Option 3 – Extended model: Proportion of birds at collision height calculated by integrating risk across a turbine blade at different points along a generic flight height distribution.
    1. Estimates using the Band model were generated using the Excel spreadsheet tool accompanying the Band (2012) publication. Estimates using the sCRM were generated using the underlying R code from the web-based sCRM tool[1] so that multiple scenarios could be run more efficiently. Code can be made available on request.
    2. The three model Options are equivalent between the Band (2012) spreadsheet implementation and the sCRM tool.
    3. The estimates from using the Band and sCRM approaches are broadly similar, prior to application of the respective avoidance rates (see Table 3.6   Open ▸ and Table 3.7   Open ▸ ). However, the sCRM integrates stochasticity by way of random sampling from statistical distributions of input parameters.

3.2. Overview of modelled scenarios

  1. The expected installation date for the Project means that it is anticipated that new technological advancements in turbine technology may occur between the consent application and the installation of turbines.
  2. For this reason, physical parameters for future possible turbines have been determined by the Applicant’s engineering team through an assessment of existing technology and research into turbine Developer plans and expectations with respect to future models. This exercise identified five potential turbine ratings to be considered in the Project Design Envelope (PDE) and determined likely worst-case parameters for each. The engineering team identified two potential variations of each of the three larger future turbine ratings with different rotor design concepts, resulting in eight types in total (Section 3.4.1).
  3. The scenarios for the Band (2012) model included the eight turbine types and Options 2 (basic model) and 3 (extended model) only, with Option 3 limited to the large gull species following SNCBs guidance (SNCBs, 2014).
  4. Model Options 2 and 3 make use of the generic flight height data of Johnston et al. (2014a; 2014b) as advised in the Scoping Opinion (4 February 2022). In addition, collision estimates for kittiwakes were modelled using Option 1 of the basic Band model, using site-based specific flight heights gathered during boat-based surveys, as advised in the Scoping Opinion and presented in Annex B (Technical Appendix 11.7: Comparison of boat-based and digital video aerial survey methods for seabirds).
  5. It was agreed through the Ornithological Road Map process (RM2, 9 August 2021) that scenarios for all turbine types would be modelled to determine which represents the worst-case for each species considered in terms of predicted collision mortality.
  6. The turbine and biological parameters of the worst-case scenario from the Band (2012) model for each species, were also modelled using the sCRM and provided for context (Annex C).
  7. The Applicant has for the most part adopted the advice on ornithological assessment parameters advised in the Scoping Opinion (Volume 3, Appendix 11.8), for the purposes of conducting a CRM assessment on offshore ornithology for the EIA. Nevertheless, the Applicant considers elements of the Scoping Opinion to be over-precautionary and a departure from standard advice/practice. As such, the Applicant determined to undertake a ‘dual assessment’ approach of the collision risk posed by the proposed Development:
  • The ‘Scoping Approach’; and
  • The ‘Developer Approach’.
    1. With respect to estimating collision risk, the two approaches differ only in their use of input monthly density estimates of flying birds of the assessed species within the proposed Development.
    2. The Scoping Approach is based on the Scoping Consultation responses from NatureScot and Marine Scotland Science (Volume 3, Appendix 11.8) which advised the use of monthly maximum density of relevant seabird species within the proposed Development Array area in the CRMs.
    3. However, guidance on the use of the CRM suggests that model predictions should be based upon the mean monthly densities of flying birds estimated within the array area (Band 2012)[2] and, to the best of the Applicant’s knowledge, this approach has been applied in all recent UK offshore wind farm assessments (i.e. from at least the Round 3 and Scottish territorial waters leasing rounds onwards).
    4. Despite this, the Scoping Opinion advised that the CRMs for the Proposed Development should use the maximum monthly densities of flying birds within the array area. In part at least, this advice appeared to derive from a decision that it was not possible to use the stochastic version of the CRM (sCRM; McGregor et al., 2018) due to an absence of recommended avoidance rates, meaning that the resultant collision estimates for the proposed Development (as generated from the deterministic CRM) would not account for variation and uncertainty in input information, including baseline densities (K. Bell, email 02/03/2022; Volume 3, Appendix 11.8). Further advice in the Scoping Opinion was that sCRM outputs should just be presented for context. However, the use of the maximum monthly densities does not actually address this issue since a full measure of uncertainty would be limited to differences between the two density calculations (i.e. mean density and maximum density). Furthermore, it is also the case that guidance from Natural England accepts that option 2 of the sCRM can be used with the same species-specific avoidance rates as for option 2 of the deterministic CRM (Parker et al., 2022).
    5. The Developer Approach follows that recommended in the industry guidance (Band, 2012) and as undertaken in all recent UK offshore wind farm assessments that the Applicant is aware of. This approach uses the monthly mean of the relevant two annual estimates of the density of flying birds within the proposed Development Array area. The Applicant is unaware of any change to the evidence base to support a change from this approach, noting that in their advice for the revised designs of the Forth and Tay projects Marine Scotland Science stated that an approach of using the maximum monthly density values within the CRM “runs the very high risk of producing an estimated effect that is highly likely to be unreasonable and unrealistically high.” (Marine Scotland, 2017a, Marine Scotland, 2017b).
    6. The Applicant considers the Developer Approach to be scientifically robust, suitably precautionary and reflective of current methods of assessment and recommends that it can and should be reasonably relied upon by the decision maker for the purposes of assessment. The Applicant has therefore provided the necessary information to support a decision based on the Developer Approach.
    7. Nevertheless, cognisant of the advice given in the Scoping Opinion, the Applicant has also provided all necessary information to support a decision based on the Scoping Opinion

