Embedding and Extending GIS for Exploratory Analysis of Large-Scale Species Distribution Data

Embedding and Extending GIS for Exploratory Analysis

of Large-Scale Species Distribution Data

Jianting Zhang, Dept. of Computer Science

The City College of the City University of New York

Le Gruenwald, School of Computer Science

The University of Oklahoma

Outline

•Background and Motivation

•Modeling/Representation for Data Integration

•LEEASP: The Prototype System for Visual Exploration

•Related Works and Discussions

NEON Infrastructure Overview

William K. Michener Deborah Estrin, http://www.projectscience.org/workshop7/talks/estrin.pdf

Aquatic Arrays

Terrestrial Arrays

4

Background

• Enabling Technologies– GPS technology in modern field survey

– Geo-referring technology in transforming descriptive museum records to geographical coordinates

– Internet and the cyber-infrastructure for distributed data access/integration

– Spatial databases and GIS for data management and analysis

• Species distribution analysis– Quantifying the relationship between species distributions and the

environment

– Central to ecology/biogeography theories and conservation practices

– Incorporating climate change and human impact scenarios

Background

1. Guisan, A. and N. E. Zimmermann (2000). Predictive habitat distribution models in ecology. Ecological Modelling 135(2-3): 147-186.

2. Waide, R. B., M. R. Willig, et al. (1999). The relationship between productivity and species richness. Annual Review of Ecology and Systematics, 30, pp. 257-300.

3. Stockwell, D. R. B. and D. P. Peters (1999). The GARP modelling system: problems and solutions to automated spatial prediction. International Journal of Geographic Information Systems 13(2): 143-158.

4. Hirzel, A. H., Hausser, J., Chessel, D.,Perrin, N., 2002. Ecological-niche factor analysis: How to compute habitat-suitability maps without absence data? Ecology, 83(7), 2027-2036.

Total 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999

1 751 143 166 166 100 85 46 35 9

2 383 22 56 58 64 47 52 41 29 14

3 240 35 60 59 24 22 23 10 4 1 1

4 123 33 37 25 12 9 7

Background

•USDA PLANT Database •89759 plant species in 3141 US counties

WWF Wildfinder database:•29112 species, 4815 genus, 445 families, 69 orders in 4 classes (amphibians, reptiles, birds, and mammals) among the world’s 845 ecoregions•350045 species-ecoregion records

•USGS•Little tree species distribution data: 679

•NatureServe species distribution maps•5743 amphibians species worldwide •4273 birds species of the western hemisphere •1786 mammals species of the western hemisphere

The availability of compiled digital datasets

Background

EnvironmentSpecies

Taxonomic (Linnaean ranks) Kingdom Phylum Class Order Family Genus Species SubSpecies

Phylogenentic

Area

Water-Energy

Latitude

Altitude

Productivity

Environmental Gradient

Community – Ecosystem – Biome – Biosphere

Phylogeography

Background

Geographical

Distribution

Correlation

DistributionConfiguration

EnvironmentalTaxonomic

Motivations•We aim at developing an integrated data model/representation that seamlessly links geographical, taxonomic and environmental data.

•We utilize state-of-the-art visualization techniques to build a prototype to allow visual explorations between and among relevant data:

•Embedding GIS for visualizing geographical maps

•Incorporating Graph/tree visualization for taxonomic trees and ecoregion hierarchies

•Using Sortable Table, Parallel Coordinate Plot (PCP) and other techniques for multivariate environmental data

Data Modeling/Representation

GIS Data ModelLayer 1

Layer 2

Layer n

Species 1

Species 2

Species n

Using Traditional GIS Data Model

Data Modeling/Representation

•The relationships among the geographical units in different layers are not a part of the traditional GIS data models.

•To use the layer-based GIS data model for managing multiple species distribution data, the geographical and the environmental data need to be joined for each layer, either permanently or dynamically.

•While it is possible to arrange the species layers into groups in modern GIS to mimic the taxonomic hierarchy, it is difficult to identify/visualize query results that involve multiple layers back in the layer list.

