4th International Conference on Integrating GIS and Environmental Modeling (GIS/EM4):
Problems, Prospects and Research Needs. Banff, Alberta, Canada, September 2-8, 2000.
An interdisciplinary approach to addressing neighborhood
scale air quality concerns:
the integration of GIS, urban morphology, predictive
meteorology, and air quality monitoring tools
Ronald M. Cionco
Richard A. Ellefsen
Alan H. Huber
James Gallagher
Abstract
The paper describes a project that combines the capabilities of urban geography, raster-based
GIS, predictive meteorological and air pollutant diffusion modeling, to support a neighborhood-
scale air quality monitoring pilot study under the U.S. EPA EMPACT Program. The study has
resulted in the establishment of a raster-based GIS urban morphology data set centered on Rodeo,
CA and a large series of predictive microscale airflow (and diffusion) simulations in support of
an operational mobile air quality monitoring system.
Keywords
GIS, morphology data, terrain, air quality monitoring, high resolution wind fields, urban terrain
zones, neighborhood scale, predictive modeling, digital orthophoto quadrangles
Introduction
This paper presents an overview of an ongoing environmental monitoring project within Contra
Costa County, C A that has included environmental modeling supported by geographical
information. The project is being supported under the U.S. Environmental Protection Agency's
program under the national Environmental Monitoring for Public Access and Community
Tracking (EMPACT) Program. EMPACT is a unique and pioneering Presidential initiative to
provide time-relevant information and effective public access of environmental data to citizens in
86 of the nation's largest metropolitan areas. This project is providing technical assistance to
Contra Costa County, CA to complement the EPA EMPACT Community Air Toxics Monitoring
Via "Bucket Brigade" project. The "Bucket Brigade" project is empowering citizens to collect
air samples in their community.
This project provides important scientific assistance to both solve technical problems and
provide analytical support for the ongoing community air toxics monitoring by the Contra Costa
-------
County Hazardous Materials Division (HazMat). Under the present Bucket Brigade project, the
county is supporting an effort by community groups to take air samples when they perceive a
toxic air quality incident. The county HazMat program also supports county wide risk
management and incident response programs. The county recognizes the need for real-time
measurement of community air toxics concentrations both to establish baseline values and
provide support during incidents. The county has several incident response vehicles equipped
with sampling instruments. One of these vehicles has been upgraded and outfitted as a technical
support van, the BMP ACT Mobile Air Analyses Van (EM AAV). Information capabilities
developed in this project will be critical to the county in more completely characterizing
community exposures to air toxics. This paper presents what is being done to combine the
capabilities of urban geography, raster-based GIS, predictive meteorological and air pollutant
diffusion modeling, to support neighborhood-scale air quality monitoring. Local air pollution
concentrations are very much affected by the local terrain and morphology.
Morphological information
Overview
A raster-based urban morphological data set provides requisite morphology data on the ground
surface — urban features and vegetation — of a 10km x 10km test area in northern Contra Costa
County and adjacent southern Solano County. The study area (see figure 1) contains a portion of
San Pablo Bay, the hilly area of northern Contra Costa County, the Straits of Carquinez, and the
metropolitan area of the city of Vallejo.
Physically, the study area is quite diverse. Elevations range from sea level at San Pablo
Bay and the Straits to peaks reaching approximately 260m. Slopes are generally steep at the
water's edge where a narrow shelf on the south side of the Straits is used by Southern Pacific
Railroad. Franklin Canyon is used by the Burlington Northern and Santa Fe Railroad, and
California State Route 4. A prominent feature in Solano County is Mare Island. The island
features level land on its eastern side and an area of marsh bordered by about a kilometer-wide
band of tidal mud flat on the west (San Pablo) side. In between, hilly land reaches an elevation
of 90m. Level, or gently sloping land occurs on the coast in the southwest corner of the study
area where the towns of Rodeo and Pinole are located.
Native vegetation consists of grasslands with areas of trees mainly of broadleaf evergreen
live oak trees located on east-facing slopes and along some stream courses. Other native trees
present are the broadleaf evergreen California Laurel and the broadleaf deciduous Valley Oaks.