3.3. Species for modelling

  1. Collision risk estimates are presented for eight seabird species considered to be vulnerable to collision at OWFs (Furness et al., 2013): kittiwake Rissa tridactyla, herring gull Larus argentatus, lesser black-backed gull Larus fuscus, gannet Morus bassanus, Arctic tern Sterna paradisaea, common tern Sterna hirundo, little gull Hydrocoloeus minutus and great skua Stercorarius skua.
  2. Auks Alcidae and fulmar Fulmaris glacialis were not considered as they generally fly at low altitudes, well below the minimum height of turbine rotor blades (Cook et al., 2012; Johnston et al., 2014a; 2014b; Jongbloed, 2016). The list of seabird species taken forward for CRM was discussed and agreed through the Ornithology Road Map process. 

3.4. Parameterising the collision risk models

3.4.1.    Wind farm and turbine parameters

  1. Turbine and wind farm parameters were provided to HiDef from the Applicant within the PDE. Five different potential turbine designs, each of differing size, are being considered as part of the PDE: 14 MW, 15 MW, 18 MW, 21 MW and 24 MW.
  2. Information on the 14 MW turbine is based on that provided by Siemens Gamesa Renewable Energy (SGRE) and GE’s 14 MW Haliade X model. The parameters for the 15 MW turbine are based on the Vestas V236 turbine design parameters.
  3. The larger rated turbines (18 – 24 MW) are of two different rotor design types, reflecting two possible future design pathways; (A) wide chord and slower rotational speed and (B) narrower chord and faster rotational speed.
  4. The parameters for all turbine scenarios (rating and type) which are relevant to the CRM are given in Table 3.1   Open ▸ .
  5. The monthly wind availability is given in Table 3.2   Open ▸ and was calculated using the following cut-in/cut-out assumptions:
    • 14 MW: 4 – 28 ms-1 
    • 15 MW: 3 – 30 ms-1   
    • 18 MW: 3 – 30 ms-1 
    • 21 MW: 3 – 30 ms-1 
    • 24 MW: 3 – 30 ms-1 
  1. A mean monthly downtime of 3% was assumed for all turbine scenarios. The monthly time operational was calculated as the monthly wind available (%) minus 3% ( Table 3.2   Open ▸ ). Estimates of collision are provided based on the application of the large array correction factor.
Table 3.1:
Wind farm and turbine specifications for collision risk modelling for the Proposed Development.

Table 3.1: Wind farm and turbine specifications for collision risk modelling for the Proposed Development.

 

Table 3.2:
Monthly wind availability (%) (WA) and operational time (%) (OT) for turbines being considered for the Proposed Development.

Table 3.2: Monthly wind availability (%) (WA) and operational time (%) (OT) for turbines being considered for the Proposed Development.


