281
West et al. 2017) and die-off (Boyte et al. 2015; Weisberg et al. 2017). Rather than
just using images selected during green up and senescence periods, extracting
phenology metrics from vegetation indices to refine cheatgrass classifications has
also been successful (Bradley and Mustard 2008); however, in arid and semiarid
environments, these indices can be highly influenced by rock and soil and should be
used with caution (Singh and Glenn 2009). Huang and Geiger (2008) showed that a
multi-date imaging approach can be successful even when natural phenologies of
natives and nonnatives are similar. For example, native grasses and Lehmann lovegrass
(Eragrostis lehmanniana) responded differently to unusual amounts of cool season
precipitation, which allowed new tissues in invasive species to grow, making the two
grasses distinguishable with multi-date imaging (Huang and Geiger 2008).
Imaging spectroscopy to map grassland IAS also often depends on differences in
phenology, but the higher spectral resolution and typically higher spatial resolution
afforded by airborne platforms often allow for more detailed and early detection
maps. Image acquisition timing is important for species that exhibit differences in
coloration throughout the year, such as flowering species or deciduous shrubs. In the
case of leafy spurge (Euphorbia esula), hyperspectral instruments are better
equipped to detect changes in flowering and thus have a higher success rate when
compared to multispectral instruments (Mitchell and Glenn 2009). Leafy spurge has
characteristic yellow flowers that bloom in early summer, and tamarisk leaves turn
from yellow-orange to orange-brown in autumn before leaf drop. This distinct pigmentation enables remote detection using both imaging spectroscopy (Williams and
Hunt Jr 2002; Glenn et al. 2005) and multispectral data (Anderson et al. 1993;
Everitt et al. 1995; Evangelista et al. 2009). The blue-green color of new stems and
the red-brown color of older stems help detection of spotted knapweed (Centaurea
maculosa) from imaging spectroscopy (Lass et al. 2002; Lawrence et al. 2006). For
early detection of goldenrod (Solidago altissima), an invasive moist tall grass in
Japan, hyperspectral images acquired during early spring before full development
of the grass canopy make it easier to map the exposed understory (Ishii and
Washitani 2013).
Differences in canopy architecture or plant morphological traits, such as plant
height and pubescence, can also be exploited when plants share similar phenologies
or imagery is unavailable when growth cycles show key differences. Broom snakeweed (Gutierrezia sarothrae), for example, has an erect leaf canopy structure that
results in a dark image response (Everitt et al. 1987; Yang and Everitt 2010). Spotted
knapweed tends to inhibit the growth of other vegetation; the resulting increase in
visible bare soil can help identify places where spotted knapweed grows (Lass et al.
2002; Lawrence et al. 2006).
In summary, multispectral sensors that provide free and open access to global
imagery are used regularly for IAS detection in grasslands because their predefined
temporal resolution offers recurring overpasses and at the very least provides seasonal imagery. This supports time series analyses and multi-date classification techniques. Looking to the future, changes in grassland species composition are
anticipated to have the largest impact on Africa because it is home to the largest
savannas, which cover roughly 50% of the continent (Campbell 1996; Grace et al.
12 Remote Detection of Invasive Alien Species
West et al. 2017) and die-off (Boyte et al. 2015; Weisberg et al. 2017). Rather than
just using images selected during green up and senescence periods, extracting
phenology metrics from vegetation indices to refine cheatgrass classifications has
also been successful (Bradley and Mustard 2008); however, in arid and semiarid
environments, these indices can be highly influenced by rock and soil and should be
used with caution (Singh and Glenn 2009). Huang and Geiger (2008) showed that a
multi-date imaging approach can be successful even when natural phenologies of
natives and nonnatives are similar. For example, native grasses and Lehmann lovegrass
(Eragrostis lehmanniana) responded differently to unusual amounts of cool season
precipitation, which allowed new tissues in invasive species to grow, making the two
grasses distinguishable with multi-date imaging (Huang and Geiger 2008).
Imaging spectroscopy to map grassland IAS also often depends on differences in
phenology, but the higher spectral resolution and typically higher spatial resolution
afforded by airborne platforms often allow for more detailed and early detection
maps. Image acquisition timing is important for species that exhibit differences in
coloration throughout the year, such as flowering species or deciduous shrubs. In the
case of leafy spurge (Euphorbia esula), hyperspectral instruments are better
equipped to detect changes in flowering and thus have a higher success rate when
compared to multispectral instruments (Mitchell and Glenn 2009). Leafy spurge has
characteristic yellow flowers that bloom in early summer, and tamarisk leaves turn
from yellow-orange to orange-brown in autumn before leaf drop. This distinct pigmentation enables remote detection using both imaging spectroscopy (Williams and
Hunt Jr 2002; Glenn et al. 2005) and multispectral data (Anderson et al. 1993;
Everitt et al. 1995; Evangelista et al. 2009). The blue-green color of new stems and
the red-brown color of older stems help detection of spotted knapweed (Centaurea
maculosa) from imaging spectroscopy (Lass et al. 2002; Lawrence et al. 2006). For
early detection of goldenrod (Solidago altissima), an invasive moist tall grass in
Japan, hyperspectral images acquired during early spring before full development
of the grass canopy make it easier to map the exposed understory (Ishii and
Washitani 2013).
Differences in canopy architecture or plant morphological traits, such as plant
height and pubescence, can also be exploited when plants share similar phenologies
or imagery is unavailable when growth cycles show key differences. Broom snakeweed (Gutierrezia sarothrae), for example, has an erect leaf canopy structure that
results in a dark image response (Everitt et al. 1987; Yang and Everitt 2010). Spotted
knapweed tends to inhibit the growth of other vegetation; the resulting increase in
visible bare soil can help identify places where spotted knapweed grows (Lass et al.
2002; Lawrence et al. 2006).
In summary, multispectral sensors that provide free and open access to global
imagery are used regularly for IAS detection in grasslands because their predefined
temporal resolution offers recurring overpasses and at the very least provides seasonal imagery. This supports time series analyses and multi-date classification techniques. Looking to the future, changes in grassland species composition are
anticipated to have the largest impact on Africa because it is home to the largest
savannas, which cover roughly 50% of the continent (Campbell 1996; Grace et al.
12 Remote Detection of Invasive Alien Species