Transcript Alternative PM2.5 Mapping

Environmental Information Systems for
Monitoring, Assessment, and Decision-making
Stefan Falke
AAAS Science and Technology Policy Fellow
U.S. EPA - Office of Environmental Information
Environmental Information Systems
Decision-making
Monitoring
Delivery/Presentation
Storage/Description
Analysis &
Assessment
Environmental Information Systems
Decision-making
Monitoring
Spatial Analysis
Delivery/Presentation
Storage/Description
Analysis &
Assessment
Environmental Information Systems
Web-based Information Systems
Decision-making
Monitoring
Delivery/Presentation
Storage/Description
Analysis &
Assessment
Environmental Information Systems
Sensor
Webs
Decision-making
Monitoring
Delivery/Presentation
Storage/Description
Analysis &
Assessment
Mapping Air Quality
Goal: Reduce the uncertainty in mapping air quality data from
point measurements. Use a data-centric spatial interpolation that is
based on physical principles.
estimated continuous surface
point monitoring data
spatial
interpolation
ci is the estimated concentration at location i
n is the number of monitoring sites
cj is the concentration at monitoring site j
wij is the weight assigned to monitoring site j
Spatial Interpolation with Monitor Clusters
Standard interpolation
applies equal weight; each
site has 1/3 of the weight
on the estimate at i.
There is a cluster of four
sites. When applying
standard distance weighted
interpolation, the cluster
will account for 2/3 of
estimated value at i while
the two single sites each
only account for 1/6 of the
total weight.
Declustered weighting
shows the proper allocation
of the 1/3 weight to the
cluster of sites.
Declustered Interpolation
Cluster weight
Inverse distance weight
Dij 
Rij
1 rjk
CWijk  
n Rij
p
 Rij
p
wij  Dij  CWij
i
X2
X1
X1
r j1
X2
r j2
r j1
r j2
Xj
X j r X3
j3
Rij
Rij
i
i
CW~ 0.25
CW~ 1.00
r j3
X3
Variance Aided Mapping
Temporal variance is indicative of local source
influenced monitoring sites.
The higher a site’s variance, the lower its interpolation
weight and the more restricted its radius of influence
during interpolation.
n
Vj 
 x  x 
i 1
i
n 1
wij  Dij  CWij V j
1
Variance Weighting Example
Interpolation weights using distance and temporal variance of daily
maximum ozone concentrations, 1991-1995
In central Ohio, most monitoring sites experience similar temporal variance in O 3 and weights
assigned to the sites are simply R-2.
In estimating O3 near St. Louis, high variance sites (St. Louis urban sites) are used along with low
variance sites (rural sites) and their respective weights are altered from R-2.
Estimated Ozone Concentrations, 1991-1995
Estimation Error
7
Kriging
Mean Absolute Error (ppb)
6.5
DIVID
6
ID
5.5
5
4.5
4
3.5
0.9
most clustered
0.75
0.6
0.45
Clusterness
0.3
0.15
least clustered
Mean estimation error at least clustered locations
with DIVID is about 10% lower than kriging and
30% lower than inverse distance.
Barrier Aided Estimation
Pollutants are “trapped” in valleys while mountain
tops have low pollutant concentrations
• Horizontal Flow Barriers (Mountains)
• Vertical Flow Barriers (Scale Height)
PM10 in California
Without Barriers
With Barriers
AIRS PM10 data (1994-1996)
Sierra Nevada Mountains are clearly
visible with barrier aided estimation
Surrogate Aided Interpolation
1991-1995
Summer
1991-1995
Summer
Extinction Coefficient
1/r2 Interpolation
Fine Mass Concentrations
1/r2 Interpolation
1991-1995
Summer
Fine Mass
Bext
1/r2 Interpolation
1991-1995
Summer
Bext Aided FM =
Fine Mass x Bext
Bext
Satellite Imagery for PM Assessment
Spaceborne sensors allow near continuous aerosol monitoring
throughout the world. When fused with surface data they provide
information on the spatial, temporal, and chemical characteristics
of aerosols than cannot be determined from any single image or
surface observation.
Goal: Fuse SeaWiFS and TOMS satellite data with surface
observations and topographic data to describe extreme aerosol
events.
