Transcript No Slide Title

Technologies for integration and
discovery of geospatial data
Jim Quinn
Environmental Science and Policy
and Information Center for the Environment
University of California, Davis
[email protected]
http://ice.ucdavis.edu
NBII Programs
• Gap Analysis (GAP)
– Vegetation, vertebrate distributions, protected
lands
• Long term population monitoring
– ex: Breed Bird Surveys, Christmas counts
• Integrated Taxonomic Information System (ITIS)
• International programs
– IABIN, NABIN, GBIF
• Invasive Species Council
NBII Nodes
• Regional Nodes
– California
– Pacific Basin
(Hawaii)
– Northern Rockies
– Pacific Northwest
– Tennessee
– Texas
• Thematic Nodes
–
–
–
–
Avian
Fisheries
Infrastructure
(Administration)
• Proposed
– Southwest
– Invasive Species
– Taxonomic Services
Node Vision & Objectives
Principles for
environmental informatics
based on distributed
nodes:
• Environmental
information generally
should be managed at its
source
• Core data (“Darwin
Friends and Colleagues
Multinational:
MAB
IABIN
GISP
Mexico:
CONABIO, UNAM
Brazil:
Base de Dados Tropical
Venezuela:
Universidad Central de Venezuela
Russia
Komarov Botanical Institute
United States:
USGS International Programs
USGS Nonindigenous Aquatic
Species Program
Smithsonian Environmental
Research Center
Hawaiian Ecosystems at Risk
Project
NHM & Biodiversity Research
Center, University of Kansas
California:
California Biodiversity Council
California Exotic Plant Pest Council
California Food & Agriculture
California Department of
Transportation
California Biodiversity Council
• Founded in 1992
• Heads of State and
Federal Natural
Resource Agencies
• 39 Signatories
• Science Coordinating
Committee
www.ccrisp.ca.gov
Regional and Statewide
Policy Drivers
• Land Acquisition and Habitat Planning
• Fire Protection
• Clean Water Act
– Sect. 303(d) and TMDLs
• Endangered Species Act
• Invasive species
• Supporting Stakeholder Negotiations
CCRISP Methodology to Identify State
Conservation Priorities
• High priority biodiversity lands, freshwater aquatic ecosystems
and wetlands
• Prime agricultural lands
• Rangelands
• Forest lands
• Natural lands that can sustain outdoor recreational and
educational facilities and pursuits and can accommodate visitors
in a natural setting
• Sites with significant natural cultural values (archaeological and
paleontological resources)
• Critical watershed values
• Urban open space with significant natural values or potential for
significant restoration of natural values
Goal:
Integration through tools for
land use managers
Example -- Environmental effects of urban
growth in Greater Sacramento
How should
planners identify
open space to
minimize biological
effects of
urbanization,
suburbanization and
exurbanization?
Growth in Greater Sacramento -- Preserving
Agricultural Values
HePlan -- Johnston et al, 2000
Growth in Greater Sacramento -- Rare Habitats
Johnston et al., 2000
Growth in Greater Sacramento
Habitat for Rare Species
Johnston et al., 2000
Growth in Greater Sacramento
Estimated Species Protection
Johnston et al., 2000
Policy Issues
• Mix of scales
• Incomplete or
non-digital
information
• Incompatible
classifications and
attributes
Status of critical datasets
• Vegetation
– GAP is only statewide coverage -- 1km2 MMU
– At least 7 classifications in wide use
– $35M for consensus map from imagery?
• Wetlands
– About 2/3 of state available as digital coverage
– Variable MMUs and classifications
Status of critical datasets - 2
• Riparian
– No statewide map
– At least 30 agencies and organizations hold
geolocated data
• Rare species
– High quality records
– ~ 50,000 records for ~1000 species in Heritage
Database (NDDB) -- backlog of several years
– Records sparse from conservation lands
Status of critical datasets - 3
• Management and Ownership
– Statewide coverage for public ownership
• little information on private lands
• dated
– No wide-area coverages for
• conservation easements
• habitat conservation and species recovery plans
• privately protected lands
– Conservation projects only as a point coverage
Example:
Non-Point Source Pollution
• Clean Water Act
303(d) and TMDL
• Tied to basins and
waterways
• 1:100,000 USGS base
map
Data are tied to waterbodies and watersheds
Calwater
Hydrological
subareas
and State Water
Board Waterbodies
Clean Water Act
Waterbody Assessments
Over 300,000
river miles in
California -less than 20,000
assessed
Beneficial uses and causes of impairment
chosen from elements in basin plan
Example
Drinking Water Source Protection
Safe Drinking Water Act
• 17, 000 wells, 1200 surface intakes, 10% GPSed
• Crude hydrogeology of wells
• Catalog of potential contaminating activities in “protection
area” from predefined list
– industrial solvents and dry cleaners
– mines (3000 of 30,000 mapped)
– leaking underground storage tanks
– agricultural pesticide use
Local Planning -- Geospatial Framework
Required for CEQA Compliance
• 103 Required Elements for California
Environmental Quality Act (Environmental Impact
Reports)
• Some framework geospatial data required for each
• No framework dataset is complete and
standardized among all California counties
Source: Robert Ball
California Geographic Information Association
Framework data for CEQA required elements
“Quasi-spatial” data
drives public policy
Environmental data is (almost)
always tied to geolocators, but
primary occurrences are frequently
point data with complex non-spatial
attributes
How do we
integrate
geospatial and
quasi-spatial
data?
