Guidance Note 7: Guidance to support Offshore Wind Applications: Marine Ornithology - Advice for assessing collision risk of marine birds

Published: 2023
Version 3: April 2025

This document is part of a series of guidance notes available on Advice on marine renewables development - marine ornithology.

This guidance note sets out our advice on collision risk modelling. Please see Guidance Note 1 - Guidance to support Offshore Wind Applications: Marine Ornithology which provides the context within which this guidance note is provided. Check the log of updates in guidance note 1 to ensure the most current version is being followed. We expect each and every developer to adhere to this guidance, including the recommended parameters to provide consistent and comparable results for all Scottish projects to be assessed as an individual project and cumulatively.  This guidance note should be used in conjunction with the Cumulative Effects Framework (CEF).

Please note this guidance was written prior to the development of a NatureScot position on how to account for the ongoing Highly Pathogenic Avian Influenza (HPAI) mortality event within an impact assessment. We are currently developing this advice and we will provide updates and guidance as they become available. In the interim please contact marineenergy@nature.scot  should you have any specific queries.

1. Introduction

Collision risk modelling (CRM) is the approach used to quantify the risk of collision with offshore wind farm turbines. This part of the assessment process for offshore wind farms has been the focus of continual research and development in the last decade. In addition to the 2024 Joint advice note from the Statutory Nature Conservation Bodies (SNCBs) regarding bird collision risk modelling for offshore wind developments , this guidance note specifically sets out  NatureScot’s recommendations for good practice impact assessment using CRM and signposts the most recent resources we expect to be used within the impact assessments for Scottish casework.  

We acknowledge that there is ongoing research on this topic and that this note will therefore be subject to updates. Check the log of updates in our guidance note 1 to ensure the most current version is being followed. 

This guidance note covers all seabirds except Procellariforms. Should you be required to undertake an assessment on procellariforms we advise you to contact NatureScot.  Migratory birds (other than seabirds) are assessed differently, using a migratory CRM – this work is being led by Marine Directorate. The second version of this guidance note takes account of the Ozsanlev-Harris et al (2023) and Pavat et al. (2023) studies, and provides input parameters for additional species.

2. Collision Risk Models, Options and Scenarios

NatureScot recommends the use of the stochastic collision risk model (sCRM) developed by Masden, E. 2015. Developing an avian collision risk model to incorporate variability and uncertainty. Scottish Marine and Freshwater Science. 6(14) and expanded into the sCRM tool (MacGregor et al. 2018), commissioned by Marine Scotland.

We advise the use of:

• the 2022 update to the sCRM tool shiny app (Caneco 2022). This update should also be used to run deterministic output (with seed specified to enable repeatability). We require that outputs for both stochastic and deterministic CRM are presented using this tool. 

The sCRM outputs are provided for the basic Band (options 1 and 2) and extended (option 3) models. We note there are discrepancies between the original offshore Band (2012) descriptions of options and those used in the sCRM. These are described by the sCRM user manual as:  

• Option 1 uses the ‘basic’ model and uses a simple proportion of birds within the rotor reach / collision height also known as proportion at collision height (PCH), applied uniformly over the rotor (the proportion is entered into the tool either using generic or site-specific data).
• Option 2 uses the ‘basic’ model but uses the flight height distributions entered by the user (and therefore can be derived from site-specific or generic datasets). However, this is an addition to option 1 as it uses Q2r (proportion of birds flying within at risk height band) instead of PCH; where Q2r is a calculated PCH from a flight height distribution.
• Option 3 uses the ‘extended’ model and uses the flight height distributions entered by the user (and therefore can be derived from site-specific or generic datasets). This is a further addition to option 2 as it includes a flux value within the rotor swept area, based on the flight height distribution.
• ‘Option 4’ from the original offshore Band model has not been included in either Masden 2015 or the sCRM tool but is the equivalent of using the extended model with site-specific flight height distributions. 