3.4.2.    Seabird Densities

  1. The monthly densities of flying birds in the Development Array only (excluding the 16 km buffer of the Offshore Ornithology Study Area; Figure 3.1   Open ▸ ) were estimated using design-based strip transect methods from the HiDef digital aerial surveys conducted between March 2019 – April 2021. The estimates for all species were based on counts that had been apportioned for non-identified birds during the surveys; detail is provided in Technical Appendix 11.2: Ornithology Baseline Technical Report.
  2. Estimates of mean (Developer Approach) and maximum (Scoping Approach) monthly densities and pooled standard deviations (the latter only required for sCRM) for flying birds only have been used as input to the CRMs ( Table 3.3   Open ▸ and Table 3.4   Open ▸ ).

Figure 3.1:
Offshore Ornithological Study Area comprising the proposed Berwick Bank Development Array and 16 km buffer flown using digital video strip transects by HiDef March 2019 – April 2021.

Figure 3.1: Offshore Ornithological Study Area comprising the proposed Berwick Bank Development Array and 16 km buffer flown using digital video strip transects by HiDef March 2019 – April 2021.

 

Table 3.3:
Developer Approach: Mean monthly densities and pooled standard deviation (SD) of flying birds for eight species in the Proposed Development Array area only estimated from 25 months of baseline data collection.

Table 3.3:  Developer Approach: Mean monthly densities and pooled standard deviation (SD) of flying birds for eight species in the Proposed Development Array area only estimated from 25 months of baseline data collection.

 

Table 3.4:
Scoping Approach: Maximum monthly densities of flying birds for eight species in the proposed Development Array area only estimated from 25 months of baseline data collection.

Table 3.4: Scoping Approach: Maximum monthly densities of flying birds for eight species in the proposed Development Array area only estimated from 25 months of baseline data collection.

 

3.4.3.    Seabird biological parameters

  1. Discussions through the Ornithology Road Map process (RM 3, 28 September 2021; NatureScot advice, 7 October 2021) were used to decide upon sources of seabird morphological and behavioural parameters for (e.g., flight speed, wingspan; Table 3.5   Open ▸ ) to parameterise the CRMs.
  2. Body length, wingspan and flight speed measurements are sourced from Robinson (2005), Pennycuick (1997) and Alerstam et al. (2007). This information was not available for Arctic tern, so the morphological and behavioural parameters for common tern were used instead as they were considered a suitable proxy.
  3. Nocturnal activity is defined over the period of nautical twilight (the time when the centre of the sun is between 6° and 12° below the horizon). Nocturnal activity is converted to a 1-5 scale in the Band spreadsheet where 1 = 0%; 2= 25%; 3 = 50%; 4 = 75%; and 5 = 100%. NatureScot provided advice for gannet based on an analysis of nocturnal activity of tagged birds which showed there to be very low levels of activity after dark (Furness et al., 2018 and references therein). Furness et al. (2018) noted that nocturnal activity during the breeding season was 8%, and 3% during the non-breeding season; given these values, in the Band spreadsheet nocturnal activity during the breeding season is entered as 1.32, and during the non-breeding season as 1.12 following the 1-5 scale. For herring, lesser black-backed and little gulls, Arctic and common terns and great skua, the nocturnal activity scores were taken from Garthe and Hüppop (2004). The value for kittiwake is taken from the previously accepted Seagreen 1 (EIA Optimised Project Addendum 2018). All values follow the Scoping Opinion and agreement reached at the Ornithology Road Map 6 (10 May 2022) meeting.
  4. Flight type was set as flapping for all species except gannet, which was set to gliding following advice from NatureScot in their Scoping Consultation response (7 December 2021).

 

Table 3.5:
Morphological and behavioural parameters for the eight relevant species for the collision risk modelling for Berwick Bank.

Table 3.5:  Morphological and behavioural parameters for the eight relevant species for the collision risk modelling for Berwick Bank.   

*Robinson (2005); **Pennycuick (1997) and Alerstam et al., (2007); ***Seagreen 1 Scoping Report (EIA Optimised Project Addendum 2018);  †Hüppop (2004); ‡Furness et al., (2018).

