MODELLING SEABIRD COLLISION RISK WITH OFF-SHORE WINDFARMS M. Mateos, - PowerPoint PPT Presentation

MODELLING SEABIRD COLLISION RISK WITH OFF-SHORE WINDFARMS M. Mateos, G.M. Arroyo, J.J. Alonso del Rosario

Objectives To develop a stochastic model of avian collision risk at wind farms A case study To obtain Factors probabilities of collision risk To estimate mortality rates

Objectives To develop a stochastic model of avian collision risk at wind farms A case study To obtain Factors probabilities of collision risk To estimate mortality rates

THE MODEL Stochastic character, based on Montecarlo simulation

Case study: The Strait of Gibraltar NORTHEN GANNET Morus bassanus CORY’S SHEARWATER BALEARIC SHEARWATER Calonectris diomedea Puffinus mauretanicus

Case study: The Strait of Gibraltar Strait of Gibraltar Gibraltar Ceuta Algeciras TARIFA ISLAND VISUAL & RADAR CENSUSES Antenna 12 ft TX 3.050 MHz 286 dB 22-28 rpm 30 KW

THE MODEL: The wind farm as a risk window RISK WINDOW

THE MODEL: The wind farm as a risk window Wind farm Distance to coast Columns dimensions Rows Number of rows: 3, 6, 9 Number of columns: 6, 10, 14 Distance between rows: 400, 700, 1000 m Distance between columns: 300, 600, 900m Distance to coast: 1, 5, 10 km

THE MODEL: The wind farm as a risk window Distance to coast • Different species Columns F(y) • Different wind conditions • Headwind low (0 to 5 in Beaufort scale) Flight • Headwind high (6 to 9 in Beaufort scale) Rows direction • Tailwind low (0 to 5 in Beaufort scale) • Tailwind high (6 to 9 in Beaufort scale) • Uniform distribution Horizontal distribution of the migratory passage

THE MODEL: The wind farm as a risk window Distance to coast Columns F(y) Rows Montecarlo simulation

THE MODEL: The wind farm as a risk window Distance to coast Columns F(y) Rows • 60% • 70% Wind farm • 80% Avoidance rate Based on avoidance • 90% of the wind literature farm as a whole rate • 0% Species Avoidance rate From Common Eiders WFAR 82% Desholm and Kahlert 2005

THE MODEL: The wind farm as a risk window Distance to coast Columns F(y) Random Rows number Wind farm avoidance rate

THE MODEL: The wind farm as a risk window Survival Distance to coast Columns F(y) Rows Survival Wind farm avoidance rate

THE MODEL: The wind farm as a risk window Distance to coast Columns F(y) Rows

THE MODEL: The wind farm as a risk window Distance to coast Columns F(y) Rows

THE MODEL: The wind farm as a risk window Distance to coast Columns F(y) Left Rows Ahead Right TAR Birds in front of a turbine: Turbine Avoidance Rate (TAR)

THE MODEL: The wind farm as a risk window Species Avoidance rate From Common Eiders 94.6% Desholm and Kahlert 2005 Species Waterfowl and waders 97.5% Winkelman 1992, 1994 Distance to coast Columns Gulls, waders 97% Winkelman 1985 max. WFAR F(y) Bewick’s Swan 99.5% Percival 2004 min. Rows Gulls 99.9% Everaert et al. 2002 WFAR Common terns 99.8% Everaert et al. 2002 Barnacle, Greylag, White-fronted Geese 100% Percival 1998 max. WFAR TAR • 0% • 95% Birds in front of a • 96% Based on • 97% literature turbine: Turbine • 98% Avoidance Rate (TAR) • 99%

THE MODEL: The wind farm as a risk window Distance to coast Columns F(y) Random min. Rows number WFAR TAR

THE MODEL: The wind farm as a risk window Distance to coast Columns F(y) Rows TAR

THE MODEL: The wind farm as a risk window Distance to coast Following Band et al. 2007: Columns F(y) • Different species Rows • Different wind conditions • Headwind low (0 to 5 in Beaufort scale) • Headwind high (6 to 9 in Beaufort scale) • Tailwind low (0 to 5 in Beaufort scale) • Tailwind high (6 to 9 in Beaufort scale) TAR Birds within reach of Probability of safely passing the rotor blades turbine: Probability by by chance chance (Turker 1996, Band et al. 2007)

THE MODEL: The wind farm as a risk window Distance to coast Columns F(y) Rows Random number TAR Birds within reach of Probability of safely passing the rotor blades turbine: Probability by by chance chance (Turker 1996, Band et al. 2007)

THE MODEL: The wind farm as a risk window Distance to coast Columns F(y) Survival Rows TAR Probability of safely passing the rotor blades by chance

THE MODEL: The wind farm as a risk window Distance to coast Columns F(y) Rows Collision TAR Probability of safely passing the rotor blades by chance

THE MODEL: The wind farm as a risk window Distance to coast Columns F(y) PERCENTAGE Rows OF Survival MORTALITY TAR Probability of safely passing the rotor blades by chance

