Presentation is loading. Please wait.

Presentation is loading. Please wait.

Pat Halpin Nicholas School of the Environment Duke University

Similar presentations


Presentation on theme: "Pat Halpin Nicholas School of the Environment Duke University"— Presentation transcript:

1 Integrating Ocean Observing Data to Enhance Protected Species Spatial Decision Support Systems
Pat Halpin Nicholas School of the Environment Duke University Karin Forney Protected Species Division Southwest Fisheries Science Center NOAA NASA Biodiversity Team Meeting New York, NY May 7, 2009 (presented by Jason Roberts, Dave Foley and Elizabeth Becker) First year of our NASA Ecological Forecasting research program. Cooperative team = the Marine Geospatial Ecology Lab at Duke University and the NOAA Southwest Fisheries Science Center (includes the Protected Resources Division in La Jolla, CA and the ERD in Pacific Grove, CA). Effort builds on the existing spatial decision support system (SDSS) developed under the Strategic Environmental Research and Development Program.

2 Project Team Members Ben Best, Ei Fujioka, Pat Halpin, and Jason Roberts Marine Geospatial Ecology Lab Nicholas School of the Environment, Duke University Lisa Ballance, Jay Barlow, Elizabeth Becker, Steven Bograd, Karin Forney, and Jessica Redfern Southwest Fisheries Science Center NOAA - National Marine Fisheries Service Dave Foley and Daniel Palacios Joint Institute for Marine and Atmospheric Research, University of Hawai`i at Manoa Grant/Cooperative Agreement Number:  NNX08AK73G

3 Spatial Decision Support System (SDSS)
Provides spatially explicit, quantitative predictions of marine mammal habitat (probability of occurrence) and species density. Average summer density Predictions based on habitat variables derived from remotely sensed products. This decision support system provides end-users with a dynamic tool to directly query marine animal observation data, oceanographic data, model results, statistical analyses and decision support tools. The SDSS is of vital importance to addressing critical marine conservation and protected species management issues for a wide sector of end-users and is directly relevant to national and international management priorities. EXAMPLE: Slide shows average summer density predictions for the northern right whale dolphin off the US west coast. To avoid interactions with this species, users could plan their activities for the southern, offshore areas during the summer season. (Black dots show actual observations) Allows environmental planners to estimate and avoid potential interactions (e.g., ship strikes, oil drilling noise, Navy exercises) with protected marine species.

4 Study Area: East and West Coasts, GoMex, ETP
90 mammal species 100,000 sightings This decision support system provides end-users with a dynamic tool to directly query marine animal observation data, oceanographic data, model results, statistical analyses and decision support tools. The SDSS is of vital importance to addressing critical marine conservation and protected species management issues for a wide sector of end-users and is directly relevant to national and international management priorities. EXAMPLE: Slide shows average summer density predictions for the northern right whale dolphin off the US west coast. To avoid interactions with this species, users could plan their activities for the southern, offshore areas during the summer season. (Black dots show actual observations)

5 Technical Approach Marine Mammal Data:
Marine Mammal Survey Data Habitat Data Statistical Models of Species Distribution & Density Marine Mammal Data: Ship and aerial surveys Habitat Data: Remotely sensed data Existing SDSS uses SST and chlorophyll concentration. Statistical models are developed using habitat variables as predictors of marine mammal occurrence and density. Due to the synoptic and near-real time nature of remotely sensed data, they provide the ideal variables for predictive models used for management purposes. In addition to remotely sensed data, habitat predictors include more static variables such as bathymetry, bottom slope, distance to shore, and distance to other features such as islands and the 2,000m isobath.

6 Spatial Decision Support System Website
Drill into model type, region, species, and season

7 Spatial Decision Support System Website
Click to turn on observation points and survey tracks

8 Spatial Decision Support System Website
Click for details

9 Spatial Decision Support System Website
Model summary statistics Species info (click for more) Dataset info (click for more) Scroll for statistical plots

10 Spatial Decision Support System Website

11 Expansion and Enhancement of the SDSS
Goal: Expand the use of earth observing information to increase the functionality and utility of this decision support system. Incorporate a wider range of remotely-sensed earth observations into species habitat and density models Incorporate more ecologically important parameters (e.g. frontal activity) derived from remotely-sensed data Specific parameters will differ between regions due to differences in ecologically important oceanic processes in the Atlantic, Pacific, and Gulf of Mexico Release important analyses and algorithms in a package of open-source desktop GIS tools, to facilitate reuse by others Explore the implemention of now-cast and forecast capabilities in the SDSS. Transition from intro to region-specific talks.

