You are here
Harvard Forest Symposium Abstract 2005
- Title: The Use of Waveform Lidar to Measure Temperate Mixed Deciduous Forest
- Primary Author: Jeanne Anderson (University of New Hampshire - Main Campus)
- Additional Authors: J.Bryan Blair (NASA Goddard Space Flight Center); Ralph Dubayah (University of Maryland); Michelle Hofton (University of Maryland); Peter Hyde (University of Maryland); Robert Knox (NASA Goddard Space Flight Center); Mary Martin (University of New Hampshire - Main Campus); Birgit Peterson (University of Maryland); Marie-Louise Smith (USDA Forest Service)
- Abstract:
The direct retrieval of canopy height and the estimation of aboveground
biomass are two important measures of forest structure that can be
quantified by airborne laser scanning at landscape scales. These and other
metrics are central to studies attempting to quantify global carbon cycles
and to improve understanding of the spatial variation in forest structure
evident within differing biomes. Waveform lidar imagery was acquired on
September 26, 1999 and July 18-26, 2003 over the Bartlett Experimental
Forest (BEF) in central New Hampshire (USA) using LVIS (Laser Vegetation
Imaging Sensor), the airborne VCL (Vegetation Canopy Lidar) simulator. We
found strong agreement between field and lidar measurements of height (LHT)
and quadratic mean stem diameter (QMSD) with 1999 LVIS metrics (LHT: r2 =
0.89, RMSE = 3.28 m., p < 0.000; QMSD: r2 = 0.88, RMSE = 3.41 cm., p <
0.000) at the footprint level. Allometric calculations of aboveground
biomass (AGBM) and basal area (BA) and 1999 LVIS metrics (AGBM: r2 = 0.76,
RMSE = 54.75 Mg ha-1, p < 0.000; BA: r2 = 0.71, RMSE = 3.8 m2 ha -1, p <
0.000) also showed good agreement at the footprint level. 1999 LVIS height
metrics and resulting models did, however, show an overall tendency towards
underestimation of actual height. Comparisons of Forest Service plot
measurements of AGBM and QMSD from unmanaged forest tracts against
predicted values based on 2003 LVIS metrics were within the range of error
predicted by the models in 72% and 77.5% of the cases respectively. But
direct plot-level relationships between Forest Service inventory plots and
aggregated 2003 LVIS metrics were not as strong. Sampling limitations and
variability in aggregated metrics weakened these relationships. Agreement
improved when plot land-use history and species composition were factored
into the analysis. - Research Category: