Estimation of Aboveground Forest Biomass Using Airborne Scanning Discrete Return Lidar in Douglas-fir

Models based on laser height metrics matching different quantiles of the distribution of laser canopy heights should be similar to one another with respect to their predictive capabilities of aboveground biomass providing: 1) the allometric relationships in the trees studied remain consistent; and 2) the vertical distributions of laser canopy heights and needle area/mass are related according to a simple quantile-quantile relationship. To explore the use of canopy-based quantile estimators in an application of aboveground biomass estimation in Douglas-fir, models based on laser height metrics corresponding to deciles of the distribution of laser canopy heights are compared. Because robust allometric equations were not available, the influence of using two different sets of allometric equations (referred to as BAR and TER equations) on model results is examined. The coefficient of determination (r) of decile models in the BAR and TER groups ranged from 0.25 to 0.43 and 0.34 to 0.50, respectively, whereas the root-mean-square error (RMSE) ranged from 29 to 34 and 32 to 36 Mg/ha, respectively. The greatest difference in RMSE between models was less than 5 Mg/ha. Decile models for each of the BAR and TER groups were not overtly different from one another. * Corresponding author. 1. RATIONALE FOR STUDY Our previous research in sugar maple (Acer saccharum Marsh.) in central Ontario, Canada demonstrated the potential of several, yet different canopy-based quantile estimators derived from airborne scanning lidar data to estimate aboveground biomass (AB) (Lim et al., in press). Models based on different canopy-based quantile estimators were found to be comparable to one another with respect to their predictive capabilities (i.e., coefficient of determination (r) and root-mean-square error (RMSE)) (Lim et al., in press). To explore how well canopy-based quantile estimators can be generalized to other forest types for applications of AB estimation and to avoid having to make an assumption of how the vertical distributions of laser returns and needle area are related, models based on canopy-based decile estimators were developed for Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) in western Canada using the same ground reference and airborne scanning lidar data reported in Magnussen and Boudewyn (1998). 2. MATERIALS AND METHODS