Harbor Seal - Tidal Turbine Collision Risk Models. An Assessment of Sensitivities.


Title: Harbor Seal - Tidal Turbine Collision Risk Models. An Assessment of Sensitivities.
Publication Date:
March 14, 2016
Pages: 57
Technology Type:

Document Access

Attachment: Access File
(10 MB)


Wood, J.; Joy, R.; Sparling, C. (2016). Harbor Seal - Tidal Turbine Collision Risk Models. An Assessment of Sensitivities. Report by SMRU Consulting. pp 57.

There has been growing interest in generating electricity from tidal currents, but there are still concerns about the potential environmental effects of tidal turbines. One of these concerns is the risk of collision by marine mammals with spinning tidal turbines. Most estimates of marine mammal collision risk with tidal turbines have used either an Encounter Risk Model (ERM) which is based on a predator-prey model, or a Collision Risk Model (CRM) which was first developed for predicting bird collisions with wind turbines. CRM estimates are based on transit rates of the animals and the probability of collision for each transit.


In order to explore the sensitivities of the collision risk models to various inputs, we analyzed data that were available to us from the Strangford Lough MCT SeaGen project. We had access to tagged seal data from 2006, 2008 and 2010 as well as turbine and current data starting after the installation of the turbine in 2008 through 2010. Seal tags provided information on the timing and location of seals as well as their dive depth. We also had measures of current speed, direction and turbine RPM. Based on tidal data for Strangford Lough, we estimated current speed and direction at other times during which we had seal tag data. Based on analyses of the seal tag and current data, the following patterns emerged:

  • Seal vectors (i.e. seal swim speed and direction from GPS tag locations):
    • Given the dominant aspect of current in a tidal environment, the direct use of seal vectors over ground (as measured from one GPS location to the next) is cautioned.
    • Calculating seal vectors through the water, but accounting for the current vector and the seal vector over ground is more appropriate.
    • In this dataset, seals swam in all directions over ground and in relation to the current, however, seals almost exclusively swam into the current.
    • Seal swim speed through the water tended to increase with increasing current velocity.
  • Seal dives:
    • Seal dives in Strangford Lough tended to follow a ‘U’ shape and were similar to dives reported at other sites.
  • Seal habitat use:
    • Seal habitat use across Strangford Lough was not uniform.
    • Seals in Strangford Lough tended to do most of their diving in areas outside of the highest current flow areas.
  • Seal Avoidance:
    • Using Brownian Bridge methods to interpolate seal movement between GPS locations suggests that within 200 m of the turbine, ~66% of seals in 2008 and 2010 avoided the area, when compared to 2006.
    • Caution should be used in interpreting this estimate of avoidance as there is a great deal of inter-individual difference between seals in each year of data, this trend is inferred from few tagged seals.


The exploration of collision risk model sensitivities showed the following trends:

  • Use of the tip speed ratio of the SeaGen turbine decreased collision risk by 12%
  • Use of the turbine RPM across a tidal cycle instead of average RPM increased collision risk by 5%. 
  • Use of seal swim speeds through the water measured in Strangford Lough increased collision risk by 3%.
  • Seal swim direction (upstream vs downstream) increased CRM estimates by 10%
  • The assumption of a ‘U’ shaped as opposed to a ‘V’ shaped dive, decreased collision risk estimates by 63%, but it was clear that the dive data from Strangford Lough, and other tidal sites, that seals consistently use ‘U’ shaped dives in these areas.
  • Seal ‘density’ had the biggest effect on collision risk estimates. Use of measured transit rates in Strangford Lough, reduced collision risk estimates by 27% from estimates based on average seal density.
  • Avoidance has a direct multiplicative effect and therefore reduced collision risk by 66%.


It is clear from the Strangford Lough data, that seal tag data can greatly inform and improve collision risk estimates, however, it would be preferable for the growth of the tidal turbine industry if such in depth research were not needed at every tidal turbine site. There are four other potential tidal turbine sites, in addition the Strangford Lough, where seals (harbor or grey seals) have been tagged. These data could provide needed information to verify if seals use these different habitats in similar or different ways. To date, the analyses of these data have shown the following trends:

  • There is a high degree of inter-individual variation in the use of tidal areas.
  • Local abundance varies by tidal cycle and thus collision risk is not equal over the tidal cycle.
  • Depth distributions of diving seals are similar across sites with most time spent at the surface or seabed.


Seal morphometric data from the San Juan County Marine Mammal Stranding Network may help inform models that predict the consequences of collision. The harbor seals in this dataset are on average small compared to other populations. The majority of adults were classed as in good nutritional health while the majority of subadults and pups were in poor nutritional state.


Based on the above findings we suggest the following priorities for future work:

  • Avoidance needs to be further investigated to understand how it varies with distance from turbine, across individual, tidal state and locations.
  • Fine scale habitat use needs further refinement as this also has large implications in collision risk. This includes inter-individual variability, as well as variation across tidal state, current speeds, current directions and across sites.
Find Tethys on FacebookFind Tethys on Twitter
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.