3.4.4.    Avoidance rates

  1. Avoidance rates advised in the joint response of SNCBs (2014) are based on the Marine Scotland Science Avoidance Rate review (Cook et al., 2014). The review included quantitative and qualitative analyses of all available evidence for five priority species: gannet, kittiwake, lesser black-backed gull, herring gull and great black backed gull. The review built upon previous work of Cook et al. (2012).
  2. Bowgen and Cook (2018) utilise the results of the Bird Collision Avoidance (BCA) study which collected data on empirical estimates of bird behaviour at the operational Thanet Offshore Wind Farm (Skov et al., 2018) to propose seabird collision and avoidance rates.
  3. For the deterministic Band model, avoidance rates for all species were sourced from the SNCBs joint response on approved avoidance rates (SNCBs, 2014; Cook et al., 2014; Table 3.6   Open ▸ ). Use of SNCBs (2014) avoidance rates for the primary assessment was advised in the Scoping Opinion (4 February 2022). Furthermore, a 98% avoidance rate for gannet was used following RSPB’s consultation representation ( Table 3.6   Open ▸ ). 
  4. There are no SNCBs endorsed avoidance rates for kittiwake or gannet for the extended Band model (Option 3). Therefore, avoidance rates from Bowgen and Cook (2018) were used for comparison ( Table 3.7   Open ▸ ), noting that an avoidance rate for use in the extended model is not provided.
  5. For the sCRM, avoidance rates for kittiwake, gannet, herring gull and lesser black-backed gull were taken from Bowgen and Cook (2018; Table 3.7   Open ▸ ). Currently SNCBs advice on preferred avoidance rates for sCRM is not available, but agreement to use Bowgen and Cook (2018) was obtained through the Ornithology Road Map process and confirmed in the Scoping Opinion 4 February 2022. Avoidance rates for sCRM for the terns, little gull and great skua were set at 0.980.
  6. Avoidance rates from Cook (2021) were initially recommended for consideration during the Ornithology Road Map process (RM3, 28 September 2021). However, these avoidance rates have not been adopted by SNCBs to date, given concerns about the data sources and methodology used to produce them. Therefore, Cook (2021) avoidance rates have not been used.

 

Table 3.6:
Avoidance rates (± 2 SD) used for the deterministic basic (Options 1 and 2) and extended (Option 3) Band model (2012) as per SNCBs advice (SNCBs, 2014). An additional avoidance rate of 98% was included for gannets. Avoidance rates used in the deterministic Band model for the terns, little gull and great skua are taken from SNCBs (2014) and were set at 0.980.

Table 3.6:  Avoidance rates (± 2 SD) used for the deterministic basic (Options 1 and 2) and extended (Option 3) Band model (2012) as per SNCBs advice (SNCBs, 2014). An additional avoidance rate of 98% was included for gannets. Avoidance rates used in the deterministic Band model for the terns, little gull and great skua are taken from SNCBs (2014) and were set at 0.980.

 

Table 3.7:
Avoidance rates for each species used in the deterministic basic (Options 1 and 2) and extended (Option 3) Band model (2012) and stochastic collision risk models (with 95% confidence intervals) as per Bowgen and Cook (2018). For terns, little gull and great skua, avoidance rates were set at 0.980.

Table 3.7: Avoidance rates for each species used in the deterministic basic (Options 1 and 2) and extended (Option 3) Band model (2012) and stochastic collision risk models (with 95% confidence intervals) as per Bowgen and Cook (2018). For terns, little gull and great skua, avoidance rates were set at 0.980.

 

3.4.5.    Seabird flight height

  1. It was agreed through the Ornithology Road Map process (RM4, 8 December 2021) that the CRM should utilise the generic modelled flight heights from Johnston et al. (2014a; 2014b) for the primary assessment (Band Option 2 and 3). These flight height data were collated from seabird surveys at 32 OWFs in the UK and Europe. Most surveys were boat-based, and height measurements taken visually and assigned into height bands, to derive continuous flight height distributions for 25 seabird species. Site-specific flight height data for kittiwake collected during boat surveys within the proposed Development Array area are considered in Band Option 1 in Annex A.

