Collision Risk Models

From Parameter Uncertainty to Process Simulation: Rethinking Collision Risk Models

Offshore Wind and Ornithology Conference, Glasgow, 1 December 2026

Conventional collision risk models (CRMs) play an important role in assessing the potential impacts of offshore wind developments on birds. Band-based approaches estimate expected collisions using assumptions about exposure, flight heights, avoidance behaviour and turbine characteristics.

My current research explores how agent-based modelling (ABM) might complement these established approaches by making the processes that generate collision risk explicit.

Rather than replacing the Band collision calculation, the model retains it as a conditional probability calculation within a simulation of individual birds moving through a turbine array. This allows us to investigate how spatial arrangements, encounter histories and behavioural responses influence exposure and expected collisions.

The initial proof of concept, developed in NetLogo, simulates kittiwake movements through a nine-turbine array. It explores three related mechanisms:

• Geometry: Different flight trajectories generate different patterns of rotor exposure.

• Selection: When collisions remove birds, the composition of surviving trajectories can change downstream.

• Behaviour: Simple local turning responses alter trajectories and subsequent encounters.

The illustrations below explain the modelling approach and present results from an initial paired simulation comparing straight-line flight with local turning responses.

These results are exploratory. They demonstrate how changes in movement rules can alter simulated exposure and expected collisions, not measured avoidance rates or validated predictions of collision mortality.

The longer-term objective is to develop empirically informed process models that complement existing assessment methods and support investigation of alternative turbine configurations and potential mitigation strategies.