near-surface remote sensors (Richardson et al. 2007, 2009a; Graham et al. 2010;
Hufkens et al. 2012). In particular, networked webcams have been used to record
repeat canopy phenology in association with the existing eddy covariance flux
tower sites (cf. PhenoCam program, http://phenocam.sr.unh.edu/webcam/). Webcam-based phenology observation has the advantages of being more objective
(free from observer biases) and low-cost in logistics, which allows for consistent
and continuous monitoring of forest canopy conditions (Sonnentag et al. 2012).
The tradeoff is that it provides optical signals akin to satellite remote sensing and
does not give particular details of phenological development (e.g., bud break) that
traditional observer-based protocols offer.
In particular, Huemmrich et al. (1999) and Richardson et al. (2007) employed a
broad-band NDVI measured from flux tower-based radiometric measurements.
The broad-band NDVI utilizes the entire visible spectrum ([NIR-VIS]/[NIR ? -
VIS]) instead of the red band as is typical for satellite-derived NDVI. However,
more widely available near-surface remote sensing analyses have relied on visible
light digital camera measurements. The visible light digital photos contain separate
color bands (Red [R], Green [G], and Blue [B] respectively), which were used to
derive band algebra-based greenness indices, such as the excess green (2G-R-B)
and green chromatic coordinate (G/[R ? G ? B]; Sonnentag et al. 2012). Hufkens
et al. (2012) performed a comparison of near-surface remote sensing-based phenology with satellite (i.e. MODIS) remote sensing-based phenology at four PhenoCam sites across the United States. Results from the study showed relatively
consistent correspondence between the MODIS VI time series and camera excess
green time series and suggested that the mismatch of camera field of view and
satellite pixel-covered areas may contribute the major uncertainty in linking the
two types of phenological measurements. Related studies also utilized webcam
data to compare with MODIS-based LSP (Graham et al. 2010) and Landsat-based
LSP (Elmore et al. 2012). The continuous monitoring of near-surface remote
sensing also provided opportunities for validation in both the spring and autumn
seasons (Hufkens et al. 2012; Elmore et al. 2012). Elmore et al. (2012) utilized a
different index ([G-R]/[G ? R]) other than the excess green utilized in other
studies for deriving phenological information and attempted to use aerial photos to
compare with Landsat-based autumn phenology. In summary, the webcam phenology data provide a valuable source of in situ observation for validating LSP.
With the improvement of technology and data analyzing techniques, we anticipate
the ground-based radiometric measurements of tree canopy phenology will provide more insights on the links between phenological processes of plants and
corresponding remote sensing indices.
Recent validation efforts also include a study utilizing extensive data from the
Canadian phenology network (PlantWatch) and AVHRR and MERIS-integrated
LSP (Pouliot et al. 2011). Given that high resolution intensive phenology data are
only available at specific study sites, the growing extensive ground phenology data
from crowd-sourcing observation networks, such as PlantWatch (http://
www.naturewatch.ca) and the USA National Phenology Network (http://
www.usanpn.org), with real-time forest canopy monitoring using near-surface
114
J. M. Hanes et al.
Hufkens et al. 2012). In particular, networked webcams have been used to record
repeat canopy phenology in association with the existing eddy covariance flux
tower sites (cf. PhenoCam program, http://phenocam.sr.unh.edu/webcam/). Webcam-based phenology observation has the advantages of being more objective
(free from observer biases) and low-cost in logistics, which allows for consistent
and continuous monitoring of forest canopy conditions (Sonnentag et al. 2012).
The tradeoff is that it provides optical signals akin to satellite remote sensing and
does not give particular details of phenological development (e.g., bud break) that
traditional observer-based protocols offer.
In particular, Huemmrich et al. (1999) and Richardson et al. (2007) employed a
broad-band NDVI measured from flux tower-based radiometric measurements.
The broad-band NDVI utilizes the entire visible spectrum ([NIR-VIS]/[NIR ? -
VIS]) instead of the red band as is typical for satellite-derived NDVI. However,
more widely available near-surface remote sensing analyses have relied on visible
light digital camera measurements. The visible light digital photos contain separate
color bands (Red [R], Green [G], and Blue [B] respectively), which were used to
derive band algebra-based greenness indices, such as the excess green (2G-R-B)
and green chromatic coordinate (G/[R ? G ? B]; Sonnentag et al. 2012). Hufkens
et al. (2012) performed a comparison of near-surface remote sensing-based phenology with satellite (i.e. MODIS) remote sensing-based phenology at four PhenoCam sites across the United States. Results from the study showed relatively
consistent correspondence between the MODIS VI time series and camera excess
green time series and suggested that the mismatch of camera field of view and
satellite pixel-covered areas may contribute the major uncertainty in linking the
two types of phenological measurements. Related studies also utilized webcam
data to compare with MODIS-based LSP (Graham et al. 2010) and Landsat-based
LSP (Elmore et al. 2012). The continuous monitoring of near-surface remote
sensing also provided opportunities for validation in both the spring and autumn
seasons (Hufkens et al. 2012; Elmore et al. 2012). Elmore et al. (2012) utilized a
different index ([G-R]/[G ? R]) other than the excess green utilized in other
studies for deriving phenological information and attempted to use aerial photos to
compare with Landsat-based autumn phenology. In summary, the webcam phenology data provide a valuable source of in situ observation for validating LSP.
With the improvement of technology and data analyzing techniques, we anticipate
the ground-based radiometric measurements of tree canopy phenology will provide more insights on the links between phenological processes of plants and
corresponding remote sensing indices.
Recent validation efforts also include a study utilizing extensive data from the
Canadian phenology network (PlantWatch) and AVHRR and MERIS-integrated
LSP (Pouliot et al. 2011). Given that high resolution intensive phenology data are
only available at specific study sites, the growing extensive ground phenology data
from crowd-sourcing observation networks, such as PlantWatch (http://
www.naturewatch.ca) and the USA National Phenology Network (http://
www.usanpn.org), with real-time forest canopy monitoring using near-surface
114
J. M. Hanes et al.