Other trees are exotic introductions and are mainly varieties of the Australian Eucalyptus, and
pines. The usual mix of ornamentals is found in the urban areas. Trees range in height from
about five meters in thickets of live oaks to the tall Eucalyptus of about 20 meters. A common
height for live oak stands is ten meters. A similar average is seen with street and yard trees. A
high proportion of residential land use is in areas of housing tracts built with the last ten to
twenty years. These areas have many trees that have not yet reached mature height.
The unit of observation and measurement is the one hectare (100 x 100 meters) cell. Recorded
by each cell are 23 attributes that are used in the same manner as they would be with vector-
based GIS polygons. These data items are used individually to serve the purposes of the airflow
-------
and diffusion models. Single thematic maps are produced, using Arc View's Spatial Analyst.
Urban morphology features consist of: (1) industries requiring docks and dockside locations
(e.g., an oil refinery); (2) evolved commercial and residential areas of the area's older towns; (3)
recent suburban housing tracts; and (4) related modern institutional and commercial
developments.
Method
Spatial information and location control were provided by US Geological Survey Digital
Orthophoto Quadrangles (DOQs). A one hectare grid was drawn using topographical map tick
marks to bound the cells for the measurements that were made (see figures 1 & 2 for the 10 km
study area and a single square kilometer with its 100 meter grid). The Orthophotos, based on the
1983 North American Datum (NAD) were transformed to match the 1927 NAD of the
topographic maps and the wind model.
Systematic Oblique Air Photos: A systematic series of air photos was taken — from a high-wing
Cessna 150 flying at 645m altitude — for the dual purposes of: (1) providing a profile view of
urban features and trees to facilitate estimation of height and (2) to provide current information
for updating the 1993 DOQs. Photos were taken using ASA 100 color film and a 50mm lens on
a Nikon body. A total of 220 photographs were taken. All flight lines were from east to west to
achieve a northward view to aid the interpreters. Side overlap and top-bottom overlap of the
photographs were achieved.
Data Spread Sheets: All data were reported in a professionally tailored spreadsheet.
Programmed instructions disallowed incorrect entries. The program also organized the
sequential list of square kilometers by UTM coordinates thus eliminating the tedious (and error
prone) task of setting up each square kilometer prior to its interpretation. In addition, a checking
feature revealed any errors of omission or commission of data items that needed to sum to one
hundred. Color shaded data cells turned to white when datum items were entered.
Data were reported for each hectare in the study area. The 10 x 10 km area consists of 10,000
hectares. With 23 items possible for each hectare, a square kilometer had 2300 data items
possible of 230,000 for the test area. Items not occurring in a particular hectare were given a
value of zero.
Working with ground surface data (urban and vegetation) in support of meteorological models at
this one hectare resolution is micro scale, as opposed to the common meso scale at a ten km
resolution or a macro scale at a coarser resolution. However, even at the one hectare resolution,
considerable generalization of urban features must be made. Resolution of a quarter hectare (50
x 50 meters) would be a more suitable scale for the actual size of most urban features, but it
would be a very laborious effort.
Physical attributes are recorded as follows:
Building type (using an Urban Terrain Zone classification scheme)
Building density (percent of ground cover)
Building height, in meters
-------
Building orientation (to the nearest 15 degrees)
Roof pitch (flat or pitched)
Roof reflectivity (bright or dark)
Impervious surface (percent of ground covered)
Impervious surface reflectivity (bright or dark)
Bare ground (percent of cover within a hectare)
Cropland (percent of cover within a hectare)
Grassland (percent of cover within a hectare)
Marsh (percent of cover within a hectare)
Water (percent of cover within a hectare)
Coniferous trees (percent of ground covered by tree canopy within a hectare)
Coniferous trees, height in meters (to the nearest 5 meters)
Broadleaf evergreen trees (percent within a hectare)
Broadleaf evergreen trees, height in meters (to the nearest 5 meters)
Broadleaf deciduous trees (percent within a hectare)
Broadleaf deciduous trees, height in meters (to the nearest 5 meters)
Mixed trees (percent within a hectare)
Mixed trees, height in meters (to the nearest 5 meters)
Shrubs (percent within a hectare)
Shrubs, height in meters (generalized a two meters)
Using Arc View's Spatial Analyst, thematic maps were prepared of each of the attributes.