1998 Asian Dust Storm
The underlying color
image is the surface
reflectance derived from
SeaWiFS.
The TOMS absorbing
aerosol index (level 2.0)
is superimposed as
green contours.
The red contours
represent the surface
wind speed from the
NRL surface
observation data base.
The blue circles are
also from the NRL
database and indicate
locations where dust
was observed.
The high wind speeds generated the large dust front seen
in the SeaWiFS, TOMS, and surface observation data.
2000 Saharan Dust
A massive dust storm transports dust
off the west coast of Africa into the
Atlantic Ocean and across the Canary
Islands.
Fuerteventura and Lanzarote Islands are fully
blanketed by the murky yellow colored dust
plume. Gran Canaria and Tenerife are partly
covered by the dust layer but their higher
elevations appear to protrude above the dust layer
at about 1200m.
Future Research Interests
•Spatial and temporal interpolation
•Uncertainty / Estimation Error Maps
•Integration of surface and satellite data
•Development of web-based spatio-temporal tools
AAAS Fellowship Program
American Association for the Advancement of Science
(AAAS) fellowship program to bring science and
engineering PhDs to D.C. and the policy process
Fellows are placed in federal agencies (EPA, State Dept.,
NSF, NIH, USAID…) and in Congress
Goal is to provide scientific expertise to offices and to gain
first hand experience in the policy process
http://fellowships.aaas.org
Interoperable Environmental Information Systems
Advances in monitoring and information technology have
resulted in the collection and archival of large quantities of
environmental data.
However, stove-piped systems, independently developed
applications, and multiple data formats have prevented
these data and the systems that serve them from being
shared.
Interoperable environmental information systems offer the
potential for attaining systems of shared information and
applications within a distributed environment.
Environmental Monitoring for Public Access
and Community Tracking (EMPACT)
Assists communities in providing sustainable public access to
environmental monitoring data and information that are
clearly-communicated, available in near real-time, useful, and
accurate
A funded EMPACT project had three
required components:
 Real Time Environmental Monitoring
 Data Analysis & Visualization
 Information Dissemination Technology
(Internet, Kiosks, Newspaper, TV, etc.)
EMPACT Project Locations
Distributed Environmental Information Network
Data
Sources
States
Publish – Make data and tools
available to the Web
Data
Users
EPA
CDX Portal
Others
Europe
EI
CEC
EI
GEIA
Web Portal
Minimize
Burden
Find – Enable the discovery of data
and tools through Web-based search
engines
Bind - Connect data and tools to
user applications for value added
processing
Maximize
Transparency
Data and Tool Description
Data
Data
Description
(Metadata)
XML
Web
Services
Wrappers
Tools
Tool
Description
Network
Distributed Environmental Information Systems
Integrated View
Parcels
Roads
Images
Boundaries
...
Whoville Cedar Lake
Queries
extract
data from
diverse
sources
Catalog
View
Whoville Cedar Lake
Web Services
Internet
Data
Wrapping
Common interfaces enable interoperability
Clearinghouse
Data Vendor
XML
Data
Metadata
City Agency
Data
Metadata
State Agency
Data
Metadata
Fed. Agency
Data
Metadata
Catalog that
indexes data,
similar to WWW’s
html search engines
Chesapeake Bay GIS Project
Participants:
- National Aquarium
- Towson University
- Maryland DNR
- Chesapeake Bay
Program
WMS
Connector
ArcIMS Server
AIRNOW
Oracle Database
Internet/Intranet
WMS
Applet
Web-based Visibility Information System
Project with EPA/OEI/EMPACT, Washington University/CAPITA,
and Sonoma Technology, Inc
Objective: To develop a web-based, near real time visibility and
PM2.5 mapping system
Phase 1: Map visibility every 6 hours using Naval
Research Lab’s Surface Observation Data
Phase 2: Incorporate ASOS Data into mapping
system
Phase 3: Use visibility as a surrogate for mapping
PM2.5
Quebec Fires,
July 6, 2002
SeaWiFS satellite and
METAR surface haze shown in
the Voyager distributed data
browser
Satellite data are fetched from
NASA GSFC; surface data
from NWS/CAPITA servers
SeaWiFS, METAR
and TOMS Index
superimposed
5-year EPA Geospatial Architecture Vision
Users
I
n
t
e
r
o
p
e
r
a
b
l
e
Data Sources
EPA
Geo
Services
Catalog
W
e
b
T
o
o
l
s
EPA
Geo
Services
Georeporting
Geoprocessing
Mapping
Geo Data &
Tools Indexes
States/
Tribes
Others
S
e
r
v
e
r
s
EPA
Enterprise
Portal
EPA
CDX Portal
System of Access
NSDI Node
EPA
Feds
Industry
Geospatial
One-Stop
States
Civilian
Locals
GeoMetadata
•
•
•
Feds
Geography Network
Red arrows and dotted lines indicate information flow based on standards, such as XML
The Open GIS Consortium (OGC)
• The Open GIS Consortium (OGC) is a
not-for-profit, international consortium
whose 250+ industry, government, and
university members work to make
geographic information an integral part
of information systems of all kinds.
