REGIONALLY APPLICABLE EQUATIONS FOR ESTIMATING ABOVEGROUND BIOMASS IN THE AMAZON BASIN FROM GEOSCIENCE LASER ALTIMETER SYSTEM WAVEFORM METRICS

Michael
A.
Lefsky, Center for Ecological Applications of Lidar, Colorado State, University, lefsky@cnr.colostate.edu
Maria
Hunter, Complex Systems Research Center, University of New, Hampshire Michael
Keller, USDA Forest Service, International Institute of Tropical Forestry, and, Complex Systems Research Center, University of New Hampshire David
Turner, Department of Forest Science, Oregon State Univerisity, david.turner@orst.edu
(Presenting)
Plinio
B.
de Camargo, Lab. De Ecologia Isotopica, CENA/USP

Lidar remote sensing has a unique capability for estimating forest
canopy height, which has a direct and increasingly well-understood
relationship to aboveground carbon storage. In our LBA work, we are
using lidar waveforms from the Geoscience Laser Altimeter System
(GLAS) to estimate canopy height and aboveground forest carbon storage
for each GLAS waveform in the Amazon basin and combining those point
estimates with images from the MODIS sensor to develop spatially
continuous datasets. We will compare the new biomass estimates with
spatially explicit data layers from previous studies in the region and
use an existing bookkeeping C flux model to evaluate the sensitivity
of C flux estimates associated with land cover change in the region to
the new biomass estimates.
Three problems must be addressed to accurately estimate aboveground
carbon storage for each GLAS waveform. First, equations representing
the waveforms to the physical structure of the forest must be
developed, so that estimates of forest canopy height can be made.
Next, an equation relating forest canopy height to field estimates of
aboveground biomass must be developed. We have previously published on
these two topics for field plots in the Tapajos National Forest, near
Santarem, Brazil. We have extended this analysis to two other sets of
field sites in the vicinity of Manaus and Tanguro, Brazil. These sites
were picked to increase the range of variability in climate that we
considered. Simple regression analysis using SRTM slope indices and
GLAS waveform height metrics indicates that a single equation
estimates maximum canopy height at all three sites with 87% of
variance explained, and RMSE of 6.8 m, and no statistically
significant biases at any site. Similarly, there is a single equation
that relates maximum canopy height to field estimated aboveground
biomass that fits all three sites. Combining the two, a single
equation relating GLAS estimates of maximum height and field estimated
aboveground biomass has been created which explains 72% of variance,
and an RMSE of 54.7 Mgha-1. At one site (Manaus) there exists a
statistically significant bias (+ 31.3 Mgha-1 or 12.5% of the site
mean) in the estimates of aboveground biomass.