Certain vegetative land cover types (e.g., deciduous shrubs, deciduous trees,
grasslands) have distinct life cycles marked by the growth and senescence of
leaves and periods of enhanced photosynthetic activity. Where these land cover
types exist, the growth and senescence of leaves cause changes in the reflectance
of NIR and visible light from the land surface. For example, the emergence and
growth of new leaves are associated with increases in chlorophyll concentration
and leaf area, which increase the absorption of red wavelengths (Richardson et al.
2007) and the reflectance of NIR. During leaf senescence, the collapse of the
mesophyll reduces the proportion of reflected NIR (Knipling 1970; but see Castro
and Sanchez-Azofeifa 2008). In addition, declining chlorophyll concentrations and
de novo synthesis of anthocyanins in senescing leaves (Lee et al. 2003) increase
the reflectance of red wavelengths (Richardson et al. 2009a). The close correspondence between leaf growth and senescence and the reflectance of NIR and
visible light makes it possible to study vegetation phenology using reflectance
measurements acquired by remote sensors.
Annual time series of satellite-derived vegetation indices [e.g., normalized
difference vegetation index (NDVI), enhanced vegetation index (EVI) (Huete et al.
2002)] and biophysical metrics [e.g., leaf area index (LAI), fraction of absorbed
photosynthetically active radiation (FPAR) (Myneni et al. 2002)] that incorporate
the reflectance of NIR and red wavelengths generally capture the spectral changes
associated with leaf growth and senescence in the large areal units monitored by
the sensor. Therefore, these metrics can be used to document the phenology of the
land surface with variable accuracy and precision (White et al. 2009; Schwartz and
Hanes 2010). With this said, it must be noted that the degree to which these
satellite-derived metrics can be used to study LSP depends on the surface characteristics of the entire areal unit measured. Considering that the large areal units
measured from space can integrate a variety of land cover types (vegetative and
non-vegetative, deciduous and evergreen), there has been some effort to restrict the
analysis of LSP to pixels containing land cover types that exhibit distinct phenologies that are most observable from space [e.g., deciduous forests (Fisher et al.
2006)] and have a strong response to climate (White et al. 2005).
4.3 Methods
Remote sensing indices characterizing vegetation conditions form the basis of
deriving LSP metrics. The most ubiquitous algorithm is the Normalized Difference
Vegetation Index (NDVI), which utilizes chlorophyll and leaf structure-induced
reflectance contrast between red and near-infrared spectral bands from live vegetation (Rouse et al. 1974). Using the normalized difference [(NIR - R)/
(NIR ? R)] rather than quantifications of single bands effectively captures the
relative reflectance difference between the red and near-infrared bands. Specifics
regarding algorithms of this and the following vegetation indices discussed here
may be found in Chap. 3 of this book. The NDVI is able to effectively differentiate
4 Land Surface Phenology
103
grasslands) have distinct life cycles marked by the growth and senescence of
leaves and periods of enhanced photosynthetic activity. Where these land cover
types exist, the growth and senescence of leaves cause changes in the reflectance
of NIR and visible light from the land surface. For example, the emergence and
growth of new leaves are associated with increases in chlorophyll concentration
and leaf area, which increase the absorption of red wavelengths (Richardson et al.
2007) and the reflectance of NIR. During leaf senescence, the collapse of the
mesophyll reduces the proportion of reflected NIR (Knipling 1970; but see Castro
and Sanchez-Azofeifa 2008). In addition, declining chlorophyll concentrations and
de novo synthesis of anthocyanins in senescing leaves (Lee et al. 2003) increase
the reflectance of red wavelengths (Richardson et al. 2009a). The close correspondence between leaf growth and senescence and the reflectance of NIR and
visible light makes it possible to study vegetation phenology using reflectance
measurements acquired by remote sensors.
Annual time series of satellite-derived vegetation indices [e.g., normalized
difference vegetation index (NDVI), enhanced vegetation index (EVI) (Huete et al.
2002)] and biophysical metrics [e.g., leaf area index (LAI), fraction of absorbed
photosynthetically active radiation (FPAR) (Myneni et al. 2002)] that incorporate
the reflectance of NIR and red wavelengths generally capture the spectral changes
associated with leaf growth and senescence in the large areal units monitored by
the sensor. Therefore, these metrics can be used to document the phenology of the
land surface with variable accuracy and precision (White et al. 2009; Schwartz and
Hanes 2010). With this said, it must be noted that the degree to which these
satellite-derived metrics can be used to study LSP depends on the surface characteristics of the entire areal unit measured. Considering that the large areal units
measured from space can integrate a variety of land cover types (vegetative and
non-vegetative, deciduous and evergreen), there has been some effort to restrict the
analysis of LSP to pixels containing land cover types that exhibit distinct phenologies that are most observable from space [e.g., deciduous forests (Fisher et al.
2006)] and have a strong response to climate (White et al. 2005).
4.3 Methods
Remote sensing indices characterizing vegetation conditions form the basis of
deriving LSP metrics. The most ubiquitous algorithm is the Normalized Difference
Vegetation Index (NDVI), which utilizes chlorophyll and leaf structure-induced
reflectance contrast between red and near-infrared spectral bands from live vegetation (Rouse et al. 1974). Using the normalized difference [(NIR - R)/
(NIR ? R)] rather than quantifications of single bands effectively captures the
relative reflectance difference between the red and near-infrared bands. Specifics
regarding algorithms of this and the following vegetation indices discussed here
may be found in Chap. 3 of this book. The NDVI is able to effectively differentiate
4 Land Surface Phenology
103