Example photo analyses for tree types and building height, are shown in figure 3. These
attributes and others were used to support wind and diffusion modeling. In GIS fashion, the
other attributes could serve other purposes.
Meteorological and diffusion models
High resolution wind model/code
HRW is a 2-dimensional, diagnostic, time independent model that simulates airflow over
complex terrain including the effects of vegetation, buildings, and simple surfaces with a high
computational resolution - such as 100 m (40 to 400 m) and for a very local area - such as 5 km x
5 km (2 km to 20 km on a side) as described by Cionco (1985). The solution is provided at 10 m
agl. As a stand-alone code, HRW is initialized with single values (at 10 m agl) of wind speed,
wind direction, temperature, pressure (usually at 2 m), and buoyancy (stability computed from an
upper air sounding and the 10 m temperature) derived from field observations or output from a
coarser mesoscale model analysis. Digitized terrain elevation also is required at each grid point
within the domain as is digitized land morphology data sets of land feature height and type
(Cionco and Ellefscn, 1998 and Ellefsen, 1999). The model's output is composed of the u and v
wind components, a vector field, calculated streamline field, temperature field, friction velocity
field, Richardson Number field, Power Law Exponent field, and a field of the partial component
of the vertical motion (not always the total w component). Each parameter is calculated and
tabulated for the entire computational array.
Physically, calculations are performed on an array of air parcels in a pressure field such that
accelerations of these parcels are determined as they negotiate the changing slopes of the terrain
and the added thermal lift or suppression component imparted by buoyancy. Computations are
for the array of cells as flux boxes defined by four adjacent grid points. The procedure makes
-------
use of Gauss' Principle of Least Constraints (Lanczos, 1962) that requires the forces to be
minimized in order to satisfy the equations of motion. Mass is conserved during the calculations.
Numerically, simulation values are obtained by direct variational relaxation of the wind field in
the layer near the surface when the numerical minimum is reached in all of the computational
cells in the domain. The calculations are completed in an iterative manner. The method of
vector of steepest decent method is used during the calculations to approach the minimum. The
solution is reached when the internal constraints forces imposed by the warped terrain surface,
thermal structure, and requirements for flow continuity are minimized. HRW has been validated
(Cionco and Byers, 1995) using the MADONA Field Study data base (Cionco et al, 1999).
Diffusion model/code
Most recently, HRW is used in tandem with a Gaussian Puff diffusion code (Cionco, 1999)
where HRW's deformed, terrain-influenced and morphology-influenced wind fields drive the
downwind diffusion of puffs and plumes modified by the interaction with the underlying
complex terrain (Cionco et al, 1997) and morphology features (Cionco, 1999) such as canopies
and buildings. RMPUFF is the Gaussian Puff diffusion model/code developed by Mikkelsen
(Thykier-Nielsen and Mikkelsen, 1993) that computes the downwind behavior of aerosol puffs
and plumes in a deformed wind field such as is provided by HRW. The code also has a puff
splitting feature to deal with plume bifurcation and flow divergence due to channeling, slope
flow, and inversion effects in complex terrain. Puff/plume diffusion processes are controlled by
local turbulence levels, either provided directly from on-site measurements or provided from a
pre-processor calculations code. RIMPUFF is further equipped with standard plume rise
formulas, inversion and ground-based reflections, gamma dose algorithms, and wet/dry
depletion. The code outputs concentration and dosage values downwind from the release point.
Multiple releases can be handled within one simulation case. A 'coupled' set of HRW and
Gaussian puff diffusion code simulations are presented to exhibit their combined effects.