12 Enhancements to the SDSS workflow Ecologically important parameters
Species observations More Earth observations Earth observations Algorithms Statistical models Ecologically important parameters Predicted distributions Summary plots SDSS website

13 Desktop analysis with Marine Geospatial Ecology Tools
Marine Geospatial Ecology Tools (MGET) contains modular, pluggable versions of our tools in a free, open-source collection Species observations Earth observations Statistical models Predicted distributions Summary plots SDSS website Ecologically important parameters Algorithms More Earth observations

14 Example: Species distribution modeling with MGET
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 True positive rate Cutoff = 0.020 Binary classification (range map) Predicted probability of presence

15 Ecologically-important parameters: fronts and eddies
Transition from intro to region-specific talks. Image from

16 Detecting fronts with the Cayula-Cornillon algorithm
~120 km Pathfinder Daytime SST 3-Jan-2005 28.0 °C 25.8 °C Front Step 1: Histogram analysis Example output Bimodal Optimal break 27.0 °C Frequency Temperature Step 2: Spatial cohesion test Parameters for models: Distance to front Frontal activity index Strong cohesion  front present Weak cohesion  no front

17 Detecting eddy cores with the Okubo-Weiss parameter
SSH anomaly Example output Aviso DT-MSLA 27-Jan-1993 Red: Anticyclonic Blue: Cyclonic Negative W at eddy core Figures: Henson and Thomas (2007)

18 Completed transect lines 1991-2005
SWFSC CCE Shipboard Surveys Cetacean surveys conducted in summer/fall of 1991, 1993, 1996, 2001, & 2005. Surveys typically conducted July thru early Dec. Study area extends from the coast out to ~300 nm offshore. Observers on the flying bridge (highest deck on the ship) search continuously during daylight hours. Two teams of three observers rotate at 2-hour intervals among starboard observer, port observer, and data recorder positions that are located on the flying bridge of the ship (pic – highest deck on the ship). The starboard and port observers search for animals using pedestal-mounted 25x150 binoculars (pic) while the data recorder searches using unaided eye and 7x50 handheld binoculars. In addition to sighting data, changes in searching conditions, such as Beaufort sea state, are entered on a laptop computer connected to the ship’s navigation system. For NASA project we’ll have another year of data = 2008. Systematic line-transect methods were used on all surveys. Completed transect lines

19 Generalized Additive Models (GAMs) Balaenoptera physalus
Sample Density Model Results Generalized Additive Models (GAMs) Fin whale Balaenoptera physalus Plots show density predictions for each individual year BASED ON ENVIRONMENTAL CONDITIONS. Right bottom panel shows average density predictions for 5 years - Grid cells for each of the individual survey years were averaged across all years to calculate average species density. The segment specific predictions from the model were interpolated to the entire study area using inverse distance weighting to the power of 2 using Surfer 8.0 (Golden Software, Inc ). Black dots show actual sighting locations. Key Parameters dist. to 2,000m isobath, depth, SST, SVI, Beaufort

20 Improving and Refining Satellite Parameters
Derive satellite proxies to replace in situ model parameters Sea Surface Salinity (waiting for Aquarius) Thermocline depth in the CCE SSH and SST Deep Scattering Layers (prey field) Insolation and Attenuation coeffients Use improved satellite and model products Blended Satellite SST ROMS SST fields Predicted average density (AveDens) – same plot as lower right corner on previous slide, standard error (SE(Dens), and upper and lower lognormal 90% confidence limits(Lo90% and Hi90%) based on the final CCE model for fin whale. Standard errors and upper and lower lognormal 90% confidence limits were calculated from the grid cell averages and variances using standard formulae. Predicted values were then smoothed using inverse distance weighting as noted on previous slide.

21 Moving Forward - Satellite NRT
GHRSST SST fields from Remote Sensing Systems Inc. Blended SST MW/IR/OI Sep 27, 2005 Pathfinder Jun - Dec 2005 Fin whale (Balaenoptera physalus)

22 Moving Forward Redux GHRSST SST near real time data from October 15, 2008 Survey Sightings from Sep Oct 16 Sightings Sep. 27 - Oct. 16 only

23 Fin Whale Hindcast with ROMS: Linkages to Other NASA Efforts
ROMS output from the NASA FAST Project (Chavez, Chao, Chai and Barber) - will soon be providing forecasts with up to 9 months lead time. Pathfinder Sep - Dec 2005 ROMS Oct 2005 Fin whale (Balaenoptera physalus)

24 Cutting Edge: Prediction for 2008
GHRSST SST near real time data from October 15, 2008 Predicted Fin Whale density is quite unusual, and thus an excellent test. Blended SST Oct 15, 2008

25 Cutting Edge: Observed for 2008
GHRSST SST with 2008 survey sightings Very promising preliminary result Perhaps get NOAA started on providing operational support Blended SST Oct 15, 2008 Sightings Oct Data provided by Jay Barlow & team Fin whale (Balaenoptera physalus)

26 Ongoing and Future Work
East Coast team will continue to develop data products for the North Atlantic, and tools for data access and manipulation West Coast team will develop predictor variables for the CCE, and continue to explore the use of NRT Satellite data and model output to transition these products to operational status. The two teams will continue to coordinate and leverage the strengths of each in order to enhance the capabilities of the SDSS


Download ppt "Pat Halpin Nicholas School of the Environment Duke University"

Similar presentations


Ads by Google