Abstract
Collision Risk Models (CRM) are used to assess impacts on seabird populations in all offshore wind farms Environmental Impact Assessments (‘EIA‘) and Habitats Regulations Appraisals (‘HRA‘) in the UK. Existing models are unable to properly incorporate uncertainty in the input parameters into calculations of uncertainty in the collision prediction and consequently are not expressed in the outputs. Uncertainty in predicted collision has resulted in the delayed deployment of offshore wind projects, with projects being reduced in size or even cancelled. Not incorporating uncertainty when it is known to occur may be failing to meet the requirement from the European Court of justice to use, “…the best scientific knowledge in the field…”. This project aimed to create a CRM that incorporates variability in input parameters correctly into a predicted collision impact with estimated variability. In order to produce a model that was fit for purpose, stakeholders were consulted through a questionnaire-based survey. The survey results section was in seven parts, each asking about different aspects of the CRM. These were: CRM concept, user experience, CRM inputs, CRM operation, CRM outputs, CRM error checking and CRM improvements. The survey, while taking in to account the scope of the project, resulted in the following changes requested by stakeholders:
• Create a user-friendly interface for non-R users; • Speed up the code; • The number of turbines should be a user input; • Output predicted collision probability data; • Seasonal (as well as monthly & annual) assessment (default + user defined); • Error checking inputs and collision probability; and, • Monthly or seasonal flight height inputs.
The new stochastic CRM (sCRM) was based on the code written by Masden (2015), but had to be compatible with the Band (2012) offshore CRM. Testing showed that the predictions of the Masden (2015) code matched the predictions of the Band (2012) Excel spreadsheets for Option 1, but that differences in outputs for Options 2 and 3 arose because of a calculation error in Masden (2015) code. Consequently, the sCRM was based on an updated, and streamlined, version of the Masden (2015) code. The new sCRM was produced in two forms: Firstly, a Shiny app based on the R-code, available as an online tool, which can be run from: https://dmpstats.shinyapps.io/avian_stochcrm/ Secondly, the Shiny app can be downloaded as a package and run locally in a browser. It can be downloaded from: https://github.com/dmpstats/stochCRM