TRAPPER’S POINT WILDLIFE DETECTION TRAPPER’S POINT WILDLIFE DETECTION SYSTEM: TECHNOLOGICAL AND SYSTEM: TECHNOLOGICAL AND SYSTEM: TECHNOLOGICAL AND SYSTEM: TECHNOLOGICAL AND
INSTITUTIONAL CHALLENGESINSTITUTIONAL CHALLENGES
Rhonda Young, University of Wyoming
Trappers PointTrappers Pointpppp
On 6000 year migration path from summer grounds On 6000 year migration path from summer grounds in Grand Teton NP to winter grounds in Red Desert
12’ Lanes, 4’ Shoulders
Speed Limit = 65 mph
Trapper’s PointTrapper’s Pointpppp
29 reported accidents between 1995 and 2005 29 reported accidents between 1995 and 2005 23 (79%) involved collisions with animals 1 antelope 1 antelope 1 cow 21 deer
~ 2.3 animal-vehicle collisions per year
93 carcasses were collected between 1999 and 2005 92% mule deer ~ 15.8 animal-vehicle collisions per year
Trapper’s PointTrapper’s Pointpppp
Traffic volumes increasingg
AADT by Year
2000
2200
2400
affic
(vpd
)
AADT by Year
1600
1800
2000
Ann
ual D
aily
Tra
1000
1200
1400
992 994 996 998 2000 2002 2004 2006
Ave
rage
A
1992 1994 1996 1998 2000 2002 2004 2006
Year
Trapper’s PointTrapper’s Pointpppp
Pronghorn, mule deer, elk, and moose inhabit the area.g , , , Mule deer population between 22,000 and 36,000.
Estimated Population of Wildlife by species and year
50,000
60,000
Estimated Population of Wildlife, by species and year
30,000
40,000
mat
ed P
opul
atio
n
Pronghorn
Mule Deer
Elk
0
10,000
20,000Estim Elk
Moose
01990 1992 1994 1996 1998 2000 2002 2004 2006
Year
Trapper’s PointTrapper’s Pointpppp
T ’ P i t C liti Trapper’s Point Coalition More than 30 membersGas Industry Officials Ranchers WYDOT WY Game & Fish Gas Industry Officials, Ranchers, WYDOT, WY Game & Fish,
Governor’s Office,
Decided to construct a wildlife-sensing driver warning g gsystem to reduce deer-vehicle collisions. Used technology similar to system at Nugget Canyon Installed in October, 2005. Cost approximately $1,000,000.
TechnologyTechnologygygy
Wildlife detection systemy EIDS (Electronic Intrusion Detection System) Developed by Telonics, Inc for the military System worked at Nugget Canyon, WY
Geophones (seismic) and Passive Infrared sensors
TechnologyTechnology
Active Driver Warning System
gygy
g y 3 Signs in Each Direction Originally set to flash for 2 minutes for each detection
but later reduced to 30 seconds
Data CollectionData Collection
Detection Data Requires detection on two sensors (geophone and
infrared) to qualify
Relays data to storage deviceP i Printer
Laptop
False DetectionsFalse Detections
Passing tractor trailers were causing false positive g g pdetections At the time, two infrared sensors and two geophone
sensors were attached to each processor / transmitter. A detection could be caused when two infrared sensors
h i dor two geophone sensors were tripped. Unforeseen software loophole
Major change in layout Processors were consolidated so that only one infrared Processors were consolidated so that only one infrared
sensor and one geophone were attached to each.
Old System LayoutOld System Layouty yy y
33312927
25
23FE
19
17
21
26 28 30 32 34N
1
3
17
15
13
11 20
22
24
D
3
5 7
11
9
18
20
14
16A
B
C
2
4
6 8 10
12
Current System LayoutCurrent System Layouty yy y
E
23
N
33
19
21
22
24
29
31
17
15
18
20
22
25
27
29
32
34
D
17
13
11
9
18
14
163
5
28
30
C
810
12
14
2 4
6
26B
Source: Kevin Cox, WYDOT
False DetectionsFalse Detections
Infrared Filters Added to limit false detections during sunrise, sunset.
Breakdown of Detections by Transmitter 1 by half-hour periods, April 9th, 2007
12
14
er h
alf
hour
2007
6
8
10
ber o
f D
etec
tions
p
0
2
4Num
12:00 AM 6:00 AM 12:00 PM 6:00 PM 12:00 AM
Time of Day
False DetectionsFalse Detections
Qualify Times have varied between 2 and 10 sec. Qualify Times have varied between 2 and 10 sec. WYDOT, Telonics, and UW have adjusted many times Telonics recommended a qualify time of 4 to 5 seconds. Telonics recommended a qualify time of 4 to 5 seconds. UW set all of the sensors to the following:Qualify Time = 5 seconds IR Sensitivity = 4Geophone Sensitivity = 5
Problematic sensors were identified.Qualify time lowered to 2 secondsGeophone and infrared sensitivity lowered to 3Geophone and infrared sensitivity lowered to 3
Handheld Radio IssuesHandheld Radio Issues
Handheld radio was Handheld radio was not receiving signals from most of the sensors
Metal cabinet wascausing poor reception.
