HR: 08:45h
AN: B41C-03    [Abstracts]
TI: Forest Gap Formation Dynamics and Carbon Balance at the Landscape Scale
AU: * Chambers, J Q
EM: chambers@tulane.edu
AF: Tulane University, Ecol Evol Biology 310 Dinwiddie Hall, New Orleans, LA 70118 United States
AU: * Chambers, J Q
EM: chambers@tulane.edu
AF: INPA, CP 478, Manaus, AM 69011-970 Brazil
AU: Santos, J d
EM: joca@inpa.gov.br
AF: INPA, CP 478, Manaus, AM 69011-970 Brazil
AU: Higuchi, N
EM: niro@inpa.gov.br
AF: INPA, CP 478, Manaus, AM 69011-970 Brazil
AB: The carbon balance of landscapes is critically dependent on the frequency and spatial distribution of large canopy gaps ($>$ 400 m$^{2}$) which are difficult to quantify using traditional forest inventory methods. For this project, both modeling and remote sensing investigations were employed to better understand how carbon cycling in Amazon forests varies with large gap dynamics. A site specific model was parameterized based on recent studies from across the entire Amazon basin, and modified to include spatial and temporal variability in disturbance rates, including data from a logging experiment in the Central Amazon. Remote sensing investigations were employed to quantify spatial variability in large gap dynamics. Due to the combination of large quantities of downed wood and upturned soil, large gaps have unique spectral features than can be identified using hyperspectral imaging methods. Large gaps (blowdowns) were clearly identified on both IKONOS and HYPERION images for the Central Amazon near Manaus. HYPERION data were minimum noise fraction (MNF) transformed to reduce spectral dimensions for analyses, and pixels representing pure blowdown endmembers were extracted. Spectral unmixing was carried out using a robust method, and the fractional abundance of large gaps in each pixel was mapped using a conservative threshold of 30%. Additional HYPERION images were obtained for other Amazon sites, and a similar spectral unmixing of large gap fraction was carried out. Results demonstrate that carbon balance dynamics at the landscape-scale can be quite different than those measured on relatively small forest inventory plots, and underscore the important of improving our understanding of gap formation dynamics.
DE: 1615 Biogeochemical processes (4805)
DE: 1640 Remote sensing
DE: 0400 Biogeosciences
SC: Biogeosciences [B]
MN: 2004 AGU Fall Meeting