Marine Geospatial Ecology Tools Jason Roberts Ben Best

Marine Geospatial Ecology Tools Jason Roberts, Ben Best, Daniel Dunn, and Pat Halpin, Duke University Marine Geospatial Ecology Laboratory, Durham, NC The Marine Geospatial Ecology Tools (MGET) MGET is targeted at coastal and marine researchers and GIS analysts who work with spatially-explicit oceanographic and ecological data in scientific or managerial workflows. The initial MGET releases focus on tools useful in habitat studies, including tools for processing and sampling remotely-sensed oceanographic data and mapping and filtering ARGOS satellite telemetry. Many tools have both single-input and multi-input (batch processing) implementations. Key features • • Free, open-source software written mainly in Python, R and MATLAB Distributed as a self-installing setup program, for easy installation All tools include full user documentation For easy execution from many environments, each tool is exposed from: ◦ A Python class ◦ A dual-interface Microsoft COM class (on Windows) ◦ An Arc. GIS geoprocessing toolbox • Verbose logging system eases troubleshooting of difficult failures • All tools written to maximize reliability, interoperability and performance A more complex tool: Identify SST Fronts in a GOES 10/12 Image To test theories that animals associate with sea surface temperature (SST) fronts, ecologists need automated methods for identifying them in SST images. The tool presented here implements the 1992 Cayula-Cornillon algorithm. It is agnostic about which SST data are used. This example uses GOES data from NASA JPL. In collaboration with the Inter-American Tropical Tuna Commission (IATTC), we are investigating the habitats of pelagic sea turtles in the Eastern Tropical Pacific (ETP). In this simplified example, we sampled the NOAA NODC 4 km AVHRR v 5. 0 SST images and pixel quality flags at purse seine set points where fisheries observers recorded the presence of sea turtles. To explore the turtles’ possible SST preferences, we plotted SST histograms for high-quality pixels. Arc. GIS geoprocessing model Cayula-Cornillon edge detection algorithm The algorithm passes a moving window over the SST image, flagging windows that exhibit bimodal, spatially-cohesive distributions of pixel temperatures, and tracing the SST values that optimally separate the two populations. Example application: Sea turtle habitat in the ETP Mexico Step 1 28. 0 °C For more information: http: //code. env. duke. edu/projects/mget jason. roberts@duke. edu Tools under development In 2006, we developed 81 tools in a prototype tools package. We elected not to release this package because it lacked an installer, documentation, and was too tightly coupled to Arc. GIS. In late 2006, we developed a new tool framework suitable for public distribution and began rewriting the tools for it. We anticipate completion in July 2007. Pre-release builds are available for download at our website. Many of the tools listed below are still being rewritten for the new framework. If you need one now but it’s still being rewritten, we can give you the old implementation. Tools currently under development (partial listing) For each purse seine set point, given its date, generate the file names of the daily SST and pixel quality images Sea. Wi. FS Chl-a Front 25. 8 °C 120 km Tool parameters Step 2 Test 1: Bimodal distribution of pixel temperatures in the window Frequency A simple tool: HDF SDS to Arc. GIS Raster Aviso SSH, currents Optimal break 27. 0 °C Step 3 From Cayula and Cornillon (1992) Many oceanographic remote-sensing data are distributed in HDF format, but Arc. GIS 9. x cannot read them. This tool converts a Scientific Data Set (SDS) contained in an HDF file to an Arc. GIS raster. Pixel SST Test 2: Spatial cohesion of the two temperature populations Invoking the tool from Arc. GIS geoprocessing models Strong cohesion front present Batch-convert SST and pixel quality images to Arc. GIS raster format Batch-sample the rasters at purse seine set points GOES 10/12 SST Weak cohesion no front For more information about the modeling tools, attend these Geo. Tools talks: Example output • E 03. Habitat and Connectivity Arc. GIS Toolboxes (Habitat. Toolbox And Connectivity. Toolbox) for Multivariate Regression and Graph-Theoretic Marine Applications – Benjamin Best and Pat Halpin – Wednesday March 8, 10: 30 -12: 00, Kensington A Results This exploratory analysis suggests that in the ETP, olive ridley turtles inhabit slightly warmer water than green turtles. Because the sampling design was determined by fishing effort, spatial, temporal and other biases in it must be considered before robust conclusions are drawn. from Geo. Eco. Data. Management. HDFs import HDF. SDSTo. Arc. GISRaster(u'c: \temp 16\199001. s 04 m 1 pfv 50 -sst-16 b. hdf', u'c: \temp 16\sst 199001', u'sst', -180, -90, 0. 0439453125, 0) From VBScript and other languages via Microsoft COM Nearly all modern languages can call Microsoft COM components, including VBScript, VB. Net, Java, JScript, C++, C#, R, S-Plus, and MATLAB. Olive Ridley Turtles Density From Python as a native module Set hdf = WScript. Create. Object("Geo. Eco. HDF") hdf. SDSTo. Arc. GISRaster "c: \temp 16\199001. s 04 m 1 pfv 50 -sst-16 b. hdf", _ "c: \temp 16\sst 199001", _ "sst", -180, -90, 0. 0439453125, 0 • H 01. Benthic Complexity Modeling with Coarse Grain (90 m) Bathymetric Data: Is It Possible? – Daniel Dunn and Pat Halpin – Thursday March 8, 11: 00 -12: 30, Kensington E Invitation to collaborate Green Turtles Are you searching for collaborators to assist in the development of your coastal or marine geoprocessing tool? Are you searching for a release vehicle for a tool you’ve written? Please contact us to see how we can help! SST (°C) References Acknowledgements Cayula, J-F and P Cornillon. 1992. Edge detection algorithm for SST images. J. Atmos. Oceanic Technol. 9: 67 -80. The Duke researchers would like to thank M. Hall, N. Vogel, and C. Lennert of the IATTC for sharing the fishery observer data. Marine Geospatial Ecology Laboratory