HR: 0830h
AN: B51E-1015 [PDF]
TI: Mapping Forest Biomass Over Large Scales From Remote
Sensing, Topographic and Climate Data
AU: * Baccini, A
EM: abaccini@bu.edu
AF: Boston University, Department of Geography and Center for Remote Sensing, 675 Commonwealth Avenue,
Boston, MA 02215
AU: Friedl, M
EM: friedl@bu.edu
AF: Boston University, Department of Geography and Center for Remote Sensing, 675 Commonwealth Avenue,
Boston, MA 02215
AU: Woodcock, C
EM: curtis@bu.edu
AF: Boston University, Department of Geography and Center for Remote Sensing, 675 Commonwealth Avenue,
Boston, MA 02215
AU: Warbington, R
EM: rjwarbington@fs.fed.us
AF: Forest Service Remote Sensing Laboratory, Region 5, 1920 20th Street, Sacramento, CA 95814
AB:
To better understand dynamics in the global carbon cycle, improved
methods are required to quantify the carbon stored in forest
ecosystems over large areas. Remote sensing provides an obvious
means for doing this, but robust and effective methods and data
sources have proven elusive. We used a combination of remotely
sensed data, topographic data, climate variables, and statistical
methods to map above ground forest biomass for 89,000 km$^{2}$ of
National Forest land in California. To do this, 1-km Nadir BRDF
adjusted reflectances derived from the Moderate Resolution Imaging
Spectoradiometer (MODIS) were used in combination with
precipitation, temperature, and elevation data. Using these data,
tree-based models were estimated to predict biomass within each
1-km cell. The models were validated using high resolution forest
cover biomass maps produced by the United States Forest Service
for the State of California. The results show that the tree-based
models were able to predict forest biomass for land areas
characterized by shrubs, hardwood forest, and conifer forest with
a root mean square error of 44.4 tons/ha. Across all 89,000
km$^{2}$ (the whole study area) the models explain 73 percent of
the variance in biomass. These results suggest that coarse
resolution remotely sensed data, in combination with relevant
topographic and climate data, can be used to map above ground
biomass with reasonable accuracy over large areas and improve our
ability to estimate stocks of carbon in existing vegetation.
DE: 0400 Biogeosciences
DE: 1640 Remote sensing
SC: Biogeosciences [B]
MN: 2003 Fall Meeting