Environmental Engineering Reference
In-Depth Information
Collision Mortality Model Input Data
We used the Scottish Natural Heritage (SNH) “Band” model (Band et al. 2007 ) to
estimate the mortality of migratory shorebirds at the HMR.
The following inputs where required:
Total number of birds flying through the general area;
Percentage of birds flying over the wind farm;
Percentage of birds flying at rotor swept height (RSH - set between 50 and
150 m);
Wind farm and turbine parameters and the average number of turbines met by a
bird at risk;
The turbine collision risk probability for a bird flying through a rotor; and
Avoidance rates.
In the caucus statement of the Expert Shorebird Group (dated 27 April 2010) the
parties' experts agreed that a reworked 'Monte Carlo' implementation of the Band
model was appropriate as a collision risk model. The experts also agreed on the
general structure of the model and the methods of determining the inputs, and
agreed on most of the input values for the reworked model.
A common approach to allow for uncertainty of the inputs into the Band model
has been to use conservative, precautionary or worst-case scenario values as model
inputs. This compounding of precautionary values as model inputs can lead to over-
conservative estimates of bird mortality. This is because the individual values used
as inputs, sometimes already unlikely in themselves, have only an extremely small
chance of all occurring at once. One way to approach this problem and quantify the
uncertainty is to use Monte Carlo risk analysis. In the Monte Carlo approach, model
inputs are entered as distributions, which reflect the uncertainty in those values. In
the analysis, a series (usually numbering in the thousands) of possible scenarios (or
iterations) is compiled. In every iteration, each model input is picked at random
from its possible range of values with the more likely values chosen more often (this
is the input distribution). This provides the opportunity for high and low values to
counteract each other on occasion instead of always combining in the worst
possible way.
Running these iterations through the model allows an estimate of:
(a) The most likely number of casualties; and
(b) The highest value the true mortality rate is likely to take.
Because the HMR farm site is so large, the Expert Shorebird Group also agreed
that it would be more informative to design the new HMR model on a cluster
basis, effectively treating each cluster as a separate wind farm, with inputs for
each cluster. Radar flight trail data from the surveys were used (each migration
was considered and analysed separately), with 'boxes' placed around the indi-
vidual clusters to define each one. A line perpendicular to the coast was drawn
Search WWH ::




Custom Search