3.5. Seasonal considerations

  1. In this report we define biologically distinct ‘bio-seasons’ following those outlined in NatureScot guidance (2020).
  2. Bio-seasons are defined as breeding and non-breeding:
  • Breeding season: birds are strongly associated with a nest site, including nesting, egg-laying and provisioning young; and
  • Non-breeding season: birds are dispersed and no longer strongly associated with colonies. This period subsumes the short ‘pre-breeding’ seasons defined separately in NatureScot (2020).
    1. Bio-seasons for each species are given in Table 3.8   Open ▸ to Table 3.15   Open ▸ . Little gulls do not breed in Scotland; hence no breeding season is defined. However, the use of NatureScot (2020) non-breeding season definitions presents issues for non-breeding season apportioning (Technical Appendix 11.5: Ornithology Apportioning Technical Report). Since non-breeding season apportioning is reliant on information for Biologically Defined Minimum Population Scales (BDMPS) (Furness, 2015), collision mortalities were also presented for the non-breeding seasons defined in Furness (2015) and used in further PVA for kittiwake and gannet, for which the autumn and spring passage and winter periods are defined within the non-breeding season ( Table 3.8   Open ▸ and Table 3.15   Open ▸ ).
    2. Estimates of the number of collisions for each bio-season were compiled from monthly estimates. Collision estimates for seasons that encompassed half-months were allocated proportionally within the season as agreed through the Ornithology Road Map process (RM3, 28 September 2021).
    3. Where the NatureScot (2020) breeding season for kittiwake or gannet overlaps with the Furness (2015) BDMPS non-breeding seasons, the NatureScot (2020) breeding season took precedence rather than the non-breeding season. As an example, both the NatureScot (2020) breeding season and Furness (2015) autumn migration period for kittiwake include the month of August, however, when presenting collision for each bio-season, the August collisions were attributed to the breeding season rather than the Furness (2015) non-breeding autumn migration.

 

Table 3.8:
Kittiwake bio-seasons based on NatureScot (2020)* and Furness (2015)**.

Table 3.8: Kittiwake bio-seasons based on NatureScot (2020)* and Furness (2015)**.

Table 3.9:
Herring gull bio-seasons based on NatureScot (2020).

Table 3.9: Herring gull bio-seasons based on NatureScot (2020).

Table 3.10:
Lesser black-backed gull bio-seasons based on NatureScot (2020).

Table 3.10: Lesser black-backed gull bio-seasons based on NatureScot (2020).

Table 3.11:
Gannet bio-seasons based on NatureScot (2020)* and Furness (2015)**.

Table 3.11: Gannet bio-seasons based on NatureScot (2020)* and Furness (2015)**.

Table 3.12:
Arctic tern bio-seasons based on NatureScot (2020).

Table 3.12: Arctic tern bio-seasons based on NatureScot (2020).

 

Table 3.13:
Common tern bio-seasons based on NatureScot (2020).

Table 3.13: Common tern bio-seasons based on NatureScot (2020).

Table 3.14:
Little gull bio-seasons based on NatureScot (2020).

Table 3.14: Little gull bio-seasons based on NatureScot (2020).

Table 3.15:
Great skua bio-seasons based on NatureScot (2020).

Table 3.15: Great skua bio-seasons based on NatureScot (2020).

 

4. Results

  1. The modelled worst-case scenario for each species is summarised in Table 4.1   Open ▸ , for both the Developer and Scoping Approaches.
  2. In all cases, the 14 MW x 307 turbines using the deterministic Band (2012) model resulted in the worst-case scenario. The view that using fewer, larger turbines as an effective measure for reducing collision (Johnston et al., 2014a; 2014b) was borne out by the modelling undertaken here. For all species, the number of collisions tended to decrease with increasing turbine size and was lower for Type B turbines (narrower chord and faster rotational speed) compared to Type A turbines (wider chord and slower rotational speed) ( Figure 4.1   Open ▸ ).   
  3. The collision mortality estimates for the 14 MW Type A turbine based on the sCRM outputs are provided for comparison in Table 4.1   Open ▸ ), noting that the scenario is not entirely equivalent to the worst case (previous paragraph 50) due to the different avoidance rates used. The full suite of collision estimates from the sCRM are provided in Annex C.

Table 4.1:
Worst-case estimates for each species identified from the deterministic Band CRM using the generic flight height data (Options 2 & 3) and SNCBs (2014) avoidance rates for the Developer Approach and Scoping Approach. For sCRM, the mortality estimates for the ‘equivalent’ worst case scenario are provided but noting that the avoidance rates are from Bowgen & Cook (2018).

Table 4.1:  Worst-case estimates for each species identified from the deterministic Band CRM using the generic flight height data (Options 2 & 3) and SNCBs (2014) avoidance rates for the Developer Approach and Scoping Approach. For sCRM, the mortality estimates for the ‘equivalent’ worst case scenario are provided but noting that the avoidance rates are from Bowgen & Cook (2018).

