(Lymantria dispar) damage to an Appalachian oak forest using stream nitrogen levels
as a proxy for defoliation because nitrogen is released from damaged leaves into the
watershed. The FDI was able to predict both the overall intensity and extent of forest
damage; however, nearly complete defoliation was needed for the FDI to detect
damage at the stand level (Healey et al., 2005).
High-resolution hyperspectral imagery has also been used to detect insect damage.
Leckie et al. (2005) developed an automated procedure for detecting forest defoliation
by the jack pine beetle (Dendroctonus frontalis) using airborne, 2.5-m-resolution
hyperspectral imagery from the multispectral electro-optical imaging sensor. Information from six visible bands and one NIR band was combined to detect canopy
discoloration due to beetle damage. Insect damage was detected with an overall
accuracy of 84% in stands with damage levels ranging from light to heavy. Martin
et al. (2008) used foliar nitrogen levels as a proxy for forest damage in 137 forested
plots in North America, South America, and Australia. Most plant nitrogen is
contained in pigments and enzymes in leaf chloroplasts, and defoliation results in
a marked drop in both canopy nitrogen and chlorophyll, which can be detected
spectrally. Imagery from the airborne visible/infrared imaging spectrometer accurately predicted canopy nitrogen levels measured by ground sampling. The results
were valid regardless of the source of forest damage. However, high-resolution,
hyperspectral imagery is not currently available for routine forest monitoring.
Monitoring the effects of insects on vegetation presents a challenge for remote
sensing methods due to the variety of spatial and temporal scales at which damage is
manifested. Rather than relying on a single remote sensing platform, as in the above
studies, our approach was to collect data across several scales of measurements in an
attempt to combine the strengths of each approach and minimize their individual
weaknesses. For example, digital cameras provide very high spatial and temporal
resolution over a very limited area, but they can be scaled over a wider area using
Landsat imagery, while MODIS imagery can provide high temporal resolution once
stand structure is mapped with higher resolution imagery.
5.2.2 Background on Tamarix and Tamarix Leaf Beetles
on Western U.S. Rivers
Tamarix shrubs were introduced on western rivers as erosion control plants in the
nineteenth century (Chew, 2009). They spread rapidly and now occupy several
hundred thousand hectares of riparian habitat (Nagler et al., 2010a). Starting in the
1950s they were perceived as problem plants due to assumed high water use, low
habitat value for birds, and competitive displacement of native trees (DiTomaso,
1998; Zavaleta, 2000). These defects have been challenged by more recent research
(e.g., Stromberg et al., 2009; Hultine et al., 2009, 2010; Nagler et al., 2010a,b), but in
the meantime, Tamarix leaf beetles have been released on western rivers starting in
2001 as biocontrol agents (Bean et al., 2007a,b; DeLoach and Carruthers, 2004;
DeLoach et al., 2000, 2004; Lewis et al., 2003; Dudley et al., 2006; Moran et al.,
2009), and they have also spread rapidly, raising new management issues in already
stressed western riparian corridors (Hultine et al., 2010). Among these are the effects
84
CHANGE DETECTION USING VEGETATION INDICES AND MULTIPLATFORM
as a proxy for defoliation because nitrogen is released from damaged leaves into the
watershed. The FDI was able to predict both the overall intensity and extent of forest
damage; however, nearly complete defoliation was needed for the FDI to detect
damage at the stand level (Healey et al., 2005).
High-resolution hyperspectral imagery has also been used to detect insect damage.
Leckie et al. (2005) developed an automated procedure for detecting forest defoliation
by the jack pine beetle (Dendroctonus frontalis) using airborne, 2.5-m-resolution
hyperspectral imagery from the multispectral electro-optical imaging sensor. Information from six visible bands and one NIR band was combined to detect canopy
discoloration due to beetle damage. Insect damage was detected with an overall
accuracy of 84% in stands with damage levels ranging from light to heavy. Martin
et al. (2008) used foliar nitrogen levels as a proxy for forest damage in 137 forested
plots in North America, South America, and Australia. Most plant nitrogen is
contained in pigments and enzymes in leaf chloroplasts, and defoliation results in
a marked drop in both canopy nitrogen and chlorophyll, which can be detected
spectrally. Imagery from the airborne visible/infrared imaging spectrometer accurately predicted canopy nitrogen levels measured by ground sampling. The results
were valid regardless of the source of forest damage. However, high-resolution,
hyperspectral imagery is not currently available for routine forest monitoring.
Monitoring the effects of insects on vegetation presents a challenge for remote
sensing methods due to the variety of spatial and temporal scales at which damage is
manifested. Rather than relying on a single remote sensing platform, as in the above
studies, our approach was to collect data across several scales of measurements in an
attempt to combine the strengths of each approach and minimize their individual
weaknesses. For example, digital cameras provide very high spatial and temporal
resolution over a very limited area, but they can be scaled over a wider area using
Landsat imagery, while MODIS imagery can provide high temporal resolution once
stand structure is mapped with higher resolution imagery.
5.2.2 Background on Tamarix and Tamarix Leaf Beetles
on Western U.S. Rivers
Tamarix shrubs were introduced on western rivers as erosion control plants in the
nineteenth century (Chew, 2009). They spread rapidly and now occupy several
hundred thousand hectares of riparian habitat (Nagler et al., 2010a). Starting in the
1950s they were perceived as problem plants due to assumed high water use, low
habitat value for birds, and competitive displacement of native trees (DiTomaso,
1998; Zavaleta, 2000). These defects have been challenged by more recent research
(e.g., Stromberg et al., 2009; Hultine et al., 2009, 2010; Nagler et al., 2010a,b), but in
the meantime, Tamarix leaf beetles have been released on western rivers starting in
2001 as biocontrol agents (Bean et al., 2007a,b; DeLoach and Carruthers, 2004;
DeLoach et al., 2000, 2004; Lewis et al., 2003; Dudley et al., 2006; Moran et al.,
2009), and they have also spread rapidly, raising new management issues in already
stressed western riparian corridors (Hultine et al., 2010). Among these are the effects
84
CHANGE DETECTION USING VEGETATION INDICES AND MULTIPLATFORM