Data for simulations
Data required to initialize the wind model are digitized terrain elevation, digitized morphological
features such as vegetation, buildings, and other surfaces, and local meteorological data. A
terrain extraction program is used to read the NIMA DTED CD-ROMs of terrain elevation with
100 m resolution. The morphology file cannot be produced in the same manner in that NIMA
and USGS data sets are not available in the manner that is required for the simulation. See the
section on Morphology Information for a description of the generation of the morphology data
base. To conserve space, contoured maps of terrain elevation and the combined terrain and
morphology features can be seen in subsequent figures underlying the wind field solutions.
Local meteorological data is either available from established observational sites or through
event-driven data collection efforts such as limited field studies.
Simulation strategy
The strategy for these simulations is dependent upon methodically changing the initialization
values of wind direction, wind speed, and atmospheric stability in addition to the lower boundary
-------
conditions being expressed as terrain elevation and then as morphological features added to the
same terrain elevation. Wind speed is set at 2 m/s for select runs. Wind directions are set for
every 22.5° from 0° to 360°. Atmospheric stability is set first for unstable (PSC = B) and then
stable (PSC = E) as the bulk stability of the fall domain. For these initial conditions, one can
simulate a wide variety of cases where the transport paths of aerosol materials can be readily
identified with respect to noted sources and potential downwind 'receptor' areas.
Simulation results
(Solutions presented and discussed in this limited paper are mostly for wind fields initialized
with southwest flow (SW) at 2 m/s for unstable atmospheric conditions.)
An example of airflow interaction with the terrain (and water) surface is shown in figure 4 as a
streamline flow field. Flow over the water surface is smooth and undisturbed. Over land,
however, the flow field exhibits a modest deformation of long sweeping streamlines during this
daytime simulation. When morphological features are added to the terrain field, the resultant
surface configuration becomes more roughened. Initially, the analysis with the same inputs as
the terrain-only simulation of figure 4, results in a streamline field that interacts with the
morphology features as well as the terrain to produce more deformation as shown in figure 5.
Differences are readily discerned over the 'land' area when one solution is overlaid onto the other
solution. The over water streamlines are identical. Over land with the morphology features,
however, streamlines are displaced and have shorter variations occurring especially through the
Rodeo neighborhoods.
A comparison of the effects of changing atmospheric stability conditions from daytime (unstable)
to nighttime (stable) can be seen with the inclusion of figure 6. This solution is valid for a locally
stable, nighttime atmosphere for SW flow at 2 m/s. Note that the flow field is turned somewhat
counter-clockwise compared to the flow field in figures 4 and 5. The resultant flow field differs
from each of these daytime solutions.
A further comparison of daytime conditions versus nighttime effects upon the flow field can
more readily be discerned by observing the downwind behavior of aerosol plumes over the fixed
set of combined terrain and morphology features. (Note that the morphology features for figures
7, 8, and 9 are coded for buildings as red, trees as green, grassland as yellow, marsh as blue, and
white for water surfaces except over land as bare soil/impervious surfaces. Figure 7 displays the
downwind travel (some 5 km) and behavior of a plume from the town of Hercules through the
Rodeo neighborhoods and onward to the Crockett neighborhood. The plume is broad laterally
and exhibits some compression as well as the normal expansion. The aerosol concentration
levels (five contoured values) decrease in a normal manner downwind from the source although a
modest secondary maxima occurs in the downslope flow just south of Crockett. Note that the
contour of the second highest concentration value traveled some 1000 m from the source. Figure
8 shows a notable difference for the nighttime (stable) behavior of the same plume also released
at the Hercules site. It is quite apparent that the nighttime plume travels farther (beyond 7000 m)
and is notably less wide than the daytime plume. The concentration levels fall-off normally,
however, higher concentrations travel farther downwind. By comparison, the contour of the
-------
second highest concentration value traveled some 4500 m (vs 1000 m during the daytime). It is
also worth noting that the nighttime plume misses Crockett completely as it travels more directly
to Vallejo and beyond.