 

Figure 4.1:
Estimated numbers of annual collisions for each species, turbine rating and Type A and Type B for the Developer Approach (top) and the Scoping Approach (bottom).

Chart, line chart Description automatically generatedFigure 4.1: Estimated numbers of annual collisions for each species, turbine rating and Type A and Type B for the Developer Approach (top) and the Scoping Approach (bottom).

 

4.2. Kittiwake

  1. Monthly estimates of annual collisions for the worst-case for kittiwake are presented in Table 4.2   Open ▸ for both the Developer and Scoping Approaches. The total estimated annual number of collisions of kittiwake were 685 and 986 for the Developer and Scoping Approaches respectively.
  2. Results from using site-specific flight heights for kittiwakes from rangefinder and visual observer data and modelled using option 1 of Band (2012) were considerably lower (Annex B, Table 2   Open ▸ and Table 3   Open ▸ ). Based on rangefinder data, the mean estimated annual number of collisions for kittiwake using the Developer and Scoping Approaches were 56 and 81 birds respectively. Using the visual observer collected data, the annual mean increased to 225 and 324 kittiwakes for the Developer and Scoping Approaches respectively ( Table 3   Open ▸ ).   
  3. The estimated number of collisions was highest during May, which coincides with the second and third peaks in mean and max monthly densities of flying kittiwakes, with 4.91 birds/km2 (pooled SD ±0.86) and 6.73 birds/km2 (pooled SD ±1.06) respectively ( Table 3.3   Open ▸ and Table 3.4   Open ▸ ).
  4. Combining the estimated number of collisions across bio-seasons, shows it to be highest during the breeding season ( Table 4.3   Open ▸ ). However, the numbers of estimated collisions remain relatively high year-round throughout the non-breeding season.
  5. The estimated number of collisions presented in Table 4.2   Open ▸ and Table 4.3   Open ▸ were used in population modelling reported in Technical Appendix 11.6: Ornithology Population Viability Analysis.
  6. Annual collision estimates for kittiwakes for all turbine scenarios and avoidance rates using the Developer and Scoping Approaches are presented in Table 4.4   Open ▸ and Table 4.5   Open ▸ respectively.

 

Table 4.2:
Monthly estimated annual collisions for kittiwake in the Proposed Development Array for the worst-case design scenario (SNCBs avoidance rates, turbine 14 MW, Option 2), based on the Developer and Scoping Approaches. Estimates are presented using the mean avoidance rate (0.989) and for the mean avoidance rate ±2 standard deviations (SD) (0.002).

Table 4.2:  Monthly estimated annual collisions for kittiwake in the Proposed Development Array for the worst-case design scenario (SNCBs avoidance rates, turbine 14 MW, Option 2), based on the Developer and Scoping Approaches. Estimates are presented using the mean avoidance rate (0.989) and for the mean avoidance rate ±2 standard deviations (SD) (0.002).

Table 4.3:
Estimated number of collisions for kittiwake by season in the Proposed Development Array for the worst-case scenario (SNCBs avoidance rates, turbine 14 MW, Option 2).

Table 4.3:  Estimated number of collisions for kittiwake by season in the Proposed Development Array for the worst-case scenario (SNCBs avoidance rates, turbine 14 MW, Option 2).

*Using Furness (2015) BDMPS season definition.

**Using NatureScot (2020) non-breeding season definition.


Table 4.4:
Summary of estimated number of annual collisions for kittiwake from the Band model using the Developer Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

Table 4.4:  Summary of estimated number of annual collisions for kittiwake from the Band model using the Developer Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

Table 4.5:
Summary of estimated number of annual collisions for kittiwake from the Band model using the Scoping Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

Table 4.5:  Summary of estimated number of annual collisions for kittiwake from the Band model using the Scoping Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

 

