integrating ocean observing data to enhance protected species spatial decision support systems
DESCRIPTION
Integrating Ocean Observing Data to Enhance Protected Species Spatial Decision Support Systems. Biodiversity and Ecological Forecasting Team Meeting May 17-19, 2010. Dr. Karin Forney. Dr. Pat Halpin. Presented by: Ben Best (Duke) Elizabeth Becker (NOAA). Project Team Members. - PowerPoint PPT PresentationTRANSCRIPT
Integrating Ocean Observing Data to Enhance Protected Species
Spatial Decision Support Systems
Biodiversity and Ecological Forecasting Team Meeting
May 17-19, 2010
Dr. Pat HalpinDr. Karin Forney
Presented by:
Ben Best (Duke)
Elizabeth Becker (NOAA)
Project Team MembersBen Best, Ei Fujioka, Pat Halpin, and Jason Roberts
Marine Geospatial Ecology LabNicholas School of the Environment, Duke University
Lisa Ballance, Jay Barlow, Elizabeth Becker, Steven Bograd, Karin Forney, and Jessica RedfernSouthwest Fisheries Science Center
NOAA - National Marine Fisheries Service
Dave Foley and Daniel PalaciosJoint Institute for Marine and Atmospheric Research,
University of Hawai`i at Manoa
Grant/Cooperative Agreement Number: NNX08AK73G
Cetaceans (whales, dolphins and porpoises)and anthropogenic threats
Threats include
Ship strikes
Fishery bycatch
Naval activities
Anthropogenic sound
Cetaceans protected by US laws
MMPA
ESA
Cetacean distributions are dynamic
Balaenoptera musculus
Blue whale
DataData
DecisionDecision
Marine Habitat Modeling Process
ModelingSDSS
SSTCHLDepthShelfFronts
Sample fin whale densitiesKey predictor variables
Depth, Slope, SSTBal.phy
R S M odel
R efit to 1991-2005
C reated: 08/28/08 15:52:29
30o
35o
40o
45o
1991 1993 1996
W 130o
W 125o
W 120o
30o
35o
40o
45o
2001
W 130o
W 125o
W 120o
2005
W 130o
W 125o
W 120o
0.00010
0.00150
0.00250
0.00350
0.00450
0.00550
0.00650
0.00750
D ensity (Ani/km 2)
Avg91-05
O bs. Seg. D ensity
< 1 < 5 < 13
1991 1993 1996
2001 2005 Ave.
Expansion and Enhancement of the SDSS
Challenge: Marine mammals are highly mobile; distributions change on seasonal, interannual and decadal time scales
La NiñaEl Niño
Expansion and Enhancement of the SDSS
Incorporate additional covariates derived from remotely-sensed data (varies by region)
• Explore the implementation of Nowcast and Forecast capabilities
• Update SDSS and release a package of open-source Desktop GIS tools for end-users
Whale, dolphin and porpoise species:
Pacific: 21 species and 1 guild of beaked whales
Atlantic: 12 species and five species guilds
Nowcast and Forecast Capability development
GHRSST (RSS Inc.): Blended SST Developed by Remote Sensing Systems, Santa Rosa, CA High-resolution (9 km) infrared data Microwave (data for cloudy areas) Optimal interpolation Pixel-by-pixel error characterization
ROMS = Regional Ocean Modeling SystemDeveloped for the NASA-funded FAST Project (Chavez,
Chai, Chao, Barber and Foley) Run by Yi Chao's group at JPL Uses forecast surface fluxes (NCEP) Monthly mean products with 1-9 month lead time
NOWCASTS
(tactical)
FORECASTS
(strategic)
NOWCAST – Dall’s porpoise densityfor novel 2008 survey (July-Nov)
“Daily forecast”“1991-2005 Climatology”
NOWCASTS as short-term forecasts
(weeks)
FORECAST – Striped dolphin densityROMS: Oct/Nov 2008 (as forecast in July)
Expansion and Enhancement of the SDSS
Incorporate additional covariates derived from remotely-sensed data (varies by region)
• Explore the implementation of Nowcast and Forecast capabilities
• Update SDSS and release a package of open-source Desktop GIS tools for end-users
Whale, dolphin and porpoise species:
Pacific: 21 species and 1 guild of beaked whales
Atlantic: 12 species and five species guilds
SDSS: Summarize by Region
Select Region by1. Drop-down
list (OPAREAs, MPAs, EEZ)
2. Enter polygon coordinates
3. Draw on mapReturn:
– Effort, Obs– Min, Max,
Mean– Histogram– Coordinates
ROC Curve to Binary Habitat in SDSS
Example: baleen guild (fin, blue, sei, Bryde’s) in summer
Example: Species distribution modeling with MGET(Marine Geospatial Ecology Tools)
Sample time-series imagery Invoke R from ArcGIS to create plots, etc.
Fit models with R, evaluate using ROC analysis, predict maps from satellite images
False positive rate
Tru
e po
sitiv
e ra
te
Cutoff = 0.020
Binary classification (range map)
Predicted probability of presence
Red: Anticyclonic Blue: Cyclonic
Eddies from AVISO
Additional Derived Covariates
28.0 °C
25.8 °C
Front
Fre
quen
cy
Temperature
Optimal break 27.0 °C
Fronts from Pathfinder / GHRSSTCayula, J-F and P Cornillon (1992)Isern-Fontanet et al (2006)
Hybrid Coordinate Ocean Model (HYCOM)
pros: 1/12 °, cloud-free, 3D, forecastcons: modeled, since 2003, physical only
Hybrid Coordinate Ocean Model (HYCOM)
cons: complicated projection
Minimal Loss Decision Mapping
Loss Decision
Integrated Prob(Encounter)
0-.5 .5 to 1
DecisionGo 0 3
No-go
1 0
Conclusions and Next Steps
• Nowcast and forecast capabilities exceeded expectations
• Incorporate additional ecological relevant covariates (e.g. ROMS-CoSiNE)
• Foundation for climate-response modeling
• Improved SDSS for end-user needs
• Reproducibility with desktop GIS
Thank You!
Funding
• NASA
• SERDP
• NOAA
Project Support
• Marine mammal observers, oceanographers, chief scientists, cruise leaders, officers and crew of surveys
• Yi Chao (JPL)
• Fei Chai (University of Maine)
• NOAA Northeast and Southeast Fisheries Science Centers
Websites
• SDSS – http://seamap.env.duke.edu
• MGET – http://code.env.duke.edu/projects/mget