435
A typical field campaign might focus on an optimal time to collect data from a RS
perspective (e.g., the dry season in the tropics or the summer growing season in
higher latitudes). However, biological communities are inherently dynamic, and the
visibility of different species changes with ontogeny, season, and over longer time
spans due to disturbance, invasion, succession, climate change, and other processes.
Consequently, our ability to detect species richness will vary with time, often in
ways that are poorly understood. The few studies that have investigated the temporal
aspect of biodiversity using RS or spectral reflectance demonstrate the importance
of temporal dynamics when examining the optical diversity-biodiversity links.
A study of California chaparral (Zutta 2003) found a clear seasonal dependence
in the ability to distinguish plant functional types using reflectance spectra
(Fig. 16.4a). Different methods involving contrasting spectral indices and bands
yielded distinct seasonal patterns, indicating the importance of the spectral dimension (further discussed below) and illustrating interactions between temporal and
spectral dimensions. In that study, photosynthetic and flowering phenology contributed to the seasonal patterns observed. Similarly, when evaluating several functional
leaf traits with spectral reflectance in tropical species, Chavana-Bryant et al. (2017)
found a clear seasonal effect on the trait retrievals using PLSR, again emphasizing
an interaction between temporal and spectral dimensions. A study of optical diversity for prairie species also revealed strong phenological effects that were different
at the leaf and canopy scales, demonstrating an interaction between temporal and
spatial scale (Fig. 16.4b). Clearly, the temporal dimension should be considered in
any study of the RS of biodiversity, yet most studies have been limited to a single
time frame, limiting the power to distinguish biodiversity. These examples also
demonstrate that measurement technique (e.g., instrument foreoptics) and interacting scale effects can influence optical diversity.
16.2.4 Spectral Scale
The advent of hyperspectral sensors, both imaging and nonimaging, provides rich
opportunities for exploring spectral features related to biodiversity. Approaches
range from detection of species or functional traits to methods based on the information content of the spectra themselves (Fig. 16.1; Table 16.2). All of these methods require attention to spectral scale, including spectral resolution and range,
which influence biodiversity detection in complex ways. Furthermore, our methods
of analysis, ranging from simple vegetation indices to more complex full-spectral
statistical methods (Table 16.2), explore the spectral dimension in different ways
and to varying degrees. To date, relatively few studies have explicitly addressed
spectral scale in the context of biodiversity detection, but most show that more spectral information is generally better than less (e.g., Asner et al. 2012). Consequently,
hyperspectral sensors are more informative than multiband sensors, and full-range
(VIS-SWIR) detectors are usually more useful than limited range (e.g., VIS-NIR)
detectors for detecting plant traits or biodiversity. The importance of spectral scale
16 Consideration of Scale in Remote Sensing of Biodiversity
A typical field campaign might focus on an optimal time to collect data from a RS
perspective (e.g., the dry season in the tropics or the summer growing season in
higher latitudes). However, biological communities are inherently dynamic, and the
visibility of different species changes with ontogeny, season, and over longer time
spans due to disturbance, invasion, succession, climate change, and other processes.
Consequently, our ability to detect species richness will vary with time, often in
ways that are poorly understood. The few studies that have investigated the temporal
aspect of biodiversity using RS or spectral reflectance demonstrate the importance
of temporal dynamics when examining the optical diversity-biodiversity links.
A study of California chaparral (Zutta 2003) found a clear seasonal dependence
in the ability to distinguish plant functional types using reflectance spectra
(Fig. 16.4a). Different methods involving contrasting spectral indices and bands
yielded distinct seasonal patterns, indicating the importance of the spectral dimension (further discussed below) and illustrating interactions between temporal and
spectral dimensions. In that study, photosynthetic and flowering phenology contributed to the seasonal patterns observed. Similarly, when evaluating several functional
leaf traits with spectral reflectance in tropical species, Chavana-Bryant et al. (2017)
found a clear seasonal effect on the trait retrievals using PLSR, again emphasizing
an interaction between temporal and spectral dimensions. A study of optical diversity for prairie species also revealed strong phenological effects that were different
at the leaf and canopy scales, demonstrating an interaction between temporal and
spatial scale (Fig. 16.4b). Clearly, the temporal dimension should be considered in
any study of the RS of biodiversity, yet most studies have been limited to a single
time frame, limiting the power to distinguish biodiversity. These examples also
demonstrate that measurement technique (e.g., instrument foreoptics) and interacting scale effects can influence optical diversity.
16.2.4 Spectral Scale
The advent of hyperspectral sensors, both imaging and nonimaging, provides rich
opportunities for exploring spectral features related to biodiversity. Approaches
range from detection of species or functional traits to methods based on the information content of the spectra themselves (Fig. 16.1; Table 16.2). All of these methods require attention to spectral scale, including spectral resolution and range,
which influence biodiversity detection in complex ways. Furthermore, our methods
of analysis, ranging from simple vegetation indices to more complex full-spectral
statistical methods (Table 16.2), explore the spectral dimension in different ways
and to varying degrees. To date, relatively few studies have explicitly addressed
spectral scale in the context of biodiversity detection, but most show that more spectral information is generally better than less (e.g., Asner et al. 2012). Consequently,
hyperspectral sensors are more informative than multiband sensors, and full-range
(VIS-SWIR) detectors are usually more useful than limited range (e.g., VIS-NIR)
detectors for detecting plant traits or biodiversity. The importance of spectral scale
16 Consideration of Scale in Remote Sensing of Biodiversity