Danelle Laflower, Thomas Millette, and Eugenio Marcano
Mount Holyoke College
Increasing atmospheric carbon dioxide (CO2) levels are a leading cause of climate change (Malhi et al. 2002). Most terrestrial carbon is stored in forest biomass (Olson et al. 1985) by the photosynthetic conversion of atmospheric CO2. Therefore, estimating carbon stocks helps us quantify CO2 concentrations. Ecologists calculate biomass with empirical allometric equations that use species and diameter at breast height (dbh) and divide by two to estimate carbon (Brown and Schroeder 1999, Jenkins et al. 2004).
I hypothesized that I could estimate stand-level biomass using the Airborne Imaging Multispectral Sensor’s (AIMS) high-resolution imagery and LiDAR height measurements. To test this notion, I selected a study area on Mount Holyoke College property, in South Hadley, Massachusetts and systematically sampled 366 trees for species, height, dbh, and canopy data. I obtained LiDAR-derived canopy height and high resolution imagery with the AIMS system. I averaged the LiDAR values and the sampled trees’ heights within each plot to obtain plot average height for each method. By dividing the area into 20 plots, a linear regression indicated that the LiDAR average height was a significant predictor of dominant tree average height (p=0.000, R2=0.638).
For the remote biomass estimation, in each subplot I identified species and stem density in georeferenced AIMS images. From ground data, I created linear regression models to estimate dbh from height. I used LiDAR height to estimate dbh values in the corresponding biomass equations. I multiplied these biomass values by the number of stems of each species in the plot, scaled the value to hectare, and summed the results. I compared these results with the ground biomass data. For the ground validation of biomass, within the twenty plots I created ten 900-m2 subplots, where I identified species, measured dbh for all live stems >12.4 cm, and recorded place in the canopy. I calculated biomass using the species-specific allometric biomass equations from Jenkins et al. (2004), summed the results, and scaled to hectare. I also calculated biomass using only dominant and co-dominant trees. The linear regression indicated that the remote method was a significant predictor of dominant tree ground biomass (p=0.012, R2=0.568). These results suggest that this technique can adequately predict stand-level biomass in a southern New England forest. The next step will be to expand the locations to determine the feasibility of using this method for other forest types.