Slope-adaptive waveform metrics of large footprint lidar for estimation of forest aboveground biomass

Abstract

The effect of terrain slope on large footprint lidar waveforms is still a great challenge to the accurate mapping of forest aboveground biomass (AGB), although previous studies have made significant progress for slopes lower than 15°. A new method was proposed to calculate slope-adaptive waveform metrics for forest AGB mapping over mountainous areas. First, a theoretical model was proposed to calculate lidar waveforms of bare ground with known terrain slopes. The forest waveform and its corresponding calculated bare ground waveform were then aligned by the signal ending points. Their energy quartiles were extracted separately by taking the signal ending points as references. The height differences between the corresponding energy quartiles were defined as slope-adaptive waveform metrics, which were then used as independent variables for the estimation of the forest AGB. The proposed method was examined with both simulated lidar waveforms of forest stands and real waveforms acquired by the Geoscience Laser Altimeter System (GLAS). Two typical footprint sizes (i.e., 25 m and 70 m in diameter) and 21 terrain slopes (i.e., from 0° to 40° at an interval of 2°) were used in the simulation of the lidar waveforms. The terrain slopes of the GLAS waveforms were between 5.54° and 34.08°. The results showed that the terrain stretching on the forest waveform could be compensated for by the bare ground waveform. Compared with the Gaussian decomposition and waveform parameter methods, the proposed method gave the best estimation of the forest AGB with R2 = 0.92 and RMSE = 15.95 Mg/ha using simulated waveforms with a footprint size of 70 m and R2 = 0.84 and RMSE = 34.99 Mg/ha using the GLAS waveforms. The results demonstrated that the proposed method had great potential for the estimation of forest AGB using lidar waveforms over mountainous areas.