Strategy
• Highly distributed information system
supporting locally unique data
• Network of “nodes”
• Multiple points of access or portals
• Interoperability using shared vocabularies
• Migration toward XML and W3C standards
Tools for Creating a Standard
Data Reporting Structure
• XML: Extensible Markup Language
– Emerging as a standard way to exchange
structured data
• RDF: Resource Description Framework
– Good for developing the semantics of a network
• Other technologies: Z39.50, LDAP
California geographic thesaurus elements
•
•
•
•
•
•
Bioregions
Counties
Cities
Watersheds
Waterbodies
Public land polygons (e.g.,
National Parks and
Forests)
• Place names
• Road and milepost
Research Issues
• Integrating spatial environmental data with
process models
• Monitoring strategies -- taking advantage of
remote sensing and automated change
detection
• Making polygons from (non-representative)
point data
Land Use - Anderson Valley
Mean Monthly Discharge Navarro River Watershed
1950 - 1999 USGS Gaging Station
Hyperspectral Data
• Incorporate 4m AVIRIS Hyperspectral Data into Watershed Model
• Classification of Riparian Vegetation, Stream Parameters, & Land Use
• Additional Field Work to Validate Model
Riparian-Topographic
Shading Model
Vegetation Distribution
Hourly Solar Incidence
Converted to Height by DBH Class
& Percent Hardwood / Conifer
for Critical Date: July 22
Riparian Corridor Delineation
200 meter radius from streams
1996 Aerial Photographs
Reach Averaged Values
Digital Elevation Model
10 meter resolution
attributed to linear hydrographic network
for Current Conditions & Potential Conditions
RipTopo & Aquatic Conservation
RipTopo Riparian Corridor
RipTopo
Model
Results
Current
Shading
Conditions
Potential
Shading
Conditions
Integrating spatial environmental data with
process models -- Bay-Delta fish models
• Biological monitoring data
is weekly/biweekly, but
spatially precise
• Transport models have
hour time-scales, but km
spatial resolution
• Economic models are
compartment models
Asian Longhorn Beetle
(Anoplophora glabripennis)
INPUTS
OUTPUTS
County-level
data on vascular
plants (BONAP)
Distributed
National data on
birds, mammals, and
diseases (USGS)
Information
Management and
modeling (USGS,
NASA, CSU, UCD)
Predictive models
of habitats vulnerable
to invasion
Watershed-level
data on fishes
(USGS)
•Data gathering
• Species taxonomy
• Data formatting
• Synthesis
• Predictive modeling
• Analysis and
display tools
• Data accessibility
via the web
Predictive models
of the spread of
Invasive species
Point data on public
lands (USFWS,
NPS, USGS)
Vegetation and soils
plot data (USFS,
USGS, BLM)
CLEARING
HOUSE
Web-net
National-scale maps
of non-native
species distributions
National, regional,
and local priorities
for control efforts
Reports on the
status and trends
of non-native
species in the U.S.
Current Predictive Modeling Capabilities
1.
ArcGIS: Input satellite data,
veg., soils, topography, etc.
Field Data: Invasive species
data, veg., soils, topography, etc.
2.
S-Plus: Develop Multivariate Model,screen and normalize data, test for
tolerance/multi-colinearity, and run stepwise regression.
3. S-Plus: test residuals for auto-correlation and cross-correlation (Morans-I) and
find the best model (ordinary least squares, gausian, etc. using AICC criteria).
4.
5.
6.
S-Plus/Fortran: If spatially autocorrelated, run kriging or co-kriging models.
ArcInfo GIS: develop map of model uncertainty from S-Plus
output, Monte-Carlo simulations, observed-expected values.
ArcView: produce maps of current distributions, potential distributions,
and vulnerable habitats, with known levels of uncertainty.
Future “Ecological Forecasting” Models:
Far more automated, instantaneous, and continuous!
1.
ArcView: Input satellite
data, via new sensors or
change detection models.
2.
Web-ware:
• Develop multivariate model, screen and normalize data, test for
tolerance/multi-colinearity, and run combinatorial screening.
• Test residuals for auto-correlation and cross-correlation
(Morans-I) and find the best models.
• If spatial autocorrelation exists, run kriging or co-kriging
models.
• Develop map of models uncertainty (maps with standard
errors).
• Produce maps of current distributions, potential distributions,
and vulnerable habitats, with known levels of uncertainty.
OR
Field Data: Early detection
or monitoring data, from
many sources.
3. Repeat Step 1 – always be looking for new data
Thank you