properties (Nelson et al. 1988, 2004; Lim et al. 2003; Lim and Treitz 2004; and
García et al. 2010). However, lidar intensity or height combined with intensity data
provides better biomass estimate than height metrics alone (García et al. 2010;
Lim et al. 2003).
At individual tree level, biomass is estimated based on individual tree structure
parameters. The properties are located and measured using different tree segmentation approaches (Popescu and Wynne 2003, 2004). Specifically, the
parameters, including tree height, crown width, and tree locations, are derived
from CHM using a local maximum filtering approach with a variable circular
window to locate individual trees. Crown width is estimated using polynomial
fitting on two perpendicular vertical profiles through each identified tree crown.
The crown base height for each lidar-derived tree is calculated with the lidar
voxel-based approach (Popescu and Zhao 2008).
3.3.4.2 Large Footprint Full Waveform Lidar
Large footprint full-waveform systems can accurately estimate AGB in various
forest types. The commonly used airborne lidar data over the past decade are
collected from the Scanning Lidar Imager of Canopies by Echo Recovery (SLICER) with a *15 m footprint and the Laser Vegetation Imaging Sensor (LVIS)
with a *25 m footprint (Blair et al. 1999). These data have been successfully used
to retrieve AGB over various biomes across the US, which includes SLICER for
estimating AGB in Cascade Mountain Range in Oregon and Washington States
(Lefsky et al. 2005) and in Annapolis in Maryland (Lefsky et al. 1999a), and LVIS
for AGB in La Selva (Drake et al. 2002), White Mountain (Anderson et al. 2006;
Ni-Meister et al. 2010), Sierra Nevada in California (Hyde et al. 2005; Swatantran
et al., 2011), and Costa Rica (Drake et al. 2002, 2003). In most studies, stepwise
multiple regressions are adapted to predict ground-based measures of stand
structure from both conventional canopy structure indices (including mean and
maximum canopy surface height and canopy cover) and indices derived from CHP
(the height relative to the ground elevation, at which 100, 75, 50 and 25 %,
respectively, of the accumulated full waveform energy occurs) (Blair et al. 2004).
The spaceborne Geoscience Laser Altimeter System (GLAS), part of the Ice,
Cloud and land Elevation Satellite (ICESat) mission, provides global lidar data
with a variable diameter of *70 m footprint spaced at *170 m (Zwally et al.
2002; Harding et al. 2005; Lefsky et al. 2005). GLAS is a full waveform digitizing
lidar system that acquires information of topography and vertical vegetation
structure (Zwally et al. 2002; Harding and Carabajal 2005). The GLAS data have
been applied for estimating forest biomass on ground plots in tropical, temperate
and conifer forests (Boudreau et al. 2008; Lefsky et al. 2005, 2007; and Nelson
et al. 2009). One major limitation of current spaceborne lidar systems is the lack of
imaging capabilities, which only provides sparse sampling information on the
forest structure. To overcome this problem, it has been fused with other data to
82
X. Zhang and W. Ni-meister
García et al. 2010). However, lidar intensity or height combined with intensity data
provides better biomass estimate than height metrics alone (García et al. 2010;
Lim et al. 2003).
At individual tree level, biomass is estimated based on individual tree structure
parameters. The properties are located and measured using different tree segmentation approaches (Popescu and Wynne 2003, 2004). Specifically, the
parameters, including tree height, crown width, and tree locations, are derived
from CHM using a local maximum filtering approach with a variable circular
window to locate individual trees. Crown width is estimated using polynomial
fitting on two perpendicular vertical profiles through each identified tree crown.
The crown base height for each lidar-derived tree is calculated with the lidar
voxel-based approach (Popescu and Zhao 2008).
3.3.4.2 Large Footprint Full Waveform Lidar
Large footprint full-waveform systems can accurately estimate AGB in various
forest types. The commonly used airborne lidar data over the past decade are
collected from the Scanning Lidar Imager of Canopies by Echo Recovery (SLICER) with a *15 m footprint and the Laser Vegetation Imaging Sensor (LVIS)
with a *25 m footprint (Blair et al. 1999). These data have been successfully used
to retrieve AGB over various biomes across the US, which includes SLICER for
estimating AGB in Cascade Mountain Range in Oregon and Washington States
(Lefsky et al. 2005) and in Annapolis in Maryland (Lefsky et al. 1999a), and LVIS
for AGB in La Selva (Drake et al. 2002), White Mountain (Anderson et al. 2006;
Ni-Meister et al. 2010), Sierra Nevada in California (Hyde et al. 2005; Swatantran
et al., 2011), and Costa Rica (Drake et al. 2002, 2003). In most studies, stepwise
multiple regressions are adapted to predict ground-based measures of stand
structure from both conventional canopy structure indices (including mean and
maximum canopy surface height and canopy cover) and indices derived from CHP
(the height relative to the ground elevation, at which 100, 75, 50 and 25 %,
respectively, of the accumulated full waveform energy occurs) (Blair et al. 2004).
The spaceborne Geoscience Laser Altimeter System (GLAS), part of the Ice,
Cloud and land Elevation Satellite (ICESat) mission, provides global lidar data
with a variable diameter of *70 m footprint spaced at *170 m (Zwally et al.
2002; Harding et al. 2005; Lefsky et al. 2005). GLAS is a full waveform digitizing
lidar system that acquires information of topography and vertical vegetation
structure (Zwally et al. 2002; Harding and Carabajal 2005). The GLAS data have
been applied for estimating forest biomass on ground plots in tropical, temperate
and conifer forests (Boudreau et al. 2008; Lefsky et al. 2005, 2007; and Nelson
et al. 2009). One major limitation of current spaceborne lidar systems is the lack of
imaging capabilities, which only provides sparse sampling information on the
forest structure. To overcome this problem, it has been fused with other data to
82
X. Zhang and W. Ni-meister