Loosecoupling an air dispersion model and a geographic information system GIS: Asthma and air pollut

1
Loose-coupling an air dispersion model and a
geographic information system (GIS)Asthma and
air pollution in the Bronx, New York City

• Juliana Maantay, Andrew Maroko, and Jun Tu
• presenter
• The Graduate Center of the City University of New
  York
• Lehman College, City University of New York
• NOAA Collaborators
• Dr. Ralph Ferraro and Bruce Ramsay,
• NOAA/NESDIS/STAR/Satellite Climate Studies Branch
• Prepared for NOAA-CREST conference, 2008, UPRM

2
Loose-coupling Overview

• Developed new procedures to loosely integrate an
  air dispersion model (AERMOD), and a GIS (ArcGIS)
  to simulate air dispersion from selected
  stationary point sources in the Bronx, NYC
• Five pollutants were modeled PM10, PM2.5, NOX,
  CO, and SO2.
• Plume buffers were created based on the model
  results to be used as proxies of human exposure
• Asthma hospitalization rates inside and outside
  of impact zones were compared using both plume
  buffers and fixed-distance proximity buffers
• GOAL To provide a relatively simple and feasible
  method for health scientists to take advantage of
  both air dispersion modeling and GIS by avoiding
  the need for intensive programming and
  substantial GIS expertise.

3
Asthma Background

• In 2000, Asthma was the leading cause of
  hospitalization for children in NYC
• The childhood asthma hospitalization rate in the
  Bronx was over 1.5X that of the rest of the city,
  and nearly 3X that of the U.S.
• Asthma is a complex disease, and its causes are
  not well understood
• There are a wide range of potential triggers
  for acute events including pollution, indoor air
  quality, and behavioral risk factors (e.g.
  smoking)

4
Asthma Background
Data Source NYC DOH, 2003, Asthma Facts, 2nd
Edition.
5
Proximity Analysis Previous Phases

• Work already completed in the Lehman Urban GIS
  lab has revealed associations between asthma
  hospitalization rates and proximity to known
  major pollution sources in the Bronx, NY
• Fixed-distance proximity buffers were created
  around known pollution sources (TRI, SPS, MTR,
  LAH) to quantify this association
• Asthma rates within these buffers were compared
  to those outside the buffers revealing an
  increased likelihood for hospitalizations for
  those residing within the buffers
• Even when controlling for race/ethnicity and
  income in a multiple regression, there was a
  statistically significant relationship between
  proximity to pollution sources and asthma
  hospitalization rates

6
Proximity Analysis Methodology
Fixed-distance buffers in the Bronx, NY
7
Proximity Analysis Results (from Maantay, 2005)
Odds ratios of asthma hospitalization rates
(1999) by buffer type
8
CEDS Introduction

• The Cadastral-based Expert Dasymetric System
  (CEDS)
• Developed to disaggregate population data using
  cadastral information ( residential units or
  amount of residential area) as an ancillary
  dataset
• An expert system then selects the best proxy for
  population distribution on a block-group by
  block-group basis
• Particularly useful in hyper-heterogeneous urban
  environments such as the Bronx
• Results in 150x smaller units than census block
  groups, on average

9
CEDS Methodology (from Maantay, Maroko,
Herrmann. 2007)
Methodological differences and potential
improvement of population estimation of the CEDS
method (c), over both Filtered Areal Weighting
(b),and Simple Areal Weighting (a).
10
CEDS Results (from Maantay, Maroko, Herrmann.
2007)
Percent difference between estimated block group
population and census block group population in
NYC.
11
CEDS Results (from Maantay, Maroko, Herrmann.
2007)
Simple linear regressions for estimated vs.
census block group populations(forced through
origin)
12
CEDS and Asthma Methodology (from Maantay
Maroko. Forthcoming, 2008)
13
CEDS and Asthma Methodology (from Maantay
Maroko. Forthcoming, 2008)
Asthma hospitalization rates (5-year average,
1995-1999) in the Bronx, NY comparing filtered
areal weighting (FAW) and the cadastral-based
expert dasymetric system (CEDS). Red line
indicates average rate in the Bronx.
14
Loose-Coupling Introduction (from Maantay,
Maroko Tu. Forthcoming, 2008)

• With the increased spatial resolution of the
  population data, a more realistic distribution of
  air-borne pollutant concentration was desired
• AERMOD, an air dispersion model which accounts
  for meteorological and physical variables, was
  used.
• The resultant concentration data was then
  integrated with a GIS (ArcGIS) for
  post-processing and analysis
• Plume buffers around each source were created to
  measure the local effect of air pollution on
  asthma hospitalization rates

