recorded on the ground. Specifically, field spectroradiometry deals with analysis of
the position of specific reflectance features, shape of the spectrum, spectral variability,
and similarity (Rao et al., 2007). These properties have naturally high degree of
variation for most of the land covers on Earth. As a result, field spectroradiometry is
capable of producing a “spectral fingerprint” for such variations of many land cover
types (McCoy, 2004).
On the other hand, remote sensing has been an attractive source in the determination of land cover spatial distribution, providing valuable information for delineating
the extent of land cover classes as well as for performing temporal land cover change
analysis at various scales (Kavtzoglou and Colkesen, 2009). The general circumstances that make it attractive for this purpose include its capability to provide
inexpensive and repetitive data over large regions, even for inaccessible locations, and
at a wide range of spatial and temporal scales. Remote sensing data in land cover
classification mapping started to be routinely applied from the late 1960s (Petropoulos
et al., 2012a). Since then, a wide range of spaceborne multispectral systems have been
placed in orbit. Multispectral imagery from either airborne or satellite sensors has
been widely used to map broad vegetation groups or land cover classes. Nevertheless,
those systems, although able to measure at very high spatial resolution, record
information in a small number of distinct spectral bands, thereby providing obviously
limited spectral information content on characteristics related to Earth’s surface
objects, including land cover.
Recent advances in remote sensor technology have led to the launch of hyperspectral systems. Similar to field spectroradiometry, those are able to record reflected
light from land surface objects in numerous narrow, virtually continuous spectral
bands from the visible to the short-wave infrared parts of the electromagnetic
spectrum, thereby acquiring vast amounts of spectral information observed from
higher altitudes. The potential of hyperspectral remote sensing imagery to improve
discrimination among similar land cover classes, especially in diversely covered and
landscape-fragmented regions on Earth, in comparison with traditional multispectral
images, has already been highlighted by many investigators (e.g., Rao et al., 2007;
Cho et al., 2009; Zomer et al., 2009; Petropoulos et al., 2011, 2012a,b; Elatawneh
et al., 2012). Furthermore, as the challenge in remote sensing is concerned with
identifying complex surface features from aircraft and spaceborne imagery, field
spectroradiometry emerges as a highly suitable means for such investigations. Indeed,
recent studies have clearly demonstrated the contribution of field spectroradiometry
and in situ hyperspectral libraries as a means for achieving accurate discrimination of
vegetation cover on a species level. The use of hyperspectral data on both ground and
space scales has already been presented as a relative success in mapping vegetation
cover at species level (Cho et al., 2009; Mathur et al., 2002; Price, 1994; Thenkabail
et al., 2000, 2004a; Zomer et al., 2009).
An identical radiometric principle is followed in recording data from spaceborne
and field scale observations. However, spaceborne-level studies that investigate land
cover with the support of field spectroradiometry require careful consideration of the
factors arising from the different scales. At the spaceborne scale, the radiance received
at the sensor accounts for the effects of absorption and scattering of Earth’s
286
HYPERSPECTRAL REMOTE SENSING WITH EMPHASIS ON LAND COVER MAPPING
the position of specific reflectance features, shape of the spectrum, spectral variability,
and similarity (Rao et al., 2007). These properties have naturally high degree of
variation for most of the land covers on Earth. As a result, field spectroradiometry is
capable of producing a “spectral fingerprint” for such variations of many land cover
types (McCoy, 2004).
On the other hand, remote sensing has been an attractive source in the determination of land cover spatial distribution, providing valuable information for delineating
the extent of land cover classes as well as for performing temporal land cover change
analysis at various scales (Kavtzoglou and Colkesen, 2009). The general circumstances that make it attractive for this purpose include its capability to provide
inexpensive and repetitive data over large regions, even for inaccessible locations, and
at a wide range of spatial and temporal scales. Remote sensing data in land cover
classification mapping started to be routinely applied from the late 1960s (Petropoulos
et al., 2012a). Since then, a wide range of spaceborne multispectral systems have been
placed in orbit. Multispectral imagery from either airborne or satellite sensors has
been widely used to map broad vegetation groups or land cover classes. Nevertheless,
those systems, although able to measure at very high spatial resolution, record
information in a small number of distinct spectral bands, thereby providing obviously
limited spectral information content on characteristics related to Earth’s surface
objects, including land cover.
Recent advances in remote sensor technology have led to the launch of hyperspectral systems. Similar to field spectroradiometry, those are able to record reflected
light from land surface objects in numerous narrow, virtually continuous spectral
bands from the visible to the short-wave infrared parts of the electromagnetic
spectrum, thereby acquiring vast amounts of spectral information observed from
higher altitudes. The potential of hyperspectral remote sensing imagery to improve
discrimination among similar land cover classes, especially in diversely covered and
landscape-fragmented regions on Earth, in comparison with traditional multispectral
images, has already been highlighted by many investigators (e.g., Rao et al., 2007;
Cho et al., 2009; Zomer et al., 2009; Petropoulos et al., 2011, 2012a,b; Elatawneh
et al., 2012). Furthermore, as the challenge in remote sensing is concerned with
identifying complex surface features from aircraft and spaceborne imagery, field
spectroradiometry emerges as a highly suitable means for such investigations. Indeed,
recent studies have clearly demonstrated the contribution of field spectroradiometry
and in situ hyperspectral libraries as a means for achieving accurate discrimination of
vegetation cover on a species level. The use of hyperspectral data on both ground and
space scales has already been presented as a relative success in mapping vegetation
cover at species level (Cho et al., 2009; Mathur et al., 2002; Price, 1994; Thenkabail
et al., 2000, 2004a; Zomer et al., 2009).
An identical radiometric principle is followed in recording data from spaceborne
and field scale observations. However, spaceborne-level studies that investigate land
cover with the support of field spectroradiometry require careful consideration of the
factors arising from the different scales. At the spaceborne scale, the radiance received
at the sensor accounts for the effects of absorption and scattering of Earth’s
286
HYPERSPECTRAL REMOTE SENSING WITH EMPHASIS ON LAND COVER MAPPING