293
were higher than 75% for both; the potential for spectral separability was maximized by the choice of wavelength regions, and the researchers were able to increase
accuracy using a random forest classifier, due to higher capability of under-canopy
detection (Chance et al. 2016b). Because urban ecosystems are smaller than other
ecosystems and more complex, high spatial resolution is necessary to detect IAS
within the mostly non-vegetative ground cover.
Lidar with spectral data also has proven effective for mapping vegetation within
urban areas. By combining lidar with hyperspectral imaging and a random forest
classifier to map tree species including honey locust (Gleditsia triacanthos) in
Surrey, British Columbia, Canada. Liu et al. (2017) further improved classification
accuracy, showing the power of data fusion. Other studies have combined lidar data
with IKONOS multispectral data to detect whether Chinese privet (Ligustrum
sinense) invasion changed urban forest structure in Charlotte, North Carolina, USA
(Singh et al. 2015). These researchers also found that a random forest built with
lidar-derived metrics produced the best results.
RS of urban IAS, however, has some unique challenges. Because most of the
ground is covered by manmade features, it is difficult to detect green areas and map
and identify individual species (Alonzo et al. 2014). With sufficient spatial resolution, these challenges can be overcome. The most successful approach to date is to
use a combination of hyperspectral and lidar, which yields spectral, structural, and
height information.
In summary, detection of IAS in urban environments requires high spatial resolution to differentiate natural from built environments, high spectral resolution to
identify species, and sufficient temporal resolution to detect IAS at different stages
of invasion. While this is an emerging field with a growing literature, relatively few
studies of IAS in urban environments have used RS data, and further research is
needed in different geographical settings, invasion process phases, and urban density conditions.
12.3 Summary, Conclusions, and Prospectus
Invasive species are a major direct driver of biodiversity loss because they outcompete native species for local resources, eventually replacing or displacing them.
They also cause indirect losses because they do not assume all of the ecological
roles of the replaced native species. As they spread, IAS modify nutrient availability, nutrient cycling, soil chemistry, water quality, hydrology, food webs, habitats,
and other ecosystem functions (Gordon 1998; Scheffer et al. 2003; Dukes and
Mooney 2004; Hestir et al. 2016; Khanna et al. 2018), impairing ecosystem function. In addition to causing functional changes, IAS also modify ecosystem structure by physically changing canopy structures in forests and water quality in aquatic
ecosystems. Increasing global changes related to climate, nutrient cycles, and land
use will potentially change transport and introduction mechanisms of IAS in a way
that provides a competitive advantage for new IAS, likely reducing effectiveness of
12 Remote Detection of Invasive Alien Species
were higher than 75% for both; the potential for spectral separability was maximized by the choice of wavelength regions, and the researchers were able to increase
accuracy using a random forest classifier, due to higher capability of under-canopy
detection (Chance et al. 2016b). Because urban ecosystems are smaller than other
ecosystems and more complex, high spatial resolution is necessary to detect IAS
within the mostly non-vegetative ground cover.
Lidar with spectral data also has proven effective for mapping vegetation within
urban areas. By combining lidar with hyperspectral imaging and a random forest
classifier to map tree species including honey locust (Gleditsia triacanthos) in
Surrey, British Columbia, Canada. Liu et al. (2017) further improved classification
accuracy, showing the power of data fusion. Other studies have combined lidar data
with IKONOS multispectral data to detect whether Chinese privet (Ligustrum
sinense) invasion changed urban forest structure in Charlotte, North Carolina, USA
(Singh et al. 2015). These researchers also found that a random forest built with
lidar-derived metrics produced the best results.
RS of urban IAS, however, has some unique challenges. Because most of the
ground is covered by manmade features, it is difficult to detect green areas and map
and identify individual species (Alonzo et al. 2014). With sufficient spatial resolution, these challenges can be overcome. The most successful approach to date is to
use a combination of hyperspectral and lidar, which yields spectral, structural, and
height information.
In summary, detection of IAS in urban environments requires high spatial resolution to differentiate natural from built environments, high spectral resolution to
identify species, and sufficient temporal resolution to detect IAS at different stages
of invasion. While this is an emerging field with a growing literature, relatively few
studies of IAS in urban environments have used RS data, and further research is
needed in different geographical settings, invasion process phases, and urban density conditions.
12.3 Summary, Conclusions, and Prospectus
Invasive species are a major direct driver of biodiversity loss because they outcompete native species for local resources, eventually replacing or displacing them.
They also cause indirect losses because they do not assume all of the ecological
roles of the replaced native species. As they spread, IAS modify nutrient availability, nutrient cycling, soil chemistry, water quality, hydrology, food webs, habitats,
and other ecosystem functions (Gordon 1998; Scheffer et al. 2003; Dukes and
Mooney 2004; Hestir et al. 2016; Khanna et al. 2018), impairing ecosystem function. In addition to causing functional changes, IAS also modify ecosystem structure by physically changing canopy structures in forests and water quality in aquatic
ecosystems. Increasing global changes related to climate, nutrient cycles, and land
use will potentially change transport and introduction mechanisms of IAS in a way
that provides a competitive advantage for new IAS, likely reducing effectiveness of
12 Remote Detection of Invasive Alien Species