Image Processing Reference
surface to the incident radiation, and can be considered a gross measure of landscape
“brightness” (Sabins 1997 ). The MODIS 16-day albedo product contains several
different measurements; we use the broadband visible through shortwave infrared
(0.3-5.0 mm) reflectance albedo value as this represents the full wavelength range
of surface reflectance (Schaaf et al. 2002 ).
A promising way to study urban landscapes is the spatially focused approach
of patch dynamics (Zipperer et al. 2000 ). The urban landscape is a mosaic
of biotic and abiotic patches (or land cover types) within a matrix of human-
induced settlements, technical infrastructure and other human-caused landscape
modifications. Spatial heterogeneity within an urban landscape has both natural
and human sources. This spatial heterogeneity exerts environmental pressures,
both on the “natural” areas in cities (forests, large parks and wetlands) and
beyond the boundaries of cities, by reducing the size and contiguity of
ecosystem patch types (Smith 1980 ). The impacts from urbanization around
and within cities affect areas of high economic, recreational and ecological
value such as agricultural and forest areas. The mechanisms of these impacts
include increasing run-off, deforestation, soil erosion, habitat fragmentation
and change in biodiversity (Haff 2002 ). In contrast, the creation of new green
spaces also occurs in some urban areas; this enables recreation and water
infiltration (drinking water reservoirs) possibilities.
Within the last 10 years landscape metrics have been implemented on remote
sensing data for different mapping scales to emphasize the spatial content
and patch distribution of classified remote sensing data (Wu et al. 2000 ). The
field of quantitative landscape ecology has been the primarily developer of
landscape metrics or indices. Since these metrics vary in time and space they
provide useful tools for monitoring a particular landscape by providing various
measures of the distribution and shape of ecological patches on the landscape.
Numerous metric algorithms have been developed to quantify the spatial
contiguity and shape of patches (i.e. area of contiguous pixels, perimeter vs.
area, etc.; McGarigal and Marks 1995 ). These algorithms can be applied to
both pixels and vector polygon data.
As many urban monitoring programs are searching for useful indicators of
landscape change, landscape metrics deliver a starting point for the comparison
of urban areas that is not dependant on their physical and cultural setting. Several
recent studies (Alberti and Waddell 2000 ; Barnsley and Barr 2000 ; Herold et al.
2002, 2005 ; Huang et al. 2007 ; Narumalani et al. 2004 ; Rainis 2003 ; Whitford
et al. 2001 ; Yu and Ng 2007 ) demonstrate the usefulness of the application of
landscape metrics for the assessment and evaluation of urban structure, planning
and ecology. The challenge remains to find an adequate way to standardize
recently developed approaches in order to ensure the comparability of results on
a regional, national or even worldwide scale.