4.3. Herring gull

  1. Monthly estimates of collisions for the worst-case scenario for herring gull are presented in Table 4.6   Open ▸ for both the Developer and Scoping Approaches.
  2. The estimated number of collisions was highest during July, when monthly densities of flying herring gulls were at their highest, with 0.32 birds/km2 (pooled SD ± 0.10) and 0.48 birds/km2 (pooled SD ±0.12) used in the Developer and Scoping Approach respectively ( Table 3.3   Open ▸ and Table 3.4   Open ▸ ).
  3. Combining the estimated mortality across bio-seasons, shows that the estimated number of collisions is highest during the breeding season ( Table 4.7   Open ▸ ).
  4. The estimated number of collisions presented in Table 4.6   Open ▸ and Table 4.7   Open ▸ were used in population modelling reported in Technical Appendix 11.6: Ornithology Population Viability Analysis.
  5. Annual collision estimates for herring gulls for all turbine scenarios, and avoidance rates using the Developer and Scoping Approaches are presented in Table 4.8   Open ▸ and Table 4.9   Open ▸ respectively.
Table 4.6:
Monthly estimated collisions for herring gull in the Proposed Development Array for the worst-case design scenario (SNCBs avoidance rates, turbine 14 MW, Option 2), based on the Developer and Scoping Approaches. Estimates are presented using the mean avoidance rate (0.995) and for the mean avoidance rate ±2 standard deviations (SD) (0.001).

Table 4.6: Monthly estimated collisions for herring gull in the Proposed Development Array for the worst-case design scenario (SNCBs avoidance rates, turbine 14 MW, Option 2), based on the Developer and Scoping Approaches. Estimates are presented using the mean avoidance rate (0.995) and for the mean avoidance rate ±2 standard deviations (SD) (0.001).

 

Table 4.7:
Estimated number of collisions for herring gull by season in the Proposed Development Array for the worst-case scenario (SNCBs avoidance rates, turbine 14 MW, Option 2).

Table 4.7:  Estimated number of collisions for herring gull by season in the Proposed Development Array for the worst-case scenario (SNCBs avoidance rates, turbine 14 MW, Option 2).

Table 4.8:
Summary of estimated number of annual collisions for herring gull from the Band model using the Developer Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

Table 4.8: Summary of estimated number of annual collisions for herring gull from the Band model using the Developer Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.


Table 4.9:
Summary of estimated number of annual collisions for herring gull from the Band model using the Scoping Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

Table 4.9: Summary of estimated number of annual collisions for herring gull from the Band model using the Scoping Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

 

4.4. Lesser Black-backed gull

  1. Monthly estimates of collisions for the worst-case for lesser black-backed gull are presented in Table 4.10   Open ▸ for both the Developer and Scoping Approaches.
  2. The estimated number of collisions was highest during July, when monthly densities of flying lesser black-backed gulls were at their highest, with 0.13 birds/km2 (pooled SD ±0.03) and 0.18 birds/km2 (pooled SD ±0.04) respectively ( Table 3.3   Open ▸ and Table 3.4   Open ▸ ).
  3. Combining the estimated mortality across bio-seasons, shows that collisions are exclusive to the breeding season ( Table 4.11   Open ▸ ).
  4. The estimated number of collisions presented in Table 4.10   Open ▸ and Table 4.11   Open ▸ were used in population modelling reported in Technical Appendix 11.6: Ornithology Population Viability Analysis.
  5. Annual collision estimates for lesser black-backed gulls for all turbine scenarios and avoidance rates using the Developer and Scoping Approaches are presented in Table 4.12   Open ▸ and Table 4.13   Open ▸ respectively.

Table 4.10:
Monthly estimated collisions for lesser black-backed gull in the Proposed Development Array for the worst-case scenario (SNCBs avoidance rates, turbine 14 MW, Option 2), based on the Developer and Scoping Approaches. Estimates are presented using the mean avoidance rate (0.995) and for the mean avoidance rate ±2 standard deviations (SD) (0.001).

Table 4.10: Monthly estimated collisions for lesser black-backed gull in the Proposed Development Array for the worst-case scenario (SNCBs avoidance rates, turbine 14 MW, Option 2), based on the Developer and Scoping Approaches. Estimates are presented using the mean avoidance rate (0.995) and for the mean avoidance rate ±2 standard deviations (SD) (0.001).

 

Table 4.11:
Estimated number of collisions for lesser black-backed gull by season in the Proposed Development Array for the worst-case scenario (SNCBs avoidance rates, turbine 14 MW, Option 2).

Table 4.11:  Estimated number of collisions for lesser black-backed gull by season in the Proposed Development Array for the worst-case scenario (SNCBs avoidance rates, turbine 14 MW, Option 2).

Table 4.12:
Summary of estimated number of annual collisions for lesser black-backed gull from the Band model using the Developer Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

Table 4.12:  Summary of estimated number of annual collisions for lesser black-backed gull from the Band model using the Developer Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

*The estimated collisions for 14 MW, 15 MW and 18 MW turbines type A are 6.35, 5.57 and 6.02 per annum, respectively.


