map AGB at large scales. For example, Boudreau et al. (2008) and Nelson et al.
(2009) used a multiphase sampling approach to relate GLAS waveforms to airborne profiling lidar measurements which relate to field AGB estimates. Another
issue is that the lidar waveform mixes lidar energy returns from both vegetation
and underneath topography. The impact of underneath topography needs to be
removed using waveform shapes (Lefsky et al. 2005, 2007) and other physical
approaches (Yang et al. 2011).
3.3.5 Biomass from Multisensors
Data fusion techniques combine data from multiple sensors and related information from associated databases to achieve improved accuracy. In essence, the
methods statistically combine or fuse information from multiple sensors to take
advantage of the highly detailed vertical measurements provided by lidar and the
broad scale mapping capabilities of horizontal and vertical structure afforded by
radar and passive optical remote sensing data.
The fusion methods can lie on employing an approach to stratify field plots with
lidar samples of AGB with environmental controls through a stratification or
regression approach for scaling up to regions outside of lidar coverage. Stratification of a region is employed by vegetation type, topography and other environmental datasets measured by passive optical and radar remote sensing data
(Saatchi et al. 2007a, b, 2011). These methods make it possible to map forest
structure and biomass at intermediate scales. Currently, data fusion methods of two
or more sensors are based primarily on empirical analyses (Hyde et al. 2007;
Walker et al. 2007) although it is suggested to develop physical-based models as a
needed advancement.
Most recently, a multisensor dataset has been used to produce a high-resolution
‘‘National Biomass and Carbon Dataset for the year 2000 (NBCD2000)’’
(Kellndorfer et al. 2010, 2013). This dataset includes baseline estimates of basal
area-weighted canopy height, aboveground live dry biomass, and standing carbon
stock for CONUS at a 30 m spatial resolution. The dataset was developed using an
empirical modeling approach that combines USFS FIA data with high-resolution
InSAR data acquired from the 2000 Shuttle Radar Topography Mission (SRTM) and
optical remote sensing data acquired from the Landsat ETM+ sensor. Both the USGS
National Land Cover Dataset 2001(landcover and canopy density) (NLCD 2001)
and the existing vegetation type from the LANDFIRE project as well as topographic
information from the USGS National Elevation Dataset (NED) were used as spatial
predictor layers for canopy height and biomass estimation. The USFS FIA canopy
height and biomass were used in model development and validation.
Multisensor data have also been applied to generate a ‘‘benchmark’’ map of
biomass carbon stocks over 2.5 billion ha of forests on three tropical continents
(Saatchi et al. 2007a, b, 2011). The map was developed through a combination of
data from inventory plots, samples of forest structure from ICESat GLAS in
3 Remote Sensing of Forest Biomass
83
(2009) used a multiphase sampling approach to relate GLAS waveforms to airborne profiling lidar measurements which relate to field AGB estimates. Another
issue is that the lidar waveform mixes lidar energy returns from both vegetation
and underneath topography. The impact of underneath topography needs to be
removed using waveform shapes (Lefsky et al. 2005, 2007) and other physical
approaches (Yang et al. 2011).
3.3.5 Biomass from Multisensors
Data fusion techniques combine data from multiple sensors and related information from associated databases to achieve improved accuracy. In essence, the
methods statistically combine or fuse information from multiple sensors to take
advantage of the highly detailed vertical measurements provided by lidar and the
broad scale mapping capabilities of horizontal and vertical structure afforded by
radar and passive optical remote sensing data.
The fusion methods can lie on employing an approach to stratify field plots with
lidar samples of AGB with environmental controls through a stratification or
regression approach for scaling up to regions outside of lidar coverage. Stratification of a region is employed by vegetation type, topography and other environmental datasets measured by passive optical and radar remote sensing data
(Saatchi et al. 2007a, b, 2011). These methods make it possible to map forest
structure and biomass at intermediate scales. Currently, data fusion methods of two
or more sensors are based primarily on empirical analyses (Hyde et al. 2007;
Walker et al. 2007) although it is suggested to develop physical-based models as a
needed advancement.
Most recently, a multisensor dataset has been used to produce a high-resolution
‘‘National Biomass and Carbon Dataset for the year 2000 (NBCD2000)’’
(Kellndorfer et al. 2010, 2013). This dataset includes baseline estimates of basal
area-weighted canopy height, aboveground live dry biomass, and standing carbon
stock for CONUS at a 30 m spatial resolution. The dataset was developed using an
empirical modeling approach that combines USFS FIA data with high-resolution
InSAR data acquired from the 2000 Shuttle Radar Topography Mission (SRTM) and
optical remote sensing data acquired from the Landsat ETM+ sensor. Both the USGS
National Land Cover Dataset 2001(landcover and canopy density) (NLCD 2001)
and the existing vegetation type from the LANDFIRE project as well as topographic
information from the USGS National Elevation Dataset (NED) were used as spatial
predictor layers for canopy height and biomass estimation. The USFS FIA canopy
height and biomass were used in model development and validation.
Multisensor data have also been applied to generate a ‘‘benchmark’’ map of
biomass carbon stocks over 2.5 billion ha of forests on three tropical continents
(Saatchi et al. 2007a, b, 2011). The map was developed through a combination of
data from inventory plots, samples of forest structure from ICESat GLAS in
3 Remote Sensing of Forest Biomass
83