Avian Collision Risk Models
Collision risk models (CRMs) estimate the number of bird collisions that are expected to occur at wind farms. These estimates can be used to inform wind farm planning and mitigation of expected effects. CRMs have been used in relation to both land-based and offshore wind energy development, often to assess risk before a project is built. In some situations, they are also used in place of mortality monitoring to determine likely bird mortality rates at a wind farm in order to inform mitigation requirements or assess cumulative regional effects.
The first CRM was published in 1983. As of 2024, at least 52 different CRMs had been developed (Cook et al. 2025). Models typically include:
(1) an estimate of the number of birds exposed to collision risk (e.g., total numbers of birds passing through a wind farm or turbine rotor-swept zone within a set time period) and/or
(2) an estimate of the probability that an exposed bird will collide with the turbine (Cook et al. 2025).
A range of specific parameters are used to help inform these estimates, often including data on wind farm and turbine design, bird morphology and behavior, population size and demography, and landscape/habitat characteristics.
Word cloud indicating the relative frequency with which parameters are used by different CRMs. Similar parameters have been grouped (e.g., rotor diameter also includes models that consider rotor radius, which is derived from the diameter). Copied with permission from Cook et al. (2025).
The Band Model and Derivatives
A commonly used family of CRMs derive from models developed by Bill Band (Band 2007, 2012). StochCRM, SCRAM, and several other current models (e.g., Smales et al. 2013) all derive from this CRM framework. Band-type models use data on turbine blades (size, shape, speed, percentage of time rotating, etc.) and data on the bird species of interest (size, type of flight, speed, avoidance rate, etc.) to estimate the probability of a blade and bird being present in the same physical space at the same time (i.e., colliding). This physical model remains reasonably consistent across CRMs (Cook et al. 2025), though CRMs within this group vary in how they are estimating exposure (i.e., the number of birds available to interact with a turbine blade).
Model Validation
Few CRMs have been validated to assess whether their estimates of risk are accurate, largely due to the challenges inherent to such efforts. For example, it is not possible to collect carcasses at offshore wind farms, which makes it difficult to determine actual mortality rates to assess the accuracy of CRM predictions. Even at land-based wind farms, the relative infrequency of collisions makes validation studies difficult and expensive. As a recent paper noted, “attempts to validate the predictive power of CRMs…have produced results that are, at best, ambiguous (de Lucas et al., 2008; Ferrer et al., 2012). As a result, there is often a lack of confidence in the use of model outputs for consenting decisions (Searle et al., 2023).” As such, additional data are needed from post-construction monitoring efforts to better validate collision risk models for a range of species, wind farm designs, and geographic locations.