Environmental Engineering Reference
In-Depth Information
Heynen, Perkins and Roy, (2006) used urban tree canopy
data derived from aerial photography, US Census data capturing
racial and other socioeconomic characteristics, and interviews
with key managers of the urban forest in Milwaukee, Wisconsin.
These authors find a racially inequitable distribution of tree cover,
where whites are more likely to live in neighborhoods with more
tree cover and Hispanics less so. However, these authors note
that urban forest management has complex relationships with
socioeconomic status, as the character of urban trees in socioe-
conomically distressed neighborhoods often differs substantially
from those in more elite neighborhoods. Trees in poor neigh-
borhoods, which often sprout up in unmanaged or, marginally
managed, land may be considered a nuisance, while planted trees
in managed landscapes add value to residential properties.
Another way that remote sensing may be employed for envi-
ronmental justice analysis concerns the sensing of environmental
hazards. The use of remote sensing for observing and planning for
environmental hazards is well established, including for natural
disasters such as hurricanes, volcanoes, flooding, and wildfires,
among others (Hong, Adler and Verdin, 2007). While there is
clear promise for using remote sensing to analyze the equity of
the distribution of such environmental hazards with regards to
socioeconomic character, researchers have only just begun to
exploit remotely sensed data that capture hazards as a resource
for environmental justice research. One particular hazard that
has been addressed is extreme heat. Remote sensing researchers
have analyzed the urban heat island effect for some time (Oke,
1982; Roth, Oke and Emery, 1989; Lo, Quattrochi and Luvell,
1997), and image products from sensors that record thermal
emissivity can be integrated with socioeconomic data.
Both Lo and Faber (1997) and Li and Weng (2007) use
temperature data derived from Landsat thermal band imagery
and find that higher temperature is associated with indicators of
socioeconomic disadvantage. This is not surprising, given that
temperature is positively correlated with impervious surface and
population density, lending supporting evidence that tempera-
tures are higher in inner-city urban areas, which also tend to be
home to the largest concentrations of the poor and minorities in
the US. In an analysis of Tucson, Arizona, Harlan
et al
. (2006)
found similar results, where the poor andminorities had a greater
exposure to heat stress, as well as reduced access to green vegeta-
tion and open space. These populations also had fewer resources
to mitigate the effects of extreme heat, such as air conditioning
and swimming pools.
approach has been to acquire socioeconomic data at a particu-
lar unit of data aggregation, typically as defined by the census
organization or agency responsible for collecting the data, and
aggregate (or disaggregate) other data of interest to the census
units. For example, one might record the presence or absence of
a hazardous facility within each US Census tract. Or, the total
volume, weight, or toxicity of toxic chemicals released may be
tallied for each spatial unit. Alternatively, one may define zones
of degree of risk, and tally socioeconomic characteristics within
each zone. These zones may be defined by, say, proximity to a
hazardous facility, or by an area within a modeled plume of pol-
lution from a point source. Such an approach typically requires
disaggregation of socioeconomic data that are only available at
spatial units incompatible with the spatial units that capture
environmental hazard.
Approaches for disaggregation of population data are gener-
ally captured by methods of areal interpolation and dasymetric
mapping (Eicher and Brewer, 2001; Mennis, 2003; Mennis and
Hultgren, 2006). US Census data products, for example, typically
model population and population character using a spatially
exhaustive tessellation of spatial units, which assumes a homoge-
neous distribution of population within each unit. The simplest
approach to disaggregation is simple areal weighting, where an
area of overlap with the spatial unit is apportioned a percentage
of the population of the unit proportional the percentage of
the unit's area that is occupied by the overlap. In dasymetric
mapping, an additional, ancillary, data set is used to inform the
disaggregation of population data captured in choropleth map
form to an improved model of population distribution. A variety
of remotely sensed ancillary data sources have been used for this
purpose, including pixel reflectance values and image texture, as
well as classified data products (Yuan, Smith and Limp, 1997;
Harvey, 2002; Wu, Qiu and Wang, 2005). In the case of classified
land cover data, for example, one may disaggregate data from
census units by exploiting the fact that few people live on water,
and that population density is likely to be higher in urban as
compared to agricultural land covers.
A handful of studies have employed dasymetric mapping to
disaggregate population data for environmental justice studies.
Maantay (2007) has developed the Cadastral-Based Expert Dasy-
metric Mapping System (CEDS) to disaggregate US Census data
using parcel-level data. This approach was applied to an envi-
ronmental justice analysis of flood hazard in New York City.
Results indicate that simple areal weighting tended to under-
count the population at risk, and particularly minorities, from
flooding, suggesting that the nature of spatial data integration can
substantially influence the results of environmental justice anal-
yses (Maantay and Maroko, 2009). Higgs and Langford (2009)
employ a variety of techniques to disaggregate population data
to environmental risk zones in an environmental justice study
of landfills in Wales, United Kingdom. These researchers experi-
ment with a variety of dasymetric mapping techniques and find
that evidence of environmental inequity is sensitive to themethod
by which one considers a population exposed to environmental
hazard. Other comparative studies have found similar evidence
(Mohai and Saha, 2006; Most and Sengupta, 2004).
Mennis (2002) used classified land cover data derived from
Landsat TM imagery to disaggregate U.S. Census population data
to a statistical surface representation for Philadelphia, Pennsylva-
nia. This allowed for fine resolution modeling of the relationship
between demographic character and proximity to hazardous
facilities contained in EPA databases. Results suggested that
16.4
Integrating remotely
sensed and other spatial
data using GIS
One of the major methodological challenges concerning envi-
ronmental justice concerns the technique used to assess the
population at risk from a particular hazard. GIS is used to inte-
grate various types of data and derive spatial relationships among
environmental hazards, amenities, and population character (see
Chakraborty and Armstrong, 1997). Typically the goal of the data
integration is to produce a set of observations that include associ-
ated measurements of hazard and socioeconomic character that
can be used in multivariate statistical analysis. The conventional
Search WWH ::
Custom Search