= 0.202. Thus, the performance of
SI
is better than
TH
for tree
species classification. Similar to
SI
, average
CVs
for three groups
are 9.25, 8.23, and 9.44, with standard deviation of 2.83, 2.59,
and 4.49 respectively (see Figure 5f). The ratio between-species
to within-species is highest, F(2,113) = 3.42, p = 0.036, and
resulted in the best classification. However, the fusion method
cannot significantly improve the classification for maple, oak,
and other species due to their similar crown shapes.
Conclusions
Lidar data can provide additional power in improving clas-
sification accuracy for tree species with distinct structures but
similar spectral properties(Alonzo
et al
., 2014; Dalponte
et al
.,
2012; Fassnacht
et al
., 2016). In this paper, two structure fea-
tures: crown shape index and coefficient of variation within
tree crowns were extracted from lidar-derived
CHM
, integrated
with treetop-based crown spectra for ash tree species identi-
fication, and compared with tree height for accuracy assess-
ment.
SI
illustrates the probability distribution of heights for
random points within a crown; it is efficient in computation
and effective in representing a complex crown shape.
CV
refers to a standardized measure of height dispersion within
a crown and represents the ratio of the standard derivation to
the mean, it is suitable for comparing data sets with different
means and different measures. As a result, integrating with
shape index and
CV
respectively, the treetop-based approach-
es significantly increased the accuracies for ash trees identi-
fication, but
CV
performs better than shape index because
CV
measures the degree of height dispersion while shape index
only expresses the trends of height distribution.
Ash tree species classification in deciduous forests is chal-
lenging. The spectral similarity amongst deciduous species
and spectral noise caused by mixed-pixel problem decrease
the accuracy of ash tree species classification. In contrast, the
appropriate structural features (e.g.,
SI
and
CV
) extracted from
lidar data have the ability to measure the unique crown bio-
physical properties invariable from life stage and mitigate the
spectral problems. Therefore, the fusion of the hyperspectral
imagery and lidar-derived shaped-related features can signifi-
cantly improve the ash tree species classification.
In this research, we applied discrete return lidar to ex-
tracted tree heights, crown radii and crown shape features.
Due to limitation of the conventional lidar data systems, only
surfaces with sufficiently spaced apart can be separated from
the returns. However, full-waveform systems have the ability
to digitize and record the entire backscattered signal of each
pulse, and record detailed geometric and physical properties
of target objects(Lin
et al.
, 2014).Thus the benefits of full-
waveform data, used in single tree detection, structure char-
acteristics estimations, and tree species classification (Duong,
2010; Gupta
et al
., 2010; Heinzel and Koch, 2011), can be
used to improve ash tree identification in the future.
Acknowledgments
This work was supported by University of Wisconsin Milwau-
kee Graduate School Research Committee Award and the Na-
tional Natural Science Foundation of China (No. 41371397).
References
Abdi, H. (2007). RV coefficient and congruence coefficient,
Encyclopedia of Measurement and Statistics
, 849–853.
Alonzo, M., B. Bookhagen, and D.A Roberts, 2014. Urban tree species
mapping using hyperspectral and lidar data fusion,
Remote
Sensing of Environment,
148:70–83.
Alonzo, M., K. Roth, and D. Roberts, 2013. Identifying Santa Barbara’s
urban tree species from AVIRIS imagery using canonical
discriminant analysis,
Remote Sensing Letters
, 4:513–521.
Alves, L.F., 2002. Tree allometry and crown shape of four tree species
in Atlantic Rain Forest, South-East Brazil,
Journal of Tropical
Ecology
,
18:245–260.
Antin, C.Š., P.Šlissier, G.Š.,Vincent, and P. Couteron, 2013. Crown
allometries are less responsive than stem allometry to tree
size and habitat variations in an Indian monsoon forest,
Trees,
27:1485–1495.
