Remote Sensing and Geospatial Analysis for Landscape Pattern Characterization - Landscape Ecology for Sustainable Environment and Culture - page 203

decomposition of each pixel into independent endmembers or pure materials to
conduct image classification at sub-pixel level (e.g. Verhoeye and Wulf 2002 );
(4) the incorporation of the spatial characteristics of the neighboring (contextual)
pixels to develop object-oriented classification (e.g. Walker and Briggs 2007 );
(5) the fusion of multi-sensor, multi-temporal, or multi-source data for combining
multiple spectral, spatial, and temporal features and ancillary information in the
image classification (e.g. Tottrup 2004 ); and (6) the use of artificial intelligence
technology, such as rule-based classifiers (e.g. Schmidt et al. 2004 ), artificial
neural

networks (e.g. Zhou and Yang 2011 ), and support vector machines

(e.g. Yang 2011b ), for pattern classification.

11.2.2 Landscape Pattern Quantification

Once a land cover map is available, landscape patterns can be quantified using
landscape indices or metrics. These metrics are algorithms measuring the diversity,
homogeneity or heterogeneity of a landscape. Although many earlier efforts have
been made to identify various metrics that are meaningful for ecological func-
tionality and biological diversity, it was the first primary release of the FRAG-
STATS software package in 1995 that helped landscape ecologists to revolutionize
the analysis of landscape structure (Kupfer 2012 ). FRAGSTATS defines eight
major groups of structural or functional metrics: area/density/edge, shape, core
area, isolation/proximity, contrast, contagion/interspersion, connectivity, and
diversity (McGarigal and Marks 1995 ). Structural metrics measure the physical
composition or configuration of a landscape without explicit reference to an
ecological process, while functional metrics explicitly measure landscape pattern
that is functionally relevant to the organism or process under consideration
(McGarigal 2002 ). These metrics can be commonly measured at three levels:
patch, class, and landscape. Patch-level metrics characterize the spatial context of
patches, class-level metrics integrate over all patches of a given land cover type,
and landscape-level metrics synthesize over the entire landscape. With the
development of GIS software engineering, some metrics originally defined in
FRAGSTATS have been incorporated into other software packages, such as Patch
Analysis ( http://www.cnfer.on.ca/SEP/patchanalyst/ ) and LANDISVIEW ( http://
kelab.tamu.edu/standard/restoration/restoration_tools.htm ) .
Landscape metrics offer an intuitive tool to measure landscape structure that
can be linked with specific ecological processes. Because the input data (i.e.
categorical land cover maps) can be derived from remote sensor imagery and the
software toolkit is readily available, landscape metrics have been widely used for
landscape pattern analysis. Nevertheless, there have been well documented con-
cerns in the use of landscape metrics. Firstly, despite well-documented guidelines
being given on the use of various metrics (e.g. McGarigal and Marks 1995 ;
Haines-Young and Chopping 1996 ; McGarigal et al. 2002 ) and more than two
decades of extensive research, interpreting landscape metrics beyond the simple