Extending GIS with Statistical Models to Predict Marine Species Distributions
description
Transcript of Extending GIS with Statistical Models to Predict Marine Species Distributions
![Page 1: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/1.jpg)
Extending GIS with Statistical Modelsto Predict Marine Species Distributions
Zach Hecht-LeavittNY Department of State
Division of Coastal Resources
![Page 2: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/2.jpg)
Offshore Planning
NJ
NY
Long Island
![Page 3: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/3.jpg)
Goals
• Predict the abundance of selected groundfish species as a function of:–environmental/habitat variables– spatial autocorrelation
• Assess error
• Understand ecology
![Page 4: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/4.jpg)
Goals
![Page 5: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/5.jpg)
• NOAA Northeast Fisheries Science Center bottom trawl (catches groundfish)
• Biannual 1975-2009• Standardized gear,
speed, and distance• Cleaned by Stone
Environmental• Break down by season
and life stage• 6 species selected
The Fish Data
NOAA NEFSC NOAA NEFSC
![Page 6: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/6.jpg)
The Environmental Data
• Depth• Distance from Shelf Edge• Bottom Grain Size• Slope• Sea Surface Temperature*• Chlorophyll*• Stratification*• Turbidity*• Zooplankton*• Provided by NOAA
Biogeography Branch*Long-term, seasonal average
![Page 7: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/7.jpg)
Data Exploration
• Approximate environmental relationships with linear trend• Data is extremely skewed (lots of zeroes)• Go with Zero-Inflated Generalized Linear Models (GLMs)
Adul
t Flu
ke C
ount
![Page 8: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/8.jpg)
Workflow
.CSV
Y=B1X1+B2X2…
• “Loose” coupling• With 10.x can “hard” couple via Geospatial
Modeling Environment (GME)
zeroinfl(), Achim Zeileis
![Page 9: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/9.jpg)
B1 * B3 *B2 * =++
Workflow
![Page 10: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/10.jpg)
The Residual
+4
+2
-1
+1
+2+3
+5
-4
-3
-9
![Page 11: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/11.jpg)
Kriging Interpolation
![Page 12: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/12.jpg)
Modeling Steps
=+
+ Error (unexplained variation)
Predictors (GLM) + Residual (kriging) = Mean Expectation
![Page 13: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/13.jpg)
Error Assessment
+2
-1 +2
+3
-4
-3
![Page 14: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/14.jpg)
Error Assessment
• 50/50 cross-validation• Smooth error with moving window• Final maps based on full dataset• Conservative error estimates
![Page 15: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/15.jpg)
L
LL
M
H
Final product (fluke example)
![Page 16: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/16.jpg)
Goals
• Predict the abundance of selected groundfish as a function of:–environmental/habitat variables– spatial autocorrelation
• Assess error
• Understand ecology
![Page 17: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/17.jpg)
Another Approach to Dealing with Zeroes
![Page 18: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/18.jpg)
Another Approach
• Rather than a one-size-fits all model…
• … model presence/absence and abundance with two separate stages
• May better reflect ecological reality
• More conservative approach
![Page 19: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/19.jpg)
Goals
• Predict the abundance of selected cetaceans as a function of:– spatial autocorrelation
• Assess error
![Page 20: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/20.jpg)
The Marine Mammal Data
• North Atlantic Right Whale Consortium Database
• Aerial and shipboard surveys, 1978 - 2009
• Cleaned by New England Aquarium
• SPUE• 4 species/groups
selected• No predictors this time!
NOAA SWFSC
NOAA SWFSC
![Page 21: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/21.jpg)
Stage I – Presence/Absence
4
0
0
0
1
3
5
0
4
2 4
2
3
4 0
0
2
0
0
1
0
0
0
1
1
1
0
1
1 1
1
1
1 0
0
1
0
0
1
0
0
0
1
1
1
0
1
1
![Page 22: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/22.jpg)
Stage I – Presence/Absence
1
0
0
0
1 1
1
0
1
1
ROC CurveEpi package,Carstensen et. al.
![Page 23: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/23.jpg)
Stage I * II – Abundance Where Present
![Page 24: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/24.jpg)
Error Assessment
![Page 25: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/25.jpg)
Goals
• Predict the abundance of selected cetaceans as a function of:– spatial autocorrelation
• Assess error
![Page 26: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/26.jpg)
End goal
![Page 27: Extending GIS with Statistical Models to Predict Marine Species Distributions](https://reader035.fdocuments.in/reader035/viewer/2022062500/568154c6550346895dc2cc62/html5/thumbnails/27.jpg)
Further reading• Hengl, T., G.B.M. Heuvelink and D.G. Rossiter. 2007. About regression-
kriging: From equations to case studies. Computers and Geosciences 33:1301-1315.
• Wenger, S. J. and M. C. Freeman. 2008. Estimating Species Occurrence, Abundance, And Detection Probability Using Zero-Inflated Distributions. Ecology 89:2953-2959
• Monestiez, P., L. Dubroca, E. Bonnin, J.-P. Durbec, C. Guinet. 2006. Geostatistical modeling of spatial distribution of Balaenoptera physalus in the Northwestern Mediterranean Sea from sparse count data and heterogeneous observation efforts. Ecological Modeling, 193: 615-628.
• Menza, C., B.P. Kinlan, D.S. Dorfman, M. Poti and C. Caldow (eds.). 2012. A Biogeographic Assessment of Seabirds, Deep Sea Corals and Ocean Habitats of the New York Bight: Science to Support Offshore Spatial Planning. NOAA Technical Memorandum NOS NCCOS 141. Silver Spring, MD. 224 pp.