Further Studies of Forest Structure Parameters

advertisement
Further Studies of Forest Structure Parameters Using Echidna® Ground-Based Lidar
Alan Strahler1, Tian Yao2, Feng Zhao3, Xiaoyuan Yang4, Crystal Schaaf4, Zhuosen Wang4, Zhan Li1, Curtis Woodcock1,
Darius Culvenor5, David Jupp6, Glenn Newnham5, and Jenny Lovell7
1Center
for Remote Sensing, Department of Earth and Environment, Boston University, Boston, MA; 2Remote Sensing Laboratory, Montclair State University, Montclair, NJ;
of Geographical Sciences, University of Maryland, College Park, MD; 4Department of Environmental, Earth and Ocean Sciences, University of Massachusetts
Boston, Boston, MA; 5CSIRO Land and Water, Melbourne, VIC, Australia; 6CSIRO Marine and Atmospheric Research, Canberra, ACT, Australia; 7CSIRO Marine and Atmospheric
Research, Hobart, TAS, Australia
3Department
Detecting Forest Growth and Change with EVI
Ongoing work with the Echidna® Validation Instrument
(EVI), a full-waveform, ground-based scanning lidar (1064
nm) developed by Australia’s CSIRO and deployed by
Boston University in California conifers (2008) and New
England hardwood and softwood (conifer) stands (2007,
2009, 2010), confirms the importance of slope correction in
forest structural parameter retrieval; detects growth and
disturbance over periods of 2-3 years; provides a new way
to measure the between-crown clumping factor in leaf area
index retrieval using lidar range; and retrieves foliage
profiles with more lower-canopy detail than a large-footprint
aircraft scanner (LVIS), while simulating LVIS foliage profiles
accurately from a nadir viewpoint using a 3-D point cloud.
EVI scans can detect change, including both growth and
disturbance, in periods of two to three years. We revisited
three New England forest sites scanned in 2007-2009 or 20072010. A shelterwood stand at the Howland Experimental
Forest, Howland, Maine, showed increased mean DBH,
above-ground biomass and leaf area index between 2007 and
2009. Two stands at the Harvard Forest, Petersham,
Massachusetts, suffered reduced leaf area index and reduced
stem count density as the result of an ice storm that damaged
the stands. At one stand, broken tops were visible in the 2010
point cloud canopy reconstruction.
Instrument and Data Characteristics
Slope Correction
Echidna® ground-based lidar. Lidar pulses strike a rotating mirror at an angle of
45°, providing a scan through zenith angles of ±130° in a vertical circle. As the
instrument rotates on its vertical axis, data from all azimuths are acquired.
Slope correction is important for accurate retrieval of forest
canopy structural parameters, such as mean diameter at
breast height (DBH), stem count density, basal area, and
above-ground biomass. Topographic slope can induce errors
in parameter retrievals because the horizontal plane of the
instrument scan, which is used to identify, measure, and count
tree trunks, will intersect trunks below breast height in the
uphill direction and above breast height in the downhill
direction. A test of three methods at southern Sierra Nevada
conifer sites improved the range of correlations of these EVIretrieved parameters with field measurements from 0.53-0.68
to 0.85-0.93 for the best method.
Change in structural parameters, New England
forest sites
Mean of
DBH†
(m)
Field
0.26 ± 0.01
0.26 ± 0.01
0.24 ± 0.01
0.24 ± 0.01
EVI
0.25 ± 0.01
0.25 ± 0.01
0.23 ± 0.01
0.24 ± 0.01
Field
953 ± 59
933 ± 57
556 ± 40
562 ± 24
Harvard Hemlock 2007–2009
Change in LAI: 4.52 – 3.95
Howland Shelterwood 2007–2009
Change in LAI: 3.16 – 3.29
587 ± 40
589 ± 25
These foliage profiles (foliage area volume density with height) are
retrieved from Howland Shelterwood site and the Harvard Hemlock site
. Profiles are averages of five scans, located at the corners and
midpoint of a 50 m x 50 m square. The foliage profile at the Hemlock
site shrank as a result of ice storm damage. At the shelterwood site, the
foliage profile expanded slightly.