Added an external Added an external antenna
Data CollectionData Collection
Effectiveness of Wildlife Detection System Effectiveness of Wildlife Detection System Camera, DVR, Detection Data False Negativeg System fails to detect wildlife when wildlife are present Random sampling of DVR data Sit i it t d k d i i ti i d Site visits at dusk during migration periods
False Positive System detects something other than wildlifey g Review of DVR data
Effectiveness of Driver Warning System Speed Sensors, Detection Data Impact of Flashing Signs on Observed Speeds
Effect on Driver BehaviorEffect on Driver Behavior
Avg speed and total volume for each half hour Avg speed and total volume for each half hour period was calculated in three locations
Number of detections per half hour was calculated Detections that occurred within 30 seconds of last Detections that occurred within 30 seconds of last
detection were eliminated Represents the number of times that flashing lights were p g g
activated during each half hour
Effect on Driver BehaviorEffect on Driver Behavior
If the system has an effect on driver behavior: Higher numbers of detections = lower average speed Lower numbers of detections = higher average speed
Linear regression analysis Response variable = average speed per half hour at a Response variable average speed per half hour at a
particular sign Predictor variables:
N b f d i h lf h Number of detections per half hour. Traffic volume per half hour. Lighting condition (Day = 1, Night = 0)
The number of detections per half hour was not significant in any of the models. (α=0.05)
Effect on Driver BehaviorEffect on Driver Behavior
Volume per half hour and detections per half hour p pare not linearly independent. The wildlife detection system is detecting traffic. R2 = 0.629, Pr>F = <0.0001
Detections vs. Volume4/26/07 - 5/16/07
R² = 0.62980
100
120
half
hour
4/26/07 - 5/16/07
ALL
ZONE 1
R² = 0.6131
R² = 0.6217
20
40
60
Det
ectio
ns p
er h ZONE 25
Linear (ALL)
Linear (ZONE 1)
Linear (ZONE 25)
00 20 40 60 80 100
Traffic Volume per half hour
Effectiveness of Detection SystemEffectiveness of Detection Systemyy
Detections that occurred outside the camera’s field of view were not considered.
5 hour-long periods that corresponded to handheld radio data were randomly selected and viewed.y Meant to provide data on false negatives. No wildlife was seen in any of the periods.
Effectiveness of Detection SystemEffectiveness of Detection Systemyy
Only 4 of the 14 sensors detected wildlife; y ;remaining detections were all false positives.
DETECTION ZONE
# OF TRUE DETECTIONS # OF FALSE POSITIVE DETECTIONS
% FALSE POSITIVE DETECTIONS
1 0 5 100%2 0 5 100%5 0 4 100%6 1 1 50%6 1 1 50%7 0 4 100%9 0 1 100%11 0 5 100%13 0 3 100%16 1 0 0%18 1 0 0%26 0 4 100%26 0 4 100%27 2 0 0%28 0 1 100%30 0 1 100%
Effect on CrashesEffect on Crashes
Total # of Reported Crashes Total # of Reported Crashes
Effect on CrashesEffect on Crashes
Total # of Animal-Vehicle Crashes Total # of Animal Vehicle Crashes
Effect on CrashesEffect on Crashes
Animal-Vehicle Crashes Rate Animal Vehicle Crashes Rate
ConclusionsConclusions
While visual trends indicate a reduction in animal- While visual trends indicate a reduction in animalvehicle crashes the difference in crashes (both frequency and crash rates) was not statistically q y ) ysignificant
ConclusionsConclusions
Number of detections per half hour does not have a Number of detections per half hour does not have a significant effect on the average speed Drivers do not respect signs and may have become Drivers do not respect signs and may have become
accustomed to them. Likely Cause: too many false detections
The majority of detections by the wildlife detection system are false detections Number of detections related to traffic volumes
ConclusionsConclusions
System Maintenance was high due to the large System Maintenance was high due to the large number of sensors.
Snow cover appears to make geophones insensitive Snow cover appears to make geophones insensitive and cause wiring breaks
Infrared sensors appear to be sensitive to sun and Infrared sensors appear to be sensitive to sun and clouds (east-west orientation problematic)
Lack of communication and remote location make Lack of communication and remote location make detecting system errors difficult
WYDOT personnel with knowledge and resources to WYDOT personnel with knowledge and resources to maintain system not locally available
Potential Future ActionsPotential Future Actions
Need for animal-vehicle countermeasure at location Nverified. Six potential future actions:1. Keep system in its current state and operate year
d h d d d ffround with dedicated maintenance staff2. Keep system in its current state and operate during
migration periodsg p3. Reduce scale of system and use fencing to funnel animals
to detection zonesR h d l f d f 4. Remove geophones and use only infrared sensors for detection
5. Use other animal detection technology such as “break the gybeam” systems
6. Use a wildlife under or overpass
Things to considerThings to considergg
Who will maintain the technology and how close Who will maintain the technology and how close are they to the site?
How will you know when there’s a problem with the How will you know when there s a problem with the system?
Can the technology be minimized while keeping the Can the technology be minimized while keeping the function of the system?
Can the technology be located more ideally? (snow Can the technology be located more ideally? (snow cover and drifting, east/west direction, on curves, separation from roadway)
Final ReportFinal Reportpp
Evaluation of an Active Wildlife-Sensing and Evaluation of an Active Wildlife Sensing and Driver-Warning System at Trapper's Point -FHWA-WY-09/03F /
Available on WYDOT Web Page http://www.dot.state.wy.us/wydot/planning_projp // y / y /p p jects/studies_plans/research_center_reports_3
QUESTIONSQUESTIONS
Rhonda Young
Associate Professor, Dept. of Civil & Arch. Engineering, Univ. of Wyoming, p g g, y g