431
grain size and extent of field sampling (Turner et al. 1989). There is an abundant
literature on the importance of considering scale when exploring ecological phenomena (Ehleringer and Field 1993; Levin 1992) and a number of “scaling rules”
have emerged. For example, Levin (1992) stated the importance of matching the
scale of observation to the phenomenon (and grain size or patch size) in question,
and specific rules of sample grain size have been developed (e.g., O’Neill et al.
1996). However, these rules are often violated when remotely sensing biodiversity.
In RS, scale has several aspects or dimensions (Malenovský et al. 2007). The
spatial dimension typically refers to the pixel size (grain size) and spatial extent of
a remotely sensed image. The temporal dimension can refer to the time of sampling,
repeat frequency, or temporal extent of sampling. The spectral dimension includes
band position, bandwidth (full width half maximum, FWHM, and sampling interval), and range. The directional or angular dimension, including the illumination or
viewing angle, leading to variations in anisotropic reflectance and the bidirectional
reflectance distribution function (BRDF), is also an important scale dimension in
optical RS, as it strongly influences the ability to detect signals present in reflectance spectra (Schaepman-Strub et al. 2006; Malenovský et al. 2007; Gamon 2015).
These scale dimensions often interact, with the effects of one influencing the effects
of another, so it is often best to consider multiple scale dimensions together. A consideration of sampling scale in all these dimensions is relevant to a discussion of RS
of biodiversity. Sampling across scales often reveals critical information that is not
apparent from a single-scale observation alone. Below, we consider these scale
effects in more detail by providing examples of how these dimensions impact biodiversity assessment from RS.
16.2.1 Biological Scale
Biodiversity exists across multiple scales of organization from genes to biomes. A
detailed discussion of the various “dimensions” of biodiversity at different scales of
organization is beyond the scope of this chapter and has been reviewed elsewhere
(e.g., Magurran 2004; see also Cavender-Bares et al., Chap. 2, this volume). In the
context of this chapter, a key challenge lies in matching the scale of the measurement approach to the biological scale of organization, a topic considered in the sections below.
16.2.2 Spatial Scale
Optical instruments used for sampling biodiversity range from laboratory spectrometers and proximal field spectrometers to airborne and satellite-based imaging spectrometers, with grain sizes (pixel sizes) covering approximately six orders of
magnitude (Fig. 16.2). If we include the molecular cross-section of DNA (e.g.,
16 Consideration of Scale in Remote Sensing of Biodiversity
grain size and extent of field sampling (Turner et al. 1989). There is an abundant
literature on the importance of considering scale when exploring ecological phenomena (Ehleringer and Field 1993; Levin 1992) and a number of “scaling rules”
have emerged. For example, Levin (1992) stated the importance of matching the
scale of observation to the phenomenon (and grain size or patch size) in question,
and specific rules of sample grain size have been developed (e.g., O’Neill et al.
1996). However, these rules are often violated when remotely sensing biodiversity.
In RS, scale has several aspects or dimensions (Malenovský et al. 2007). The
spatial dimension typically refers to the pixel size (grain size) and spatial extent of
a remotely sensed image. The temporal dimension can refer to the time of sampling,
repeat frequency, or temporal extent of sampling. The spectral dimension includes
band position, bandwidth (full width half maximum, FWHM, and sampling interval), and range. The directional or angular dimension, including the illumination or
viewing angle, leading to variations in anisotropic reflectance and the bidirectional
reflectance distribution function (BRDF), is also an important scale dimension in
optical RS, as it strongly influences the ability to detect signals present in reflectance spectra (Schaepman-Strub et al. 2006; Malenovský et al. 2007; Gamon 2015).
These scale dimensions often interact, with the effects of one influencing the effects
of another, so it is often best to consider multiple scale dimensions together. A consideration of sampling scale in all these dimensions is relevant to a discussion of RS
of biodiversity. Sampling across scales often reveals critical information that is not
apparent from a single-scale observation alone. Below, we consider these scale
effects in more detail by providing examples of how these dimensions impact biodiversity assessment from RS.
16.2.1 Biological Scale
Biodiversity exists across multiple scales of organization from genes to biomes. A
detailed discussion of the various “dimensions” of biodiversity at different scales of
organization is beyond the scope of this chapter and has been reviewed elsewhere
(e.g., Magurran 2004; see also Cavender-Bares et al., Chap. 2, this volume). In the
context of this chapter, a key challenge lies in matching the scale of the measurement approach to the biological scale of organization, a topic considered in the sections below.
16.2.2 Spatial Scale
Optical instruments used for sampling biodiversity range from laboratory spectrometers and proximal field spectrometers to airborne and satellite-based imaging spectrometers, with grain sizes (pixel sizes) covering approximately six orders of
magnitude (Fig. 16.2). If we include the molecular cross-section of DNA (e.g.,
16 Consideration of Scale in Remote Sensing of Biodiversity