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

To develop a stochastic model of avian collision risk at wind farms

To estimate mortality rates To obtain probabilities of collision risk Factors

A case study

Objectives

To develop a stochastic model of avian collision risk at wind farms

Objectives

A case study

To obtain probabilities of collision risk Factors To estimate mortality rates

Stochastic character,

THE MODEL

based on Montecarlo simulation

CORY’S SHEARWATER Calonectris diomedea

Case study: The Strait of Gibraltar

NORTHEN GANNET Morus bassanus

BALEARIC SHEARWATER

Puffinus mauretanicus

Case study: The Strait of Gibraltar

Antenna 12 ft TX 3.050 MHz 286 dB 22-28 rpm 30 KW

TARIFA ISLAND Gibraltar Algeciras Ceuta

Strait of Gibraltar

VISUAL & RADAR CENSUSES

THE MODEL: The wind farm as a risk window

RISK WINDOW

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

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

Wind farm dimensions

THE MODEL: The wind farm as a risk window

F(y)

Rows Distance to coast Columns

• Different species
• 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)
• Uniform distribution

Flight direction Horizontal distribution of the migratory passage

THE MODEL: The wind farm as a risk window

F(y)

Montecarlo simulation

Columns Rows Distance to coast

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

F(y)

• 60%
• 70%
• 80%
• 90%
• 0%

Species Avoidance rate From Common Eiders WFAR 82% Desholm and Kahlert 2005

Avoidance rate

• f the wind

farm as a whole

Wind farm avoidance rate

Based on literature

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

F(y)

Wind farm avoidance rate Random number

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

F(y)

Wind farm avoidance rate Survival Survival

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

F(y)

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

F(y)

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

F(y)

TAR

Birds in front of a turbine: Turbine Avoidance Rate (TAR)

Left Ahead Right

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

Species

F(y)

• max. WFAR
• max. WFAR

TAR min. WFAR

• 0%
• 95%
• 96%
• 97%
• 98%
• 99%

Based on literature Species Avoidance rate From Common Eiders 94.6% Desholm and Kahlert 2005 Waterfowl and waders 97.5% Winkelman 1992, 1994 Gulls, waders 97% Winkelman 1985 Bewick’s Swan 99.5% Percival 2004 Gulls 99.9% Everaert et al. 2002 Common terns 99.8% Everaert et al. 2002 Barnacle, Greylag, White-fronted Geese 100% Percival 1998

Birds in front of a turbine: Turbine Avoidance Rate (TAR)

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

F(y)

TAR min. WFAR

Random number

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

F(y)

TAR

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

F(y)

TAR

Probability of safely passing the rotor blades by chance

Following Band et al. 2007:

• Different species
• 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)

Birds within reach of turbine: Probability by chance (Turker 1996, Band et al. 2007)

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

F(y)

TAR

Random number

Probability of safely passing the rotor blades by chance

Birds within reach of turbine: Probability by chance (Turker 1996, Band et al. 2007)

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

F(y)

TAR

Probability of safely passing the rotor blades by chance

Survival

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

F(y)

TAR

Collision

Probability of safely passing the rotor blades by chance

THE MODEL: The wind farm as a risk window

Columns Rows Distance to coast

F(y)

TAR

Probability of safely passing the rotor blades by chance

Survival

PERCENTAGE OF MORTALITY

To develop a stochastic model of avian collision risk at wind farms

To obtain probabilities of collision risk

A case study

Objectives

Factors To estimate mortality rates

Percentage of estimated collisions 0.00 0.05 0.10 0.15 0.20 0.25 Number of cases 200 400 600 800 1000 1200 1400 1600 1800 2000

13,608 scenarios: WFAR, TAR ≠ 0 94% 0 to 8 out of 10,000 birds Factors

27,216 scenarios (also WFAR, TAR = 0) 1,000,000 events per scenario

To develop a stochastic model of avian collision risk at wind farm

To obtain probabilities of collision risk Factors

A case study

Objectives

To estimate mortality rates

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

Factors

Probability of passing safely the rotor blades by chance 1 2 3 4 Percentage of estimated collisions 0.00 0.02 0.04 0.06 0.08 0.10 0.12

