COS 116-5 - Scaling up forest allometry with lidar and radar remote sensing

Thursday, August 11, 2011: 2:50 PM
13, Austin Convention Center
Naiara Pinto1, Ralph Dubayah1, Marc Simard2, Sassan Saatchi3, Bruce Cook4 and Paul Siqueira5, (1)Department of Geography, University of Maryland, College Park, MD, (2)Radar Science and Engineering, NASA Jet Propulsion Laboratory, Pasadena, CA, (3)Radar Science and Engineering, Jet Propulsion Laboratory, Pasadena, CA, (4)NASA Goddard Space Flight Center, Greenbelt, MD, (5)Electrical and Computing Engineering, University of Massachusetts, Amherst, MA
Background/Question/Methods

Forest aboveground biomass (AGB) can be mapped regionally by using data from lidar sensors to estimate mean and maximum forest height. The relationship height x AGB is, however, derived statistically and for individual study sites. Here, we examine whether models can be generalized by including (1) the decay exponent (alpha) of the plot's size class distribution, and (2) the individual allometry relating tree size and biomass.

We combine remote sensing and field data from two mixed-leaf forest sites in Maine and New Hampshire. We use assumptions from allometric scaling theory and self-thinning as a starting point. We first perform a sensitivity analysis to examine the impact of (1) and (2) on the height x AGB relationship. Then, we use data from 44 field plots to ask whether the decay exponent (alpha) can help constrain the height x AGB relationship. Last, we examine whether alpha can be estimated remotely using radar imagery.

Results/Conclusions

Fitting an exponential decay function to our field data yielded exponents that were compatible with previous scaling papers. However, the presence of emergent trees, elevational gradients, and fire disturbance complicates the model fitting. Uncertainties about tree allometry can introduce errors on the order of 170 Mg/tree. In addition, allometric scaling theory (specifically, the alpha coeffients reported in previous studies) predicts a height x AGB relationship that is consistent with remote sensing observations. The alpha coefficient shows a non-linear relationship with the radar signal.

Our results show that ecological theory can help refine algorithms for mapping carbon stocks, by providing a framework to account for the interdependencies among forest structural parameters.

Copyright © . All rights reserved.
Banner photo by Flickr user greg westfall.