Scaling Forest Biometric Properties Derived from High Resolution Imagery to the Amazon Basin using Moderate Resolution Spectral Reflectance Data

Michael William Palace, Complex System Research Center, UNH, palace@kaos.sr.unh.edu (Presenting)
Bobby Braswell, Complex System Research Center, UNH, rob.braswell@unh.edu (Presenting)
Stephen Hagen, Complex System Research Center, UNH, hagennh@gmail.com
Michael Keller, NEON Inc., michael@kaos.sr.unh.edu

We are integrating field-measured tropical forest biometric variables with multi-scale remote sensing data from numerous sensors, for the purpose of characterizing and understanding patterns of forest structure across Amazonia. The Amazon basin contains the largest continuous tropical forest on the Earth (6 million km2) and constitutes 40% of the remaining area for this ecotype. The dynamic processes of growth and disturbance are reflected in the structural components of forests. Because Amazonia contains a large stock of biomass and because unmanaged Amazon forests currently may be a significant sink for carbon, understanding Amazonian forest dynamics reflected in forest structure is important for understanding regional and global carbon and biogeochemical cycles. A lack of comprehensive estimates of forest structural properties across the Amazon basin currently limits our ability to map carbon balances in this region. Recent observations from plots and eddy flux towers of carbon sink activity in Amazonian forests could be caused by recovery from disturbance, Because many or most of the currently studied forest plots were not randomly selected, and because their geographic distribution leaves vast areas unstudied, regional remote sensing data is required to understand the rate and frequency of forest disturbance in Amazonia and the linkage of disturbance to ecosystem carbon flux. We are using high resolution optical data to quantify forest structural properties including stem frequency, crown dimensions, and canopy gap fraction. Next, we will extrapolate these estimates of forest structure from the local and regional scale to the basin scale by linking them statistically with synoptic reflectance data from moderate resolution sensors (MODIS/MISR). This will be done annually for seven years (2002-2008) using linear and non-linear statistical methods. The resulting temporal and spatial distributions of forest structural properties will provide insight into changes in carbon cycling at regional scales.