Download - Schnick capstone presentation
Brian Schnick
MS-GISc Candidate
Denver University
A GIS Model to Predict Feral Pig (Sus scrofa)
Habitat on Vandenberg Air Force Base,
California
Pigs cause damage; need to predict habitat
Pigs are difficult to observeNocturnalBed in dense brush/coverVAFB encompasses ~400 sq km
GIS modeling proven for raptors, mammals, pests
Why a GIS model?
30 CESGeobase ManagerRange ManagerArcheologistsFacility Managers
30 SFSGame Conservation Officers
30 FSSGolf Course Management
Proposed Audience
Pigs inhabit 56/58 CA counties
Rooting!Disturbs ground 6-18” deepLandscapingArcheology sitesEndangered species
No current catalog of pig incidents/locations
Needs Assessment
Predictive modelPreferred environmental factors
Provide data to:CES, SFS, FSS
Animal Control/Damage PreventionBarriers?Prioritized site study?Population control?
Goal of Application
ESRI GIS software in-place
ArcView/ArcInfo licensesSpatial Analyst
Raster CalculatorExtract Values to PointsDistance Straight LineCommand Line: ReclassByASCIIFile_sa
ArcIMS for distribution?
Software Requirements
Personal Desktop
SensorsTrimble GeoXH handheld GPS unit Leaf River digital game cameraRaven UAVNight vision devicesPersonal spotlight
Hardware Requirements
Secondary Data: National Map Seamless Server
Primary Data: TransectsGame CamUAV
Datasets and Sources
BTS Roads NED 1/9 arc sec NAIP (3 band) UTM Zone 10
NLCD 2001 Land Cover (30m)National Atlas Vegetation Growth – Peak:2005
National Atlas Vegetation Growth – Average:2005
National Atlas Roads National Atlas Streams National Atlas Land Cover Characteristics (AVHRR 1km)
Study AreaMulti-Stage Area Sampling
3 per area
NS / EW distribution
SW corner origin
www.random.org
Transect Selection
1
5
9
1 5 9
2 areas closed for EOD operations
3 contained admin/brush/terrain problems
Rejected without replacementUAV overcomes!
Barriers
Transects and GPS3 x 2 day sessions54 presence, 47 absence points
collectedLots of ticks and poison oak
Game Camera2 x week sessions120 photos of bushes, 1 photo of a
deer collected = 2 absence points
Data Collection
Raven UAV1 session of 2 hours @ duskIR sensor (low resolution)14 presence points collected
Data Collection, continued
UAV IR Video Clip
YouTube
Food/water/shelter = Water/Vegetation/Roads
Factor weights
Application Development
Environmental factor Weight
Distance to water 60%
Vegetation 30%
Distance to roads 10%
((([Dist2SmWaRas2] * -0.000114) + 1) * 6) + (([Dist2RoadRas] * 33.33) * 4)+ ([nlcd_reclass] * 0.33)
Vegetation weights
Application Development, cont.
Veg type Weight
Shrubland, herbaceous grassland, woody wetland, herbaceous wetland emergent
3
Recreational grasses urban, low intensity residential, evergreen forest, mixed forestland
2
Open water, high intensity residential, commercial-industrial-roads, transitional barren
1
Dis
trib
uti
on M
odel
The Data
Dist to Roads (m)
Dist to Water (m)
0.00
200.00
400.00
600.00
800.00
1000.00
1200.00
PresenceAbsence
Low in
tens
ity re
siden
tial
Everg
reen
fore
st
Urban
, rec
reat
iona
l gra
sses
Emer
gent
, her
bace
ous w
etla
nd
Shrub
land
Grass
es, h
erba
ceou
s0.00
0.10
0.20
0.30
0.40
0.50
0.60
PresenceAbsence
Logistic regression: R2=.04
Process Validation
Logistic regression: R2=.31
Simulated data
Environment factors correlate to presence
Prediction correlates to presence: R2=.22
Moderate success creating GIS model for feral pig distribution!
Primary issue: data collection
Future research/model refinement
Disbursed data collectionRadio collars!Seasonal patterns
Discussion
Tim BeltonDarryl York, Paul Vincent, Ken LucasChris Ryan, Robert PetersonEd Panas, Wayne MosesTSgt Vail, SSgt Johnson, SSgt Fay
Acknowledgements