A closer look at the flow field, as vectors, responding to the non-uniform array of morphology
features (buildings, trees, grasslands, impervious surfaces) for a nominal one km2 neighborhood
area is shown in figure 9. The vectors, plotted as arrows, show that the wind is accelerating and
decelerating as well as changing directions throughout the neighborhood. (Page limitations do
not permit additional examples) Coarser scale meteorological models are not capable of
generating these high resolution variations in the flow field and the resultant diffusion field on
the neighborhood scale.
Example field application
An example application of the EMAAV is presented here to demonstrate how the measurement
and modeling can be applied. On June 16 the EMAAV was operated around the refinery and was
able to identify the source of the MTBE was not from the refinery but from a tank storage area
downwind from the refinery. This conclusion is possible by examining the relationship between
potential sources and the meteorology during the event. The location of the EMAAV
measurements is identified in figure 10. Wind speed and direction measurements as collected at
two locations are presented in figure 11. No data was collected for the time period at location 3
Information for each location is summarized below. The van also continuously collects Carbon
Monoxide and Polycyclic Aromatic Hydrocarbon measurement. These data are not presented
here due to page limitation and they were not critical to this example application.
Location 1 : Cummings Skyway and ISO (downwind) 221ft. elev.
Ran BTEX and detected 0.053ppm MTBE
1203 hours to 1223 hours
Location 2 : California St. and San Pablo Ave., Rodeo (upwind) 6ft, elev.
Ran BTEX and detected (1) unknown
1233hrs,tol245hrs.
Location 3 : Hillcrest Elementary School east parking lot (upwind) 37ft. elev
No runs made
12S4hrstol259hrs.
Location 4 : ISO adjacent to the sulfur recovery unit of Tosco's refinery (downwind) 215ft, elev.
Ran BTEX with no detection
Ran E & M mercaptan and got 0.021ppm ethylmercaptan
1313 hrs. to 1330 hrs.
Potential exposure to the people living near the refinery can be assessed by examining the wind
models. In addition population exposure assessments are being supported by using an Arc View
application containing demographic information on the population.
Summary
The results of this study clearly show that:
1. the HRW model/code produces different transport (vectors and streamlines) fields for terrain-
-------
only scenarios versus the combined morphology and terrain scenarios;
2. the HRW model/code also produces different transport fields for stable and unstable
atmospheric conditions;
3, simulated transport fields readily identify the path that aerosol plumes travel with respect to
source locations versus downwind neighborhood 'receptor' areas;
4. the HRW transport fields can serve as guidance to the mobile air quality monitoring system
for the tracking and sampling of the public's reporting of neighborhood scale air quality
problems and incidents;
5. the optional effort to exemplify the downwind behavior of am air pollutant plumes with a
diffusion model further verifies the credibility of the HRW simulation capabilities for local air
quality situations. A prime example of this would be that parts of neighborhoods may report air
quality incidents where as adjacent neighborhoods just some 500m away may not experience the
same air quality incidents.
Acknowledgments
The work of two San Jose State University graduate students was vital to the project: Maureen
Kelley transformed the orthophoto quads to the correct location format, and Cheryl Anderson
prepared the spreadsheets. Both also participated in the inventorying process.
References used
(Due to page limitations here, please contact the authors for a complete list)
Note: Figures provided in HTML format can be examined in greater detail by expanding the
small format necessary for this paper
Authors
Ronald M. Cionco, Research Micrometeorologist, US Army Research Laboratory,
ATTN: AMSRL-IS-E, White Sands Missile Range, NM 88002
rcionco@arl.mil, Tel: 505 678-1572
Richard A. Ellefsen, Professor Department of Geography, San Jose State University,
One Washington Square, San Jose CA 95192
Tel: 408 924-5480, FAX: 408 924-5477
Alan H. Huber, Physical Scientist, Atmospheric Sciences Modeling Division, ARL/NOAA,
RTF, NC 27711. On assignment to the National Exposure Research Laboratory, US EPA.
huber.alan@epa.gov , Tel: 919 541-1338
James Gallagher, Hazardous Materials Specialist, Contra Costa County Health Department,
4333 Paacheco Boulevard, Martinez, CA 94553
igallagh@hsd.co.contra-costa.ca.us, Tel: 925 646-2286
Disclaimer; The U. S. Environmental Protection Agency through its Office of Research and Development funded in
part the research described here. It has been subjected to Agency review and approved for publication. Mention of
trade names or commercial products does not constitute and endorsement or recommendation for use.