Balearic Shearwater Cory's Shearwater Gannet

TAR (%) 95 96 97 98 99 Percentage of estimated collisions 0.00 0.02 0.04 0.06 0.08 0.10

) )

Number of columns

1 2 3

Percentage of estimated collisions

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 900 m between columns 600 m between columns 300 m between columns 6 10 14

• WFAR: 20%
• TAR: 23.5%
• Probability by chance : 20.8%
• Spatial distribution of the birds

entering passage: 18.4%

• Wind farm dimensions: 5.9%

Factors

Avoidance rates are the most important factors assessing the risk of bird collision

(thus confirming Desholm and Kahlert 2005, Chamberlain et al. 2006)

It’s necessary to consider the specific bird passage input spatial distribution

Percentage of birds entering the wind farm (step 0) 5 10 15 20 25 30 35 40 45 Percentage of estimated collisions 0.000 0.005 0.010 0.015 0.020 0.025

90 80 70 60 WFAR (%)

To develop a stochastic model of avian collision risk at wind farm

To obtain probabilities of collision risk Factors

A case study

Objectives Number of birds collided per time period Bird volume Flight altitude

To estimate mortality rates

Estimating the mortality rates: FLIGHT ALTITUDE

FLIGHT ALTITUDE

Following Krüger and Garthe 2001, We obtained the proportion of birds flying in each height layer for:

• Different species
• Different wind conditions

Risk of collision

Layer 3 Layer 2 Layer 1

Autumn migration volume in the north side of the Strait of Gibraltar

BIRD VOLUME

Non-evasive scenario + TAR + WFAR + Flight Altitude 1,340 ± 433 46 ± 15 11.6 ± 3.7 2.3 ± 0.8 306 ± 73 11 ± 3 2.6 ± 0.6 0.2 ± 0.1 203 ± 43 7 ± 2 1.8 ± 0.4 0.6 ± 0.1

Percentage of Cory’s Shearwater flying at Layer 1 Layer 2 Layer 3 n E1 99.1% 0.6% 0.2% 2,160 E2 99.4% 0.6% 0.0% 36 W1 94.6% 5.3% 0.1% 3,262 W2 100.0% 0.0% 0.0% 1,195

30 4 5 Estimated number of collided birds per autumn season Estimating the mortality rates

Non-evasive scenario + TAR + WFAR + Flight Altitude 1,340 ± 433 46 ± 15 11.6 ± 3.7 2.3 ± 0.8 306 ± 73 11 ± 3 2.6 ± 0.6 0.2 ± 0.1 203 ± 43 7 ± 2 1.8 ± 0.4 0.6 ± 0.1

Percentage of Balearic Shearwater flying at 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 13 Estimated number of collided birds per autumn season

Non-evasive scenario + TAR + WFAR + Flight Altitude 1,340 ± 433 46 ± 15 11.6 ± 3.7 2.3 ± 0.8 306 ± 73 11 ± 3 2.6 ± 0.6 0.2 ± 0.1 203 ± 43 7 ± 2 1.8 ± 0.4 0.6 ± 0.1

Percentage of Northern Gannet flying at Layer 1 Layer 2 Layer 3 n E1 90.5% 8.4% 1.1% 577 E2 97.2% 2.8% 0.0% 156 W1 76.4% 20.6% 3.0% 718 W2 91.2% 8.3% 0.6% 223

3 Estimating the mortality rates Estimated number of collided birds per autumn season

Avoidance rates are the most important factors assessing the risk of bird collision Altitudes

• f

migration  strongly influence the probability of collision 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 These parameters should be considered as priorities to be addressed in post-construction studies

Conclusions: THE CASE STUDY

A collision model considering the wind farm area as a risk window was constructed for avian migrants. Due to its very fast run velocity, it is possible to test a huge number of scenarios in a relatively short period of time. The possibility of testing so many cases, linked to its stochastic character and its high flexibility, give to the estimated probabilities of collision a high level of statistical confidence.

Conclusions: THE MODEL

THANKS VERY MUCH FOR YOUR ATTENTION