Ben-Arie, J.R., G.J. Hay, R.P. Powers, G. Castilla, G.,and B. St-Onge,
2009. Development of a pit filling algorithm for LiDAR canopy
height models,
Computers & Geosciences
, 35:1940–1949.
Cappaert, D., D.G. McCullough, T.M. Poland, and N.W. Siegert, 2005.
Emerald ash borer in North America: A research and regulatory
challenge,
American Entomologist
, 51:152–165.
Clark, M.L., D.A. Roberts, and D.B. Clark, 2005. Hyperspectral
discrimination of tropical rain forest tree species at leaf to crown
scales,
Remote Sensing of Environment,
96:375–398.
Dalponte, M., L. Bruzzone, and D. Gianelle, 2008. Fusion of
hyperspectral and LIDAR remote sensing data for classification
of complex forest areas,
IEEE Transactions on Geoscience and
Remote Sensing,46
:1416–1427.
Dalponte, M., L.Bruzzone, L.,and D. Gianelle, 2012. Tree species
classification in the Southern Alps based on the fusion of very
high geometrical resolution multispectral/hyperspectral images
and LiDAR data,
Remote Sensing of Environment
, 123:258–270.
Deo, R., M. Russell, G. Domke, H.-E, Andersen, W. Cohen, & C.
Woodall, 2017. Evaluating site-specific and generic spatial
models of aboveground forest biomass based on Landsat time-
series and LiDAR strip samples in the Eastern USA,
Remote
Sensing
, 9:598.
Duong, V.H.,2010. Processing and application of ICESat large
footprint full waveform laser range data,
Delft University of
Technology.
Fassnacht, F.E., H. Latifi, A., Ghosh, P.K. Joshi, and B. Koch, 2014.
Assessing the potential of hyperspectral imagery to map bark
beetle-induced tree mortality,
Remote Sensing of Environment
,
140:533–548.
Fassnacht, F.E., H. Latifi, Ñ .K.Stere, A. Modzelewska, M. Lefsky,L.T.
Waser, C. Straub, and A. Ghosh, 2016. Review of studies on tree
species classification from remotely sensed data,
Remote Sensing
of Environment
:186:64–87.
Flower, C.E., K.S. Knight, and M.A. Gonzalez-Meler, 2013. Impacts of
the emerald ash borer (
Agrilus planipennis Fairmaire
) induced
ash (
Fraxinus
spp.) mortality on forest carbon cycling and
successional dynamics in the eastern United States,
Biological
Invasions
, 15: 931–944.
George, R., H. Padalia, and S. Kushwaha, 2014. Forest tree species
discrimination in western Himalaya using EO-1 Hyperion,
International Journal of Applied Earth Observation and
Geoinformation
, 28:140–149.
Ghosh, A., F.E. Fassnacht, P. Joshi, and B. Koch, 2014. A framework
for mapping tree species combining hyperspectral and LiDAR
data: Role of selected classifiers and sensor across three spatial
scales,
International Journal of Applied Earth Observation and
Geoinformation, 26
:49–63
Gupta, S., H. Weinacker, and B. Koch, 2010. Single tree detection
using full waveform laser scanner data,
Silvilaser.
Harlow, W.M.,1991.
Textbook of Dendrology: Covering the Important
Forest Trees of the United States and Canada
, McGraw-Hill.
Heinzel, J., and B. Koch, 2011. Exploring full-waveform LiDAR
parameters for tree species classification,
International Journal of
Applied Earth Observation and Geoinformation
, 13:152–160.
Herms, D., P. Bonello, D. Smitley, E. Rebek, and D. Cipolinni, 2004.
Interspecific variation in ash resistance/susceptibility to EAB,
Proceedings of the 2004 Annual Emerald Ash Borer Meeting
,
Romulus, Michigan, USA, October.
Hill, R., and A. Thomson, 2005. Mapping woodland species
composition and structure using airborne spectral and LiDAR
data,
International Journal of Remote Sensing
, 26:3763–3779.
502
August 2018
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