-------
Digital Qrthophoto Qmtdrangle Image of the Con-
tra Costa Test Site with * kilometer UTM Grid
.65300 *6bQQQ t67GQG (68000 (69QCG «7D1DO
Figure L The 10 square kilometer study area.
Figure 2. Example one square kilometer
Example: Using Photo Images to Make Morphological Measurements
633 634 835
Industrial Area OJrbam Terrain Zone class A4) a
Adjacent Residential Area (UTZ class De3>
Digital OrthophoEc Quad
These three vi«ws ef a nsw
h(M the measi^ement Astern is \jssd. For
each of the nme K«cU>«s (note UTM deaig-
Eations on the DtXI and in the extras* »f
tbesfa«adsli6e!;}theaaaljstj«si&g She
DOQ and th« obliijue air pHotOj esiiraatas
th« values for xtrto futures aud
?a^atation. Readiljwsibie features are the
buildings, tb«.r rcof type and reR^Hwtj',
tha iniE^rvious surfaces (dart), the trwsE
thar typedHea irees, negni in meters
Bimb*.
amb, heigh) in f«e(^3
hg-4-4-
Figure 3. Example photo analyses.
3:E2:
ff
The advanSpt rfasiftg cbHi^'M asiiiii photoaraplis to daenwne height of buildings and trees
is $ic*
-------
Figure 6. Example nighttime (stable) wind field. Figure 7. Example daytime (unstable) release.
Figure 8. Example nightime release.
Figure 9. Example one square km wind field.
Wind Speed and Direction
0°
3SO
330
300
270
240
210
18Q
150
120
o Wind Speed
(MPH)
a Wild Direction
1200 1210 1220 1230 12« 1250 1260 (Degrees)
Clock Time
Figure 10. Example study near Rodeo, CA Figure 11, Wind speed (1) and wind direction (r).
-------
NEBL-RTP-AMD-00-172
TECHNICAL REPORT DATA
1, BEPORT NO,
EPA/60G/A-QO/081
2.
4, TITLE AND SUBTITLE
An Interdisciplinary Approach to Addressing Neighborhood Scale Air Quality
Concerns: The integration of GIS, urban morphology, predictive meteorology,
and air quality monitoring tools
5.REPORT DATE
^PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
'Ronald M. Cionco, 2Ricbard A. Ellefsen, 3Alan H. Huber, 4James Gallagher
8.PERFORMING ORGANIZATION REPORT NO.
9, PERFORMING ORGANIZATION NAME AND ADDRESS
1 US Army Research Laboratory, White Sands Missile Range, NM 88002
2 San Jose State University, One Washington Square, San Jose, CA 95195
3Same as Block 12
"Contra Costa County Health Dept., 4333 Paacheco Boulevard, Marline, CA
94553
10.PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
National Exposure Research Laboratory
Office 'of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
13.TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/9
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The paper describes a project that combines the capabilities of urban geography, raster-based GIS, predictive meteorological
and air pollutant diffusion modeling, to support a neighborhood-scale air quality monitoring pilot study under the U.S. EPA
EMPACT Program, The study has resulted in the establishment of a raster-based GIS urban morphology data set centered on
Rodeo, CA and a large series of predictive microscale airflow (and diffusion) simulations in support of an operational mobile
air quality monitoring system.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/ OPEN ENDED TERMS
C.COSATI
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
EPA-2220
19. SECURITY CLASS (This Report)
UNCLASSIFIED
21.NO. OF PAGES
10
20. SECURITY CLASS (This Page)
UNCLASSIFIED
22. PRICE
------- |