Stem density
(tree/ha)
Basal area
2
(m /ha)
2009
EVI
921 ± 79
906 ± 71
Field
54.65 ± 6.03
54.23 ± 5.37
26.34 ± 1.01 27.89 ± 1.03
EVI
51.21 ± 3.54
49.99 ± 4.16
27.88 ± 2.08 30.62 ± 0.58
Above-ground Field
biomass
EVI
(ton/ha)
Dominant species
2007
2010
Howland
Shelterwood
2007
2009
Source
2007
260.9 ± 16.6
255.2 ± 11.2
110.0 ± 4.05 118.0 ± 4.87
235.3 ± 17.3
222.7 ± 21.0
112.2 ± 9.21 125.2 ± 4.65
Hemlock, white pine
Crown
missi
ng
Hemlock, red spruce
Before DTM correction
Sector Method Use the existing
find trunks algorithm on multiple
azimuth sectors but vary the search
height setting in each sector to
conform to the average height in
the sector.
Fitted Ground Plane Method Fit a
simple first-order model of slope and
orientation to the ground returns and
recalculate the height associated with
each range to fit the model.
After DTM correction
DTM Method Merge the scans into
a point cloud, construct a highresolution DTM from the ground
points, and adjust the elevation of
every point to the surface.
Foliage Profile
Foliage profiles retrieved from EVI scans at five Sierra Nevada
sites are closely correlated with those of the airborne Lidar
Vegetation Imaging Sensor (LVIS) when averaged over a
diameter of 100 m. At smaller diameters, the EVI scans have
more detail in lower canopy layers and the LVIS and EVI
foliage profiles are more distinct. Foliage profiles derived from
processing 3-D site point clouds with a nadir view match the
LVIS foliage profiles more closely than profiles derived from
EVI in scan mode. Removal of terrain effects significantly
enhances the match with LVIS profiles.
Harvard
Hemlock
Parameters
Canopy profile without topo correction
EVI scan layout
30 m
50 m
70 m
100 m
With topo correction
The Echidna Validation Instrument (EVI) operates at a wavelength of 1064 nm. At
this wavelength, leaves, branches, trunks, litter, and soils all have about the same
reflectance, in the range of 0.5 to 0.6.
Change in Foliage Profile
Without topo correction
Overview
3-D canopy top image
with LVIS shots
Canopy profile with topo correction
As foliage profile area increases, LVIS, EVI scan, and EVI point cloud foliage profiles become more
similar. The point cloud profile fits LVIS best. Topographically-corrected data are smoothest.
Clumping Index
Examples of EVI data for a hemlock stand at the Harvard Forest in Massachusetts
(top image) and a giant sequoia-red fir stand at Sequoia National Forest in
California (bottom image). The images are in a plate carrée projection that displays
the data by azimuth angle (x-axis) and zenith angle (y-axis).
A proper Auzzie echidna.
Acknowledgements:
This work was supported by NASA grants NNG06GI92G and
NNX08AE94A and NSF grant DBI-0923389 . We thank the
assistance of John Lee at Howland Experimental Forest,
Audrey Barker Plotkin at Harvard Forest, Andrew Richardson
at Bartlett Experimental Forest, and Ralph Dubayah at the
Sequoia National Forest. Field assistance was provided by
Zhuosen Wang, Miguel Roman, Mitchell Schull, Qingling
Zhang and Shihyan Lee in collecting field and EVI data.
A new method for retrieval of the forest canopy between-crown
clumping index from angular gaps in hemispherically-projected
EVI data traces gaps as they narrow with range from the
instrument, thus providing the approximate physical size,
rather than angular size, of the gaps. In applying this method
to a range of sites in the southern Sierra Nevada, element
clumping index values are lower (more between-crown
clumping effect) in more open stands, providing improved
results as compared to conventional hemispherical
photography. In dense stands with fewer gaps, the clumping
index values were closer.
Contact:
Alan Strahler, alan@bu.edu; Crystal Schaaf, Crystal.Schaaf@umb.edu.
25 m
50 m
75 m
100 m
135 m
Gap images with range. As range increases, gaps are
narrowed.
Within-element gap
Between-element gap
(Left) Gap fraction measured at far range by EVI correlates very well with gap fraction in hemi photos.
(Center) Hemi clumping index, based on angular gaps, is larger than EVI clumping index in more open
stands, based on physical size of gaps. This is because the EVI sees small gaps at far range as betweenelement gaps, while the hemi photo sees them as within-element gaps. (Right) Because EVI detects more
between-element clumping, EVI measures higher LAI values.
Download