ecosystems. When statistical clusters are grouped into LULC classes, in which
smaller areas (e.g., pixels) are combined into larger ones (e.g., patches), both spatial
resolution and statistical information are lost (Clapham, 2003).
4.3 OBSERVATIONAL SCALE AND IMAGE SCENE MODELS
Spatial resolution is a function of sensor altitude, detector size, focal size, and system
configuration (Jensen, 2005). Spatial resolution is closely related to the term of spatial
scale (Ju et al., 2005). As a matter of fact, spatial resolution defines the “measurement
scale” (Lam and Quattrochi, 1992) or the “observational scale” of a sensor.
Spatial resolution defines the level of spatial detail depicted in an image, and it is
often related to the size of the smallest possible feature that can be detected from an
image. This definition implies that only objects larger than the spatial resolution of a
sensor can be picked out from an image. However, a smaller feature may sometimes
be detectable if its reflectance dominates within a particular resolution cell or it has a
unique shape (e.g., linear features). Another meaning of spatial resolution is that a
ground feature should be distinguishable as a separate entity in the image. But the
separation from neighbors or background is not always sufficient to identify the
object. Therefore, the concept of spatial resolution includes both detectability and
separability. For any feature to be resolvable in an image, it involves consideration of
spatial resolution, spectral contrast, and feature shape. Jensen and Cowen (1999)
suggested that the minimum spatial resolution requirement should be one-half the
diameter of the smallest object of interest. For two major types of impervious surface,
buildings (perimeter, area, height, and property line) and roads (width) are generally
detectable with the minimum spatial resolution of 0.25–0.5 m, while road centerline
can be detected at a lower resolution of 1–30 m (Jensen and Cowen, 1999). Before
1999, lack of high-spatial-resolution images (less than 10 m) is a main reason for
scarce research on urban remote sensing before 2000 (Weng, 2012). The mediumspatial-resolution images (10–100 m), such as Landsat and SPOT, were not readily
available and were expensive to most researchers from developing countries. For a
remote sensing project, image spatial resolution should not be the only factor needed
to be considered. The relationship between the geographical scale/extent of a study
area and the spatial resolution of the remote sensing image has to be studied
(Quattrochi and Goodchild, 1997). For mapping at the continental or global scale,
coarse-spatial-resolution data are usually employed. Gamba and Herold (2009)
assessed eight major research efforts in global urban extent mapping and found
that most maps were produced at the spatial resolution of 1–2 km. When using coarseresolution images, a threshold has to be defined with respect to what constitutes an
urban/built-up pixel (Lu et al., 2008; Schneider et al., 2010).
With the advent of very high resolution satellite imagery, such as IKONOS
(launched in 1999), QuickBird (2001), and OrbView (2003) images, great efforts
have been made in the applications of these remote sensing images in urban studies.
High-resolution satellite imagery has been applied in mapping impervious surfaces in
urban areas (Cablk and Minor, 2003; Goetz et al., 2003; Lu and Weng, 2009; Wu,
OBSERVATIONAL SCALE AND IMAGE SCENE MODELS
65
smaller areas (e.g., pixels) are combined into larger ones (e.g., patches), both spatial
resolution and statistical information are lost (Clapham, 2003).
4.3 OBSERVATIONAL SCALE AND IMAGE SCENE MODELS
Spatial resolution is a function of sensor altitude, detector size, focal size, and system
configuration (Jensen, 2005). Spatial resolution is closely related to the term of spatial
scale (Ju et al., 2005). As a matter of fact, spatial resolution defines the “measurement
scale” (Lam and Quattrochi, 1992) or the “observational scale” of a sensor.
Spatial resolution defines the level of spatial detail depicted in an image, and it is
often related to the size of the smallest possible feature that can be detected from an
image. This definition implies that only objects larger than the spatial resolution of a
sensor can be picked out from an image. However, a smaller feature may sometimes
be detectable if its reflectance dominates within a particular resolution cell or it has a
unique shape (e.g., linear features). Another meaning of spatial resolution is that a
ground feature should be distinguishable as a separate entity in the image. But the
separation from neighbors or background is not always sufficient to identify the
object. Therefore, the concept of spatial resolution includes both detectability and
separability. For any feature to be resolvable in an image, it involves consideration of
spatial resolution, spectral contrast, and feature shape. Jensen and Cowen (1999)
suggested that the minimum spatial resolution requirement should be one-half the
diameter of the smallest object of interest. For two major types of impervious surface,
buildings (perimeter, area, height, and property line) and roads (width) are generally
detectable with the minimum spatial resolution of 0.25–0.5 m, while road centerline
can be detected at a lower resolution of 1–30 m (Jensen and Cowen, 1999). Before
1999, lack of high-spatial-resolution images (less than 10 m) is a main reason for
scarce research on urban remote sensing before 2000 (Weng, 2012). The mediumspatial-resolution images (10–100 m), such as Landsat and SPOT, were not readily
available and were expensive to most researchers from developing countries. For a
remote sensing project, image spatial resolution should not be the only factor needed
to be considered. The relationship between the geographical scale/extent of a study
area and the spatial resolution of the remote sensing image has to be studied
(Quattrochi and Goodchild, 1997). For mapping at the continental or global scale,
coarse-spatial-resolution data are usually employed. Gamba and Herold (2009)
assessed eight major research efforts in global urban extent mapping and found
that most maps were produced at the spatial resolution of 1–2 km. When using coarseresolution images, a threshold has to be defined with respect to what constitutes an
urban/built-up pixel (Lu et al., 2008; Schneider et al., 2010).
With the advent of very high resolution satellite imagery, such as IKONOS
(launched in 1999), QuickBird (2001), and OrbView (2003) images, great efforts
have been made in the applications of these remote sensing images in urban studies.
High-resolution satellite imagery has been applied in mapping impervious surfaces in
urban areas (Cablk and Minor, 2003; Goetz et al., 2003; Lu and Weng, 2009; Wu,
OBSERVATIONAL SCALE AND IMAGE SCENE MODELS
65