The present review uses a case study approach to illustrate some of the opportunities now available for combining observations across scales, drawing on our own
research over the past two decades. The review will explore the use of multiplatform
sensor systems to characterize ecological change, as exemplified by efforts to scale the
effects of a biocontrol insect (the leaf beetle Diorhabda carinulata) on the phenology
and water use of Tamarix shrubs (Tamarix ramosissima and related species and
hybrids) targeted for removal on western U.S. rivers, from the level of individual
leaves to the regional level of measurement. A final section will summarize the
lessons learned and will emphasize the need for ground data to calibrate and validate
remote sensing data and the types of errors inherent in scaling point data over
wide areas, illustrated with research on evapotranspiration (ET) of Tamarix using a
wide range of ground measurement and remote sensing methods. The goal of the
review is to describe emerging methods for using satellite imagery across temporal
and spatial scales in enough detail that they can be applied in other landscape studies.
5.2 COMBINING PHENOCAMS, LANDSAT, AND MODIS IMAGERY
TO MONITOR EFFECTS OF INSECT DEFOLIATION OF
VEGETATION ACROSS SPATIAL AND TEMPORAL SCALES
5.2.1 Need for Multiplatform Methods in Detecting Insect Damage
to Forests
Insect infestations cause widespread damage to forest ecosystems, and the damage is
expected to increase in areas such as the southwestern United States where climates
are expected to become warmer and drier (Dale et al., 2001; Seager et al., 2007). Insect
effects appear at multiple spatial scales, from effects on individual plants to regional
effects. Furthermore, insect infestation is often episodic, occurring over just a few
weeks, so monitoring programs must have high temporal resolution as well cover a
wide range of spatial scales. Wulder et al. (2006) pointed out that since forest
management agencies require information at several scales of measurement, no single
remote sensing approach is adequate. At the low-resolution end of the scale, Eklundh
et al. (2009) tested a method for mapping Scots pine defoliation by the pine sawfly
(Neodiprion spp.) in Norway using coarse-resolution 16-day composite imagery from
the MODIS sensors on the Terra satellite. They point out that the normalized
difference vegetation index (NDVI) values were useful in detecting areas of defoliation within the forest, but only weak relations were found between the degree of
damage and MODIS change parameters. They concluded that MODIS could be used
to detect damaged forest areas but that high-resolution imagery or fieldwork would be
needed to estimate the intensity of the infestation.
Medium-resolution Landsat imagery has also been used to detect insect damage.
Healey et al. (2005) combined six Landsat bands into a three-parameter forest
disturbance index (FDI) using a tasseled cap transformation in which brightness,
greenness, and wetness were evaluated by different band combinations. Areas of
forest clear-cut were characterized by high brightness and low greenness and wetness
values. Eshleman et al. (2009) tested the ability of the FDI to detect gypsy moth
COMBINING PHENOCAMS, LANDSAT, AND MODIS IMAGERY
83
research over the past two decades. The review will explore the use of multiplatform
sensor systems to characterize ecological change, as exemplified by efforts to scale the
effects of a biocontrol insect (the leaf beetle Diorhabda carinulata) on the phenology
and water use of Tamarix shrubs (Tamarix ramosissima and related species and
hybrids) targeted for removal on western U.S. rivers, from the level of individual
leaves to the regional level of measurement. A final section will summarize the
lessons learned and will emphasize the need for ground data to calibrate and validate
remote sensing data and the types of errors inherent in scaling point data over
wide areas, illustrated with research on evapotranspiration (ET) of Tamarix using a
wide range of ground measurement and remote sensing methods. The goal of the
review is to describe emerging methods for using satellite imagery across temporal
and spatial scales in enough detail that they can be applied in other landscape studies.
5.2 COMBINING PHENOCAMS, LANDSAT, AND MODIS IMAGERY
TO MONITOR EFFECTS OF INSECT DEFOLIATION OF
VEGETATION ACROSS SPATIAL AND TEMPORAL SCALES
5.2.1 Need for Multiplatform Methods in Detecting Insect Damage
to Forests
Insect infestations cause widespread damage to forest ecosystems, and the damage is
expected to increase in areas such as the southwestern United States where climates
are expected to become warmer and drier (Dale et al., 2001; Seager et al., 2007). Insect
effects appear at multiple spatial scales, from effects on individual plants to regional
effects. Furthermore, insect infestation is often episodic, occurring over just a few
weeks, so monitoring programs must have high temporal resolution as well cover a
wide range of spatial scales. Wulder et al. (2006) pointed out that since forest
management agencies require information at several scales of measurement, no single
remote sensing approach is adequate. At the low-resolution end of the scale, Eklundh
et al. (2009) tested a method for mapping Scots pine defoliation by the pine sawfly
(Neodiprion spp.) in Norway using coarse-resolution 16-day composite imagery from
the MODIS sensors on the Terra satellite. They point out that the normalized
difference vegetation index (NDVI) values were useful in detecting areas of defoliation within the forest, but only weak relations were found between the degree of
damage and MODIS change parameters. They concluded that MODIS could be used
to detect damaged forest areas but that high-resolution imagery or fieldwork would be
needed to estimate the intensity of the infestation.
Medium-resolution Landsat imagery has also been used to detect insect damage.
Healey et al. (2005) combined six Landsat bands into a three-parameter forest
disturbance index (FDI) using a tasseled cap transformation in which brightness,
greenness, and wetness were evaluated by different band combinations. Areas of
forest clear-cut were characterized by high brightness and low greenness and wetness
values. Eshleman et al. (2009) tested the ability of the FDI to detect gypsy moth
COMBINING PHENOCAMS, LANDSAT, AND MODIS IMAGERY
83