Environmental Engineering Reference
In-Depth Information
25.1 Introduction
Urban environments, because of their inherent fine-grained
heterogeneity of land use (i.e., the specific use(s) of the parcel)
and land cover (i.e., the vegetation, soil, rock, or built material
present) present a special problem when attempting to map and
model vegetation and land cover (Yang and Lo, 2002, Yuan
et al ., 2005), often requiring a specialized classification system
to be useful to ecologists and urban planners alike (Cadenasso,
Pickett and Schwarz, 2007). Previous attempts at mapping urban
areas with fine-grained detail have generally used either high
resolution imagery (e.g., IKONOS, Quickbird, SPOT), which
necessarily limits the spatial extent of a study area, or spectral
mixture analysis of medium resolution imagery (e.g., Landsat
Thematic Mapper [TM], MODIS). Landsat TM data in particular,
with its synoptic view of the earth, high (30 m) spatial resolution,
multidecade data stream, multispectral sensors designed specif-
ically for detecting vegetation, and low cost (now free), makes
it especially attractive to both ecological and urban and regional
planning studies. However, to make Landsat TM data useful for
classifying urban areas, specialized classification techniques are
required. Other chapters in this topic go into more detail on
satellite sensors and classification methods, but in my example
below I will briefly discuss several image classification techniques
(i.e., image segmentation, supervised classification, and spectral
mixture analysis) that are useful for deriving data to be used in
urban ecological modeling.
Several recent studies have used multi-date satellite imagery
to track changes in urban extent and form over time. For
example, Taubenbock et al . (2009) used a Landsat Multispetral
Scanner (MSS) and TM time-series to identify urban footprints
and spatial patterns of development for India's 12 largest urban
agglomerations from the 1970s until 2000. They used metrics
of landscape composition and configuration to classify different
urban growth types (e.g., sprawling or increased density, mono-
or polycentric, laminar or punctual). Tole (2008) used 1986 and
2001 Landsat TM satellite imagery and spectral mixture analysis
to document changes in the extent and pattern of urban land in
the greater Toronto area, Canada. Tole concluded that during
both time periods the area had a highly dispersed, sprawling
pattern of built-up land (i.e., urban development). Similarly,
Rashed (2008) used spectral mixture analysis and fuzzy logic
to classify two time periods of Landsat TM imagery for Los
Angeles, California, and landscape metrics to analyze changes in
the extent and shape of urban areas. Jat, Garg and Khare (2008)
used similar methods and determined that between 1977 - 2002,
developed land increased three times as fast as the growth in
population for Amjer, India. Each of these studies (and many
others) used satellite imagery to examine patterns and changes
in patterns of urban land cover over time which is often the first
step in understanding the status of ecological systems of a region.
Humans have been transforming the Earth's land area increas-
ingly in the past decades with nearly half showing evidence of
human manipulation (Houghton, 1994; Lambin et al ., 2001).
As human populations increase, there is increasing pressure to
convert land to urban land uses (Meyer and Turner, 1992). More
than 50% of humans now live in cities; up from only 10% in
1900 (Sadik, 1999). Urbanization, defined here as the conver-
sion of undeveloped or agricultural lands to build environments,
dramatically alters landscapes and substitutes natural processes
with human ones (Grimm et al ., 2000; Alberti et al ., 2003). The
intensification of land use caused by urbanization threatens the
long-term provisioning of ecosystem services (Foley et al ., 2005).
Specifically, altering land use of a parcel of land inevitably alters
the land cover (e.g., vegetation, impervious surface, etc.) present
on a site, thereby changing the species that can occupy that
land and changing many other processes (i.e., surface water flow,
groundwater penetration) as well.
In this chapter, I am specifically interested in understanding
how future urbanization may change the biodiversity of a region.
To address this question, I must be able to: (1) develop models
that predict future land cover; (2) develop models that estimate
biodiversity; (3) integrate the output from land cover change
models into models estimating biodiversity. Although none of
these steps are trivial on their own and the integration of multiple
models makes this process even more challenging, such integrated
modeling is the goal of many modeling efforts (Alberti and
Waddell, 2000). I will first introduce the major steps required to
address this issue and then provide an example drawn from my
work in western Washington State, USA.
25.1.1 Using urban remote sensing to
develop land cover maps for
ecological modeling
As urban remote sensing is the focus of this topic, I focus my
discussion here on specific techniques useful for creating data
necessary for ecological modeling. The use of satellite imagery and
remote sensing techniques, while employed by both urban plan-
ners and ecologists for over 30 years, has increased greatly within
the both fields in the last twenty years (Wilson et al ., 2003; Sud-
hira, Ramachandra and Jagadish, 2004; Croissant and York, 2005;
Fassnacht, Cohen and Spies, 2006; Jat, Garg and Khare, 2008;
Munroe, Tole, 2008). The utility of remotely sensed data to detect
patterns and change in patterns over time is well suited for both
fields and is discussed in detail in other portions of this topic.
Landsat sensors, in particular, have played an important role in
land cover change (Yuan et al ., 2005; Taubenbock et al ., 2009)
and ecological studies (Cohen and Goward, 2004). Urban and
regional planners utilize remotely sensed data in many ways
including tracking changes in land use and land cover (LULC)
over time. Ecologists in general, and landscape ecologists in
particular, primarily have been interested in the patterns and
changes in pattern of vegetation or land cover, often as a sur-
rogate for other features of interest such as wildlife habitat or
primary productivity (Xu et al ., 2007).
25.1.2 One example of ecological
modeling: modeling species habitat
Urban ecology, as a branch of ecology, has recently emerged as
an interdisciplinary field interested in understanding how urban
systems function (Grimm et al ., 2000; Pickett et al ., 2001; Alberti
et al ., 2003; Alberti, 2008). The resilience of urban systems is
linked to the dynamic interactions between social, economic,
and biophysical processes operating over multiple spatial and
temporal scales (Alberti and Marzluff, 2004). As land cover
Search WWH ::




Custom Search