• Operates a Specification Development
Program that is similar to other
Industry consortia (W3C, ISO, etc.).
• Also operates an Interoperability
Program (IP), a global, innovative,
partnership-driven, hands-on
engineering and testing program
designed to deliver proven
specifications into the Specification
Development Program.
OGC Vision
A world in which
everyone benefits from
geographic information
and services made
available
across any network,
application, or platform.
OGC Mission
To deliver
spatial interface
specifications
that are openly available
for global use.
Open GIS Web Services (OWS) Vision
• Creates evolutionary, standards-based framework
to enable seamless integration of online
geoprocessing and location services.
• Future applications assembled from multiple,
network-enabled, self-describing geoprocessing
and location services.
• Break down barriers between real world,
information about real world, and users.
Open GIS Web Services
Sponsors, Participants, and Coordinating Organizations
Participants
Compusult
CubeWerx
Coordinating Organizations
Dawn Corp.
Sponsors
Urban Logic, CIESIN, NYC DOITT, NYC DEP,
FEMA, EPA Region 2
DLR
ESRI
FGDC
Galdos Systems
GeoConnections Canada
GMU
Lockheed Martin
Common Architecture
Intergraph
NASA
Working Group
Ionic Software
NIMA
Laser-Scan
USGS
Sensor Web
Web Mapping
PCI Geomatics
US EPA
Working Group
Working Group
Polexis
USACE ERDC
SAIC
CANRI
Demo Integration
Social Change
Online
Syncline
OGC IP Team
YSI
OGC
OGC
Management Team Architecture Team
University of
Alabama
Huntsville
BAE, LMCO, NASA, TASC, GST, Image Matters, OGC Staff
Vision for NY
Sensor Webs
Sensor Webs are web-enabled sensors that can seamlessly
exchange data with other web-based applications and can
communicate with one another – leading to “dynamic networks”
Advances in micro-electronics, nanotechnology, and wireless
communication have provided the potential for the development
of environmental sensors that will provide major leaps in the
available coverage, timeliness, and resolution of monitoring
information.
Will enable spatially and temporally dense environmental
monitoring
Sensor Webs will reveal previously unobservable phenomena
since they can be placed in areas not previously suitable for
monitoring
OWS Sensor Collection Service Clients
Distributed Information System Workshops
Distributed Data Dissemination, Access, & Processing (3DAP)
July 2001
- Institutional Interoperability
Web-based Environmental Information Systems for Global
Emission Inventories (WEISGEI)
July 2002
- Bring together Information Sciences and Atmospheric Sciences
Future Research Interests
•Council on Environmental Cooperation (CEC) Integration of Emission Inventories for North America
•Development of a Fire Emissions Inventory
•Web Services (Tools) development
•Implementation of sensor webs for air quality studies
•Policy impacts of real time environmental information
Future Project Interests
•Advanced spatial and temporal interpolation techniques (surrogate data) and
corresponding estimation error maps
•Web services – going beyond placing maps on the Web interoperability
•Smart Sensors and Sensor Webs
Data
•Information driven environmental management
bases
Data Description, Format and Interface
Standards
Web-based Services
Gov’t
(Integration, Aggregation, Mapping,
Modeling)
Industry
Catalogs & Query
Tools
Browsers / Client
Applications
Public
Sensors
DIVID vs. Kriging
ASOS Visibility Measurements
Prior to 1994, visual range was recorded
hourly by human observations
Human observations were replaced with
automated light scattering instruments of the
Automated Surface Observing System
(ASOS)
The ASOS sensor measures the extinction
coefficient as one-minute averages and
calculates visual range based on a running
10-minute average of the one-minute
measurements
Lens-to-lens
3.5 feet
projector
detector
photocell
Forward scatter ASOS
visibility sensor
ASOS for Air Quality Studies
•Currently, available only at a quantized resolution of 18
binned ranges with a visual range upper bound of 10 miles,
even though the instrument can provide meaningful data up
to 20-30 miles.