Objectives To develop a stochastic model of avian collision risk at wind farms A case study To obtain Factors probabilities of collision risk To estimate mortality rates

Factors 27,216 scenarios (also WFAR, TAR = 0) 1,000,000 events per scenario 13,608 scenarios: WFAR, TAR ≠ 0 2000 1800 94% 0 to 8 out of 10,000 birds 1600 Number of cases 1400 1200 1000 800 600 400 200 0 0.00 0.05 0.10 0.15 0.20 0.25 Percentage of estimated collisions

Objectives To develop a stochastic model of avian collision risk at wind farm A case study To obtain Factors probabilities of collision risk To estimate mortality rates

Factors To assess the weighted importance of the different input variables in collision predictions Generalized Additive Model

Factors 0.025 • WFAR: 20% Percentage of estimated collisions 0.020 • TAR: 23.5% • Probability by chance : 20.8% 0.015 0.010 • Spatial distribution of the birds 0.005 entering passage: 18.4% 0.000 5 10 15 20 25 30 35 40 45 90 80 70 60 Percentage of birds entering the wind farm (step 0) • Wind farm dimensions: 5.9% WFAR (%) ) 0.10 Percentage of estimated collisions 0.08 0.06 0.07 It’s necessary to consider 0.04 Avoidance rates are the most 900 m between columns 600 m between columns 0.06 Percentage of estimated collisions 0.02 300 m between columns the specific bird passage 0.00 important factors assessing the 0.05 95 96 97 98 99 TAR (%) 0.04 input spatial distribution risk of bird collision ) 0.12 (thus Percentage of estimated collisions 0.03 Balearic Shearwater 0.10 Cory's Shearwater Gannet confirming Desholm and Kahlert 2005, 0.02 0.08 0.06 Chamberlain et al. 2006) 0.01 0.04 0.00 0.02 1 2 3 6 10 14 0.00 Number of columns 1 2 3 4 Probability of passing safely the rotor blades by chance

Objectives To develop a stochastic model of avian collision risk at wind farm A case study To obtain Factors probabilities of collision risk Bird volume Number of Flight altitude To estimate birds mortality rates collided per time period

Estimating the mortality rates: FLIGHT ALTITUDE BIRD VOLUME FLIGHT ALTITUDE Autumn migration Layer 3 volume in the north side of the Strait of Gibraltar Layer 2 Risk of collision Following Krüger and Garthe 2001, Layer 1 We obtained the proportion of birds flying in each height layer for: • Different species • Different wind conditions

Estimating the mortality rates Estimated number of collided birds per autumn season 30 4 5 Non-evasive + TAR + WFAR + Flight Altitude scenario 1,340 ± 433 46 ± 15 11.6 ± 3.7 2.3 ± 0.8 Percentage of Cory’s Shearwater flying at 306 ± 73 11 ± 3 2.6 ± 0.6 0.2 ± 0.1 Layer 1 Layer 2 Layer 3 n E1 99.1% 0.6% 0.2% 2,160 E2 99.4% 0.6% 0.0% 36 203 ± 43 7 ± 2 1.8 ± 0.4 0.6 ± 0.1 W1 94.6% 5.3% 0.1% 3,262 W2 100.0% 0.0% 0.0% 1,195

Estimating the mortality rates Estimated number of collided birds per autumn season 13 Non-evasive + TAR + WFAR + Flight Altitude scenario 1,340 ± 433 46 ± 15 11.6 ± 3.7 2.3 ± 0.8 306 ± 73 11 ± 3 2.6 ± 0.6 0.2 ± 0.1 Percentage of Balearic Shearwater flying at 203 ± 43 7 ± 2 1.8 ± 0.4 0.6 ± 0.1 Layer 1 Layer 2 Layer 3 n E1 99.7% 0.3% 0.0% 1,518 E2 100.0% 0.0% 0.0% 25 W1 97.8% 2.1% 0.1% 849 W2 100% 0% 0% 20

Estimating the mortality rates Estimated number of collided birds per autumn season 3 Non-evasive + TAR + WFAR + Flight Altitude scenario Percentage of Northern Gannet flying at 1,340 ± 433 46 ± 15 11.6 ± 3.7 2.3 ± 0.8 Layer 1 Layer 2 Layer 3 n E1 90.5% 8.4% 1.1% 577 306 ± 73 11 ± 3 2.6 ± 0.6 0.2 ± 0.1 E2 97.2% 2.8% 0.0% 156 W1 76.4% 20.6% 3.0% 718 W2 91.2% 8.3% 0.6% 223 203 ± 43 7 ± 2 1.8 ± 0.4 0.6 ± 0.1

Conclusions: THE CASE STUDY Avoidance rates are the most important factors assessing the risk of bird collision Altitudes of migration  strongly influence the probability of collision These parameters should be considered as priorities to be addressed in post-construction studies Fatalities seems to be low  To consider the synergistic effect of installing different wind farms along the same migratory route Other hazards exist to birds by the construction of off- shore wind farms, in addition to collision risk