Table 4.13:
Summary of estimated number of annual collisions for lesser black-backed gull from the Band model using the Scoping Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

Table 4.13:  Summary of estimated number of annual collisions for lesser black-backed gull from the Band model using the Scoping Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

4.5. Gannet

  1. Monthly estimates of collisions for the worst-case for gannet are presented in Table 4.14   Open ▸ for both the Developer and Scoping Approaches.
  2. The estimated number of collisions was highest during July, which coincides with the second highest monthly densities of flying gannets, estimated at 1.51 birds/km2 (pooled SD ±0.23) and 1.53 birds/km2 (pooled SD ±0.17) in the Developer and Scoping Approach respectively ( Table 3.3   Open ▸ and Table 3.4   Open ▸ ).
  3. Combining the estimated mortality across bio-seasons, shows that the estimated number of collisions is highest during the breeding season ( Table 4.15   Open ▸ ).
  4. Estimated number of collisions presented in Table 4.14   Open ▸ and Table 4.15   Open ▸ were used in population modelling reported in Technical Appendix 11.6: Ornithology Population Viability Analysis.
  5. Annual collision estimates for gannets for all turbine scenarios and avoidances rates using the Developer and Scoping Approaches are presented in Table 4.16   Open ▸ and Table 4.17   Open ▸ respectively.
  6. Monthly estimates of collisions for the breeding season for gannet using an avoidance rate of 0.980 and the Band Option 2 (SNCBs, 2014) are presented for context in Table 4.18   Open ▸ for both the Developer and Scoping Approaches.

Table 4.14:
Monthly estimated collisions for gannet in the Proposed Development Array for the worst-case scenario (SNCBs avoidance rates, turbine 14 MW, Option 2), based on the Developer and Scoping Approaches. Estimates are presented using the mean avoidance rate (0.989) and for the mean avoidance rate ±2 standard deviations (SD) (0.002).

Table 4.14:  Monthly estimated collisions for gannet in the Proposed Development Array for the worst-case scenario (SNCBs avoidance rates, turbine 14 MW, Option 2), based on the Developer and Scoping Approaches. Estimates are presented using the mean avoidance rate (0.989) and for the mean avoidance rate ±2 standard deviations (SD) (0.002).

Table 4.15:
Estimated number of collisions for gannet by season in the Proposed Development Array for the worst-case scenario (SNCBs avoidance rates, turbine 14 MW, Option 2).

Table 4.15:  Estimated number of collisions for gannet by season in the Proposed Development Array for the worst-case scenario (SNCBs avoidance rates, turbine 14 MW, Option 2).

*Using Furness (2015) BDMPS season definition.

**Using NatureScot (2020) non-breeding season definition.

Table 4.16:
Summary of estimated number of annual collisions for gliding gannet from the Band model Options 2 and 3 using the Developer Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

Table 4.16:  Summary of estimated number of annual collisions for gliding gannet from the Band model Options 2 and 3 using the Developer Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.


Table 4.17:
Summary of estimated number of annual collisions for gliding gannet from the Band model Options 2 and 3 using the Scoping Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

Table 4.17:  Summary of estimated number of annual collisions for gliding gannet from the Band model Options 2 and 3 using the Scoping Approach and generic flight height data, for turbine Type A (wide chord and slow rotational speed) and B (narrow chord and fast rotational speed). Avoidance rates are from SNCBs (2014) and Bowgen and Cook (2018). Estimates are rounded to the nearest whole.

Table 4.18:
Monthly estimated collisions for gliding gannet in the Proposed Development Array for the worst-case scenario (mean avoidance rate of 0.980, turbine 14 MW, Option 2) during the breeding season only, based on the Developer and Scoping Approaches and generic flight height.

Table 4.18: Monthly estimated collisions for gliding gannet in the Proposed Development Array for the worst-case scenario (mean avoidance rate of 0.980, turbine 14 MW, Option 2) during the breeding season only, based on the Developer and Scoping Approaches and generic flight height.

*March collision estimates presented are for the entire month. Gannet breeding season is estimated to start in mid-March (NatureScot, 2020), therefore, only half of the collisions for the month of March were counted in the total breeding season collision estimates.