Problems

Data Modeling/RepresentationThe Integrated Data Model

GIS Data Model

Data Modeling/Representation

Environmental

Relational(RDBMS)

From/To

Environmental

Geographical

Taxonomic

TaxonomicGeographical

GIS

•Object-Relational Framework

•Taxonomic data is now first-class citizen

Data Modeling/Representation

Geographical

EnvironmentalTaxonomic

T->G(+E) E->G(+T)

G->E

G->T

Supported Operations

Operations need to be formally defined!

LEEASP: Prototypehttp://www-cs.ccny.cuny.edu/~jzhang/tech/LEEASPV10.zip

Geographical View

Taxonomic View

Environmental View

Ecoregion View

Linked Environment for Exploratory Analysis of Large-Scale Species Distribution Data

LEEASP: Prototype

•USGS NA Little dataset: 679 tree species, 90 Megabytes in ESRI Shapefile format

•WorldClim 10 minutes altitude and 18 bioclimate variables

•EPA NA Ecoregion data: up to Level III

•Resolution: 0.5*0.5 Deg

•11777 valid cells

Example Data

LEEASP: PrototypeGeographical View

•Embedding GIS

•Based on open source JUMP GIS from Vividsolutions

•Designed to present the distribution information

•Follows “Focus+Context” principle

On-screen digitizing to specify environmental gradients

LEEASP: Prototype

Details

SummaryControlOverview

Environmental View

LEEASP: PrototypeTaxonomic View

1

0

1

0

0

1

0

2

31

11

0

0

1

0

0

1

0

1

0

1

0

0

0

0

OR

1

……

……

……

G->T T->G

LEEASP: PrototypeEcoregion View

•Using the same API for Taxonomic view

•Based on Prefuse (Jeffrey et al, 2005)

•Efficient Tree Layout algorithms

•Advanced information visualization functions (zoom/animation)

LEEASP: Prototype

Coordinated Multiple View

•Overview+Detail

•Focus+Context

Related Works/Discussions•USGS (1990s): Climate-Vegetation Atlas of the North America (http://pubs.usgs.gov/pp/p1650-a/)

•Prasad and Iverson (1999-ongoing):A Climate Change Atlas for 80 Forest Tree Species of the Eastern United States http://www.fs.fed.us/ne/delaware/atlas/ (Forest Service)

•Spatiotemporal data modeling and visualization (Andrienko et al 2003, Guo et al 2006)

•Tree and graph visualization research (Bongshin et al 2004, Hillis et al 2005, Graham and Kennedy 2005, Parr et al 2007)

Related Works/Discussions

•LEEASP focuses on dynamic visualizations through user interactions rather than delivering static mapping results.

•LEEASP provides multi-way mapping among geographical, ecoregion, environmental and taxonomic data

•Views in LEEASP represent the four types of data are coordinated: when a subset of data in one view is selected through the graphic user interfaces, the subset of data will be identified and highlighted in other views.

Related Works/Discussions

•Future work

•Better formalization of the integrated data model

•Conduct more thorough user evaluations by domain scientists

•Distributed data integration based SOA

•Explore “mashup” technologies

Acknowledgements•Prefuse and JUMP GIS open source development teams.

•This work is supported in part by NSF grant ITR #0225665 SEEK and NSF grant ATM #0619139 CEO:P-COMET.

•Thanks to Profs. Robert K. Peet (UNC) and Jessie Kennedy (Napier University, UK) for taxonomy help.

•Thanks to Dr. Weimin Xi (TAUM) and Anantha M. Prasad (USDA Forest Service) for evaluating the prototype and providing constructive suggestions.

•Special thanks to three anoymous ACM-GIS conference reviewers for their comments and suggestions.

• Conference travel is supported by faculty startup fund from the Grove School of Engineering, the City College of the City University of New York.

Q&A

[email protected]

27

http://www-cs.ccny.cuny.edu/~jzhang/tech/LEEASPV10.zip