Environmental Engineering Reference
In-Depth Information
membership in several hierarchical levels depending on the
spatial scale of segmentation (e.g., a Vegetation superclass, a
Canopied Vegetation class, and a Deciduous Tree subclass).
with membership in each determined by fuzzy classification
algorithms using spectral character, shape, and neighborhood
relationships across the hierarchical levels as inputs. Built mate-
rials can be identified in similar fashion (e.g., image objects with
high length to width ratios, low reflectance in the visible to
near infrared wavelengths), and location in or near an ''urban''
superclass image object could be identified as asphalt roadways.
The OBIA approach has also been used for LULC classification
of medium resolution (15 m/pixel) ASTER data of Phoenix, AZ
and Las Vegas, NV (Schopfer and Moeller, 2006), suggesting that
it may be useful for semiautomated classification using historical
Landsat datasets.
The High Ecological Resolution Classification for Urban
Landscapes and Environmental Systems (HERCULES) classi-
fication scheme of Cadenasso, Pickett and Schwarz, (2007) uses
OBIA of very high resolution aerial photography and lidar data
to classify six urban land cover types: coarse-textured vegeta-
tion, fine-textured vegetation; bare soil, pavement, buildings,
and building typology (spatial arrangement, general structural
type, and general height classes). The classification scheme is
informed by the vegetation-impervious - soil, or VIS classifica-
tion model of Ridd (1995). While the HERCULES classification
scheme is specifically designed for urban ecological studies, the
output can easily be converted into building height and vegetation
percentage format for input into climate models.
that data take geometry and power/wavelength output can be var-
ied. This is in contrast to passive optical remote sensing systems
that typically have a stable viewing and illumination geometry
that relies on the Sun for energy input to the land surface. The
spatial resolution of orbital SAR sensors such as ENVISAT, the
European Remote Sensing (ERS) satellite, and TerraSAR-X - 1 to
more than 10 m per pixel - has limited their use for urban build-
ing form and height extraction as unambiguous interpretation
of the data is difficult. Airborne systems such as TopoSAR and
the Phased Array Multifunctional Imaging Radar (PAMIR) can
provide subdecimeter resolutions that greatly facilitate mapping
of urban building formand height, as well as provide information
on the composition of the urban materialsthemselves(Brenner
and Roessing, 2008; Dell'Acqua, 2009).
Three-dimensional building data sets are increasingly derived
from airborne lidar systems. Lidar provides DEMs and DTMs,
which allow the derivation of building size, shape, orientation,
relative location to other buildings and other urban morpho-
logical features (trees, highway overpasses, etc.; also see Chapter
6, this volume). The finest resolution is typically on the order
of 1-5 m. Lidar provides a high-resolution representation of
urban morphological features for entire metropolitan areas, with
a minimal set of airplane flyovers. However, lidar is costly and
presents a data management challenge given the massive size of
datasets (Ching
et al
., 2009). Using building data in conjunc-
tion with a geographic information system (GIS) Brown
et al
.
(2002) developed scripts for automating the calculations of UCM
building-related input parameters (see Table 21.1). Those data
can then be correlated with LULC data sets, or building attributes
can be accepted by the atmospheric model in a spatially explicit
manner, i.e. they do not need to be associated with a particular
land use or land cover class.
Based on the need for advanced treatments of high resolution
urban morphological features in meteorological and air qual-
ity modeling systems, the ''Community''-based National Urban
Database and Access Portal Tool (NUDAPT) was developed
(Ching
et al
., 2009). NUDAPT is currently sponsored by the US
Environmental Protection Agency (USEPA) but involves collab-
orations and contributions from federal and state agencies, and
from national and international private and academic institu-
tions. NUDAPT contains archived copies of lidar DEMandDTM
data currently being acquired by the National Geospatial Agency
(NGA; formerly the National Imagery and Mapping Agency).
When completed, NGA will have obtained data from as many
as 133 urban areas. NUDAPT provides a web-based portal tech-
nology which facilitates the customization of data handling and
retrievals to generate gridded fields of urban canopy parameters
for UCMs. Data are currently available for Houston (Texas, USA)
and New York (New York, USA).
Techniques to build georeferenced three-dimensional build-
ing models using aerial and ground-based imagery, together
with traditional ground maps, have been developed (Gatzidis,
Liarokapis and Brujic-Okretic, 2007). A typical workflow is
to use ground mapping data to accurately position building
footprints, followed by extrusion to correct building height using
ancillary vector data. Lastly, ground or aerial imagery can be used
to construct a photorealistic representation of a given building.
Similar tools are now publicly available through Web-based
geospatial browsers, e.g. Google Building Maker (http://
sketchup.google.com/3dwh/buildingmaker.html) and provide
an alternative means of generating three-dimensional urban
models. The three-dimensional building representations must
21.3.2
Building data
As mentioned in Section 21.2 for the application of the single-
layer andmultilayer UCMs, building-relatedmodel input param-
eters are necessary. Three-dimensional building data sets and
derived building statistics have become readily available for
many cities around the world (Ratti, Sabatino and Britter, 2002,
Brown
et al
., 2002, Burian
et al
., 2006, 2008, Martilli, 2009). Print
stereographic aerial photography and manual photogrammetric
techniques have been used extensively in urban areas to derive
such parameters as building heights and dimensions for planning
purposes; an extensive review of the subject is outside the scope
of this chapter, but Jensen (2000) provides a good introduction
to the subject. More recently, photogrammetric analysis of aerial
orthophotographs derived from digital camera data has become
the state of the art. The availability of photogrammetry software,
global positioning system (GPS) and ground survey data, and
increased desktop computing capabilities has facilitated the use
of digital aerial photography for creation of both digital terrain
models (DTMs) and digital elevation models (DEMs) that allow
for the extraction of building locations and dimensions at very
high spatial resolution.
Data acquired from both synthetic aperture radar (SAR) and
interferometric synthetic aperture radar (InSAR) satellite and
airborne systems has been used to map urban building form, dis-
tribution, density, and in some cases to extract building heights by
measuring reflected energy from the land surface and roughness
elements such as buildings, trees, etc. (Brenner and Roessing,
2008; Dell'Acqua and Gamba, 2006; Gamba, Houshmand and
Saccani, 2000; Gens and Van Genderen, 1996; also see Chapter 5,
this volume). Both SAR and InSAR systems are active in the sense
Search WWH ::
Custom Search