15
Loose-Coupling Methodology (from Maantay, Maroko
Tu. Forthcoming, 2008)
Flowchart of generalized steps for loose-coupling
of AERMOD and ArcGIS.
16
Loose-Coupling Methodology (from Maantay, Maroko
Tu. Forthcoming, 2008)
AERMOD data sources and preprocessing methods.
17
Loose-Coupling Methodology (from Maantay, Maroko
Tu. Forthcoming, 2008)
NEI stationary source release points (stacks) in
the Bronx to be inputted to AERMOD
18
Loose-Coupling Methodology (from Maantay, Maroko
Tu. Forthcoming, 2008)
33 release points (19 facilities) and the output
receptor grid (500 500 feet) with 9120
individual receptors. AERMOD is run for each
pollutant and facility (95 model iterations)
19
Loose-Coupling Methodology (from Maantay, Maroko
Tu. Forthcoming, 2008)

• The outputs of AERMOD do not include the
  background pollutant concentration or other
  (non-modeled) sources, and do not represent the
  ambient air quality or the total pollution
  burden.
• In this study we only analyzed the impact of a
  stationary source on its surrounding area.
• Source Impact Index (SII) was created to
  represent the combined contribution of the five
  air pollutants from a source to the ambient air,
  by following the EPA method of calculating the
  Air Quality Index (AQI)

20
Loose-Coupling Methodology (from Maantay, Maroko
Tu. Forthcoming, 2008)

• The sub-indices for the five air pollutants at
  each receptor (point) were calculated, the
  highest sub-index is used as the SII for that
  point.
• The equation follows that of the USEPA AQI,
  expressed below

Where Ip is the sub-index for pollutant p, Cp is
the concentration of pollutant p, BPHi is the top
breakpoint that is greater than or equal to Cp,
BPLo is the bottom breakpoint that is less than
or equal to Cp, IHi is the sub-index
corresponding to BPHi, ILo is the sub-index
corresponding to BPLo.
21
Loose-Coupling Methodology (from Maantay, Maroko
Tu. Forthcoming, 2008)
Dispersion from one source. 10 of the maximum
pollutant concentration from each source is used
to define the boundary of the individual plume
buffers
22
Loose-Coupling Methodology (from Maantay, Maroko
Tu. Forthcoming, 2008)
Plume vs. Proximity Buffers
23
Loose-Coupling Results (from Maantay, Maroko
Tu. Forthcoming, 2008)
Comparison of asthma rates obtained by plume and
proximity buffers.
24
Loose-Coupling Results (from Maantay, Maroko
Tu. Forthcoming, 2008)

• Higher asthma hospitalization rates are
  associated with a higher potential exposure to
  local air pollution, rather than showing merely a
  correlation of asthma hospitalization rates and
  distance to pollution sources as earlier studies
  have found.
• The plume buffers capture more population and
  asthma hospitalizations, in absolute numbers,
  than the proximity buffers.
• This difference in captured population is not
  only due to the modeled air dispersion, but also
  because population density is not homogeneously
  distributed around each pollution source.

25
Loose-Coupling Conclusions

• The loose-coupling of AERMOD and GIS show the
  advantages of air dispersion modeling and the
  creation of plume buffers over fixed-distance
  proximity analysis.
• The case study results reinforce the earlier
  fixed-distance proximity buffer findings, but
  drive the research further, creating a more
  nuanced analysis.
• The loose-coupling provides a feasible method for
  researchers to take advantage of both air
  dispersion modeling and GIS by avoiding the need
  for intensive programming and substantial GIS
  expertise.
• The loose-coupling integrative method could have
  beneficial policy implications and value to
  environmental protection, public health, and
  other agencies in conducting impact assessments,
  and analyzing likely ramifications of land use
  decisions.

26
Loose-Coupling Future Steps

• Model additional major pollution sources
  (transportation, toxic release inventory
  facilities, etc.)
• Increase the areal extent of the study area to
  include all of New York City
• Explore possible associations between modeled air
  pollutant concentrations and additional diseases
  (e.g. cardiovascular disease)
• Explore the utility of remotely sensed air
  quality data in order to improve our
  understanding of its temporal association with
  hospitalization rates (spatio-temporal analyses,
  temporal lag between air quality change and
  change in hospitalization rates, etc.)

27
Loose-Coupling Acknowledgements

• The National Oceanic and Atmospheric
  Administrations Cooperative Remote Sensing
  Science and Technology Center (NOAA-CREST)
  provided critical support for this project under
  NOAA grant number NA17AE162.
• This research was also partially supported by
  grant number 2 R25 ES01185-05 from the National
  Institute of Environmental Health Sciences of the
  National Institutes of Health.
• The statements contained within this poster are
  not the opinions of the funding agency or the
  U.S. government, but reflect the authors
  opinions.