•In the near future, it is anticipated that ASOS data will be
available at their full resolution on the web in “real-time.”
•Even at full resolution, they are of limited use in the western
U.S. because visual range there is often in excess of 30 miles.
•The application to “real-time” mapping (hourly or less)
needs to be evaluated
Surface Observations Extinction Coefficient
Network Assessment and Network Design
Goal: Develop methods for assessing the performance of air
quality monitoring networks using a multi-objective
“information value” approach.
Five measures of network performance considered:
•Persons/Station measures the number of people in the ‘sampling zone’ of
each station.
• Spatial coverage measures the geographic surface each station covers.
• Estimation uncertainty measures the ability to estimate the
concentration at a station location using data from all other stations.
• Pollutant Concentration is a measure of the health risk.
• Deviation from NAAQS measures the station’s value for compliance
evaluation.
Estimation Error, E
•
The estimation error is determined by
– selectively removing each site from the database
– estimating the concentration at that site by spatial interpolation
– setting the error as the difference between the estimated and measured values, E = Est.-Meas.
PM2.5 Error
< -3 μg/m3
-3 - -1 μg/m3
-1 - +1 μg/m3
+1 - +3 μg/m3
> +3 μg/m3
PM2.5 Station Sampling Zones
•
•
•
•
Every location on the map is assigned to the closest monitoring station.
At the boundaries the distance to two stations is equal.
Following the above rules, the ‘sampling zone’ surrounding each site is a polygon.
The area (km2) of each polygon is calculated in ArcView.
Census Tract Population
• The population data used
for determining a station’s
population is from ESRI’s
census tract file with
estimated 1999
populations.
• The centroid of each
census tract is associated
with a station area.
• The census tract
populations for all
centroids that fall within a
station’s area are summed.
PM2.5 Network Performance Rankings
Equal weighting of measures
Red=High Ranking
Blue=Low Ranking
Bio Sketch
B.A. Physics
1992
Courses that examined science and
technology in the context of other fields such as law,
history, and political science
M.S. Engineering & Policy
1993
Courses covered economic, legal,
management, and public policy dimensions of science
and technology
Thesis examined information flow in
environmental policy making and use of
“hypermedia” in the policy making process
Basketball in German Bundesliga
1994
Bio Sketch
D.Sc. Environmental Engineering (1999)
• Mapping Air Quality
• OTAG Data Analysis Workgroup
1995-2000
• PM-Fine Data Analysis Workgroup
• Network Assessment & Design
• Taught Geostatistics and GIS Data Analysis Lab
Research Associate (2000)
• Integration of Satellite Imagery and Surface-based
monitoring data
Center for Air Pollution
Impact and Trend Analysis
Bio Sketch
American Association for the Advancement of
Science (AAAS) Fellowship (current) –Washington
D.C.
• Environmental Monitoring for Public Access and
Community Tracking (EMPACT) Program
• Data Integration and web mapping projects including:
Open GIS Consortium Standards
Visibility/PM2.5 Web-mapping
Chesapeake Bay GIS
PM2.5 Estimates using Visibility Surrogate
1998 Central American Fires
SeaWiFS, TOMS, and
visibility indicate high aerosol
concentrations from Central
America transported over the
central U.S.
The smoke is transported
north into the upper Midwest
and to the east. The extinction
coefficient is highest further
north than the highest TOMS
aerosol index.
Smoke plumes over Central
America appear over low
elevation terrain, while high
elevation regions remain
mostly smoke free.