Site Harvard Hemlock
Site 305Site Harvard EMS tower
The LAI estimates are impacted by changes in the occlusion effect at the different scales.
75m is considered as the appropriate scale due to the layout of the EVI scans.
Modeled G vs. G=0.5: more work is needed to define the transition between the two in intermediate sites
Retrieving Leaf Area Index and Foliage Profiles through Voxelized 3-D Forest Reconstruction Using Terrestrial Full Waveform Dual-Wavelength Echidna® Lidar (DWEL)
Alan Strahler1, Xiaoyuan Yang2, 3, Crystal Schaaf3, Zhan Li1, Zhuosen Wang3, Tian Yao4, Feng Zhao5, Edward Saenz3, Ian Paynter3, Ewan Douglas1, Supriya Chakrabarti6, Timothy Cook6, Jason Martel6, Glenn Howe6, David Jupp7, Darius Culvenor8, Glenn Newnham9, Jenny Lovell10
1Boston University, Boston, MA, USA; 2Sandia National Laboratory, Livermore, CA, USA; 3University of Massachusetts Boston, Boston, MA, USA; 4Montclair State University, Montclair, NJ, USA; 5University of Maryland, College Park, MD, USA; 6University of Massachusetts Lowell, Lowell, MA, USA; 7CSIRO Marine and Atmospheric Research, Canberra ACT, Australia; 8Environmental Sensing Systems, Melbourne,
Victoria 3000, Australia; 9CSIRO Land and Water, Clayton South, Victoria, Australia; 10CSIRO Marine and Atmospheric Research, Hobart, Tasmania, Australia
Foliage Area Volume Density & Leaf Area Index
, ,1, ,, ,
app
t gap
rf r
G P r
G-function Leaf reflectance
Apparent reflectance of the point
Probability of gap
Three-dimensional Forest Reconstruction
Voxelization
Voxelization transform s the irregular, unorganized cloud of data points in the 3-D forest reconstruction into volumetric datasets.
Voxel Based LAI & FAVD Estimation•FAVD Model for point cloud
•The total effective area of objects in volume V
or
•Leaf area index:
Noted: In the case of multiple observations: the average of LAI is used.
, ,, , , ,
, ,app
Vt gap
rVLA r V f rG P r
( , , )( , , )
( ) ( , , )app
Vt gap
x y zVLA x y zG P x y z
, , baseLAI LA x y z A
Modeling
Model the product from the distribution of hit values with angle in the scan data and apply the model to each hit to get LAI and FAVD profile.
1 22
max, 21 2 cos 1 sint c H HG f f
Horizontal fraction of canopy cover
Derived value of at the hinge angle
Site ID Field LAIVoxel based LAI by Diameter
G=0.550m 75m 100m
Hemlock 4.32±0.27 4.95 4.73 3.66 2.68EMS tower 4.67±0.82 5.30 4.86 3.52 2.83
305 2.03±0.50 6.54 *5.55 4.86 2.77
Measuring and monitoring canopy biophysical parameters provide a baseline for carbon flux studies related to deforestation and disturbance in forest ecosystem. Terrestrial full-waveform lidar systems, Echidna® Validation Instrument (EVI) and its successor Dual-Wavelength Echidna® Lidar (DWEL), offer rapid, accurate, and automated characterization of forest structure (Strahler et al., 2008; Yang et al., 2013). In this study, we proposed a methodology based on voxelized 3-D forest reconstruction built from EVI and DWEL scans (Douglas et al., 2012) to directly estimate two important biophysical parameters: Leaf Area Index (LAI) and foliage profiles.
Overview
Site Characteristics
Site ID Leading dominants
Top canopy height
(m)
Mean DBH (m)
Stem count density (ha–1)
Above-ground
biomass (t ha–1)
Hemlock Hemlock 22.6 0.24±0.02 906±71 234±7
EMS Tower Red maple, red oak 26.4 0.28±0.02 951±69 373±36
305 Red fir 45.2 0.58±0.02 284±40 1215±150
While the two instruments detected the top of the canopy pretty well, LVIS sees more upper canopy component while EVI sees more lower canopy component over the same forest area. (Profiles are compared at 75 m plot diameter.)
Site Harvard Hemlock
Site 305Site Harvard EMS tower
Terrestrial (EVI) vs. Airborne (LVIS) Lidar
Dual-Wavelength Echidna® Lidar (DWEL)
The Dual-Wavelength Echidna® Lidar (DWEL), the successor instrument to the EVI, emits simultaneous laser pulses at 1064 nm and 1548 nm wavelengths. DWEL scans provide the capability to separate hits of leaves from hits of trunks and branches because of the reduced response at 1548 nm due to water absorption by leaf cellular contents.
Normalized Difference Index (NDI): (1064nm – 1548nm) / (1064 nm + 1548 nm)
1064nm 1548 nm
1064 nm 1548 nm
Classification by thresholding NDI
By simply thresholding NDI of each point, the range effects in the classification is largely reduced, and the trunks are differentiated from foliage/branchlets points.
Top Related