We note that the extended model can use either site-specific or generic datasets. ‘Option 3’ is typically referred to as the use of the extended model with the generic dataset Johnston, A., Cook, A.S.C.P., Wright, L.J., Humphreys, E.M., Burton N.H.K. 2014. Modelling flight heights of marine birds to more accurately assess collision risk with offshore wind turbines – Corrigendum. Journal of Applied Ecology. 51L 1126-1130. The site specific data can be presented, but a generic data-based assessment must be presented and used as the basis for the impact assessment.

There will be a number of windfarm design scenarios considered for the application. The terms ‘most likely scenario’ (MLS) and ‘worst-case scenario’ (WCS) refer to the project design envelope and the identification of what is likely to be built out and what would be the worst-case scenario from the range of turbine design options provided within the project design envelope. Agreement should be reached with NatureScot and Marine Directorate (MDLOT) on what MLS and WCS scenarios are to be used in advance of final modelling being undertaken, noting these may not be the same for all species.

NatureScot advise that as a minimum the following two scenarios should be modelled and results presented for each CRM species: 

1. Most likely scenario (MLS) -  Option 2 (using the generic flight height dataset)
2. Worst case scenario (WCS) -  Option 2 (using the generic flight height dataset) 

We require that both breeding and non-breeding season totals for all CRM species, as well as the annual total, are calculated from the monthly outputs (see section 7 and 8).   See below for presentation guidance (sections 7 and 8).   

3. Biological parameters

We advise the values presented in Appendix 1 are used as biological parameters for the sCRM tool. BTO Birdfacts are currently the main reference source for biological (bird) parameters used in the collision models, i.e. body length and wing span. The default values for these parameters are presented in the sCRM tool.  

Nocturnal activity scores remain the same as those used in the Scottish Round 3 / STW applications (Garthe & Hüppop, 2004), other than for gannet, where values derived from the review of tagged birds (Furness et al. 2018) should be used. Any deviation from these values will require consultation with NatureScot, with any additional data sources presented including clear reference to which definition of nocturnal is being used, noting that our preference is for nautical twilight to be used. 

At present we continue to rely on published data for flight speeds (i.e. Pennycuick 1997; Alerstam et al. 2007), however we recognise more recent studies are contributing to new evidence. Any changes to flight speed would require changes to any avoidance rates used. Until more recent studies are published and agreed the default values for each species should be used. 

4. Densities

Monthly density estimates of flying birds should be derived from site characterisation surveys (as described in Guidance note 2: Baseline Characterisation Surveys and Reporting). The sCRM tool provides three approaches for estimating the variability for monthly density data. We advise that 1000 samples from a distribution of mean densities (e.g. from a bootstrapped sample) is used.

Where stochastic models have been used we require a clear statement as to which variability approach has been chosen and should the first or second approach be used, this will require justification. The bootstrapped data should be provided to enable the modelling to be re-run and the outputs checked. 

We advise estimates of the number of collisions for each season are compiled from monthly estimates. Collision estimates for seasons that encompass half-months should then be allocated proportionally within the relevant season. For months that were split between seasons, e.g. April for kittiwake, 50% of collisions were allocated to the breeding season total and 50% to the non-breeding season total.

5. Flight height

Johnston et al. (2014) Modelling of flight heights of marine birds to more accurately assess collision risk with offshore wind turbines. Journal of Applied Ecology (with associated corrigendum) currently remains the recommended reference for generic flight heights and is the default within the sCRM tool. We acknowledge uncertainty remains around best practice for flight height data collection methods, primarily due to absence of agreed validation of techniques. Further discussion and agreement on use of flight height data derived from the site-specific surveys is required. If site-specific flight height data are to be presented for context we expect a full description of method, accuracy, precision and comparison with Johnston et al. (2014), with explanation of any differences to inform discussions with NatureScot. 

We note that use of site-specific flight height for sCRM requires recalculations of avoidance rates.  Our assessment will be based on the use the generic flight height data.

6. Avoidance rates

Avoidance rate estimates calculated specifically for the stochastic CRM have been produced by Ozsanlev-Harris et al. (2023), these rates should be used within the CRM with +/- 2 standard deviations. This study incorporates collision data from all suitable terrestrial, coastal and marine offshore wind farms.  The Joint SNCB guidance note (2024) provides justification for the  value used for each species.

For species that have not previously been considered in offshore assessments, i.e. Arctic skua, we will be recommending that the ‘All Gull’ rates should be used (refer to appendix 1). Any deviations from this advice will require clear justification and evidence.

7. Macro-avoidance

For species such as kittiwake and gannet where both collision risk and distributional responses (displacement and barrier effects) are considered we are aware that work is ongoing to understand how best to combine these impacts. Pavat et al. (2023) have developed a macro avoidance rate for gannet which is applied first, for CRM modelling, before collisions are calculated (Consideration of avoidance behaviour of northern gannet (Morus bassanus) in collision risk modelling for offshore wind farm impact assessments, Natural England Commissioned Report NERC512, 2023). This can be used to alter the input densities of flying birds by a macro-avoidance rate and then run the CRMs using a within OWF only avoidance rate. 

At present we can accept this approach for the non-breeding season but not for the breeding season. We consider that there is insufficient evidence regarding gannet behaviour around wind farms near Scottish SPA’s in the breeding season and further work is required to resolve this. As a result, our current approach is precautionary, advising that both collision and displacement impacts are assessed for gannet. 

The intention is to revisit this additive approach, in light of more advanced techniques for discriminating between barrier, macro-avoidance and displacement effects. We intend to use ongoing current-post-consent monitoring in Scotland to provide valuable information and enable us to have more confidence in applying a gannet macro avoidance rate in the breeding season. 

In the interim, we advise applying macro avoidance of 70% in the non-breeding season, based on the outputs from Pavet et al. (2023)

8. Presentation of CRM - Breeding Season

Our seasonal definitions guidance note should be used for determining which months to include for each species’ breeding season. A summary of the CRM results should be presented in the main Ornithology chapter of the EIA Report and HRA. We provide an example of a summary table in Appendix 2. The breeding season total should be calculated from the monthly totals. The table (Appendix 2) should be utilised for each of the two scenarios (MLS – option 2 and WCS – option 2) and include the stochastic and deterministic outputs for each collision risk species. 

We expect the monthly outputs (as per the sCRM tool) for a deterministic model to be presented within a technical ornithology appendix in the EIAR/HRA. The seed run for the deterministic models should be clearly stated along with input parameters.  

We advise that the pdf produced by the Shiny App tool should be presented as an Appendix. Presentation of results should be accompanied by input values used, to enable NatureScot to rerun the model, if we deem it necessary to do so. The seed run should be clearly stated in the outputs to allow exactly the same model to be run again.  

9. Presentation of CRM - non breeding season

We advise that the CRM should be presented for the non-breeding season for all species at collision risk, these periods should be defined using our Guidance Note 9: Seasonal periods for birds in the Scottish Marine Environment.

The non-breeding assessment should be presented, as per the breeding season guidance (above – section 7) – however, as described in our Guidance Note 10: Advice for apportioning impacts to breeding colonies these collisions should be considered in the context of the regional populations as defined by the Furness, R.W. 2015. Non-breeding season populations of seabirds in UK waters: Population sizes for Biologically Defined Minimum Population Scales (BDMPS). Natural England Commissioned Report: 164.

10. Migratory collision risk assessments

An updated review of migratory routes and vulnerabilities across the UK has been published by Marine Directorate and The Crown Estate. This work also includes development of a stochastic migration CRM tool (known as mCRM) to enable quantitative assessment of risks to migratory SPA species including swans, geese, divers, seaduck and raptors. The Scottish Government’s strategic study of collision risk for birds on migration Work Package 1: Strategic review of birds on migration in Scottish waters should be used alongside the associated Stochastic Collision Risk Model for migratory species.

11. Future updates to this guidance note

This guidance note will be reviewed and updated as new evidence becomes available, i.e. following publication of relevant research projects, including but not limited to: 

• Post Consent Monitoring
• The Cumulative Effects Framework
• The Migratory Collision Risk work
• Updates to the joint SNCB guidance note
• Any updates to Ozsanlev-Harris et al will be reflected in an update to this guidance note.

 Appendices