Open Journal of Forestry
2012. Vol.2, No.3, 89-96
Published Online July 2012 in SciRes (http://www.SciRP.org/journal/ojf) http://dx.doi.org/10.4236/ojf.2012.23012
Estimating Vertical Distribution of Vegetation Cover in
Temperate Heterogeneous Forests Using Airborne Laser
Scanning Data
Keiko Ioki1, Junichi Imanishi2, Takeshi Sasaki2, Youngkeun Song2,
Yukihiro Morimoto2, Hisashi Hasegawa3
1Graduate School of Agriculture, Kyoto University, Kyoto, Japan
2Graduate School of Global Environmental Studies, Kyoto University, Kyoto, Japan
3Field Science Education and Research Center, Kyoto University, Kyoto, Japan
Received April 26th, 2012; revised May 28th, 2012; accepted June 8th, 2012
Vertical structure is important for understanding forest environment, yet difficult to characterize, espe-
cially in temperate heterogeneous forests where the structure is complex. This study used data from a
small-footprint airborne laser scanning (ALS) to estimate vegetation coverage in four stratum ranges in a
warm temperate forest in Japan: >12 m, 8 - 12 m, 4 - 8 m, and 0 - 4 m in height. Field data were collected
in 17 broad-leaved and 12 coniferous sample plots, consisting of the proportion of vegetation cover in
each stratum range. The field and ALS measurements were conducted in summer, during leaf-on condi-
tions. Using echo attributes (first, last, intermediate, and only), we calculated the vegetation coverage in-
dex (VCI) at 1-m height intervals. The cumulative sum of the VCI (CUMVCI) was then computed and
compared with field observations. Linear regression analysis showed that the ALS data gave reasonable
estimates of vegetation coverage in the upper two or three stratum ranges in broad-leaved stands, and in
the upper two stratum ranges in coniferous stands. The model gave reproducible estimates until approxi-
mately 95% of the total returns had been applied. We conclude that ALS data can provide useful informa-
tion on natural habitats in the management of warm temperate forest.
Keywords: Forest Structure; Remote Sensing; Airborne Laser Scanning
Introduction
Remote sensing is an effective tool for obtaining spatial in-
formation overlarge areas, which enables the development and
assessment of management plans for forests and other ecosys-
tems.It can be an attractive alternative to conventional field
survey because it has the ability to obtain the measurements
from the areas that are limited by accessibility and because the
data collection/processing are efficient. Airborne laser scanner
(ALS) offers distinct advantages in the representation of three-
dimensional spatial structures (Lefsky et al., 2002), and numer-
ous studies have successfully used this technique to estimate
the characteristics of forest stands (e.g., Nilsson, 1996; Mag-
nussen & Boudewyn, 1998; Lefsky et al., 1999; Næsset, 2002;
Næsset & Økland, 2002).
Traditionally, researchers have described forest structure in
terms of stand attributes such as tree height, stem density, di-
ameter at breast height (DBH), and basal area (BA). These
attributes are especially useful when applied to homogeneous
forests (e.g., boreal coniferous forests); however, they may not
fully describe the complex spatial patterns in heterogeneous
forests in the warm temperate zone. Vertical distribution of
foliage and woody materials is one of the attributes most com-
monly used to represent forest stand structure(Brokaw and Lent
1999) and the first in which a quantitative relationship was
established between an element of structure and a measure of
faunal diversity (McElhinny et al., 2005). In a landmark study
on forest birds, MacArthur and MacArthur (1961) established a
linear relationship between foliage height diversity (FHD),
which describes the distribution of foliage within different
stratum ranges, and bird species diversity. In warm temperate
forests, vertical distribution of foliage and woody materials
could be a good attribute for assessing forest wildlife habitat.
Several studies have attempted to characterize the vertical
structure of forests using ALS (Zimble et al., 2003; Clark et al.,
2004; Maltamo et al., 2005; Hill & Broughton, 2009). These
studies aimed at distinguishing single-storied and multi-storied
stand structures, or characterizing the understory in terms of
tree height and density. Few studies have used ALS data to
estimate the stand structure including the vertical distribution of
foliage. Coops et al. (2007) found that ALS-based models of
vertical foliage distribution are robust when compared with
field observations in Canadian coniferous forests, and Hashi-
moto et al. (2004) investigated the use of leaf-off ALS data to
derive indices describing forest habitat for birds in a deciduous
forest in Japan. Miura and Jones (2010) demonstrated the abil-
ity to characterize ecological structure in a dry Eucalypt forest
in Australia. These studies indicate the possibility of estimating
the vertical distribution of forest foliage using ALS data; how-
ever, a potential of using leaf-on ALS data in a warm temperate
broad-leaved forest has yet to be investigated.
More recently, high-resolution full-waveform recording ALS
system became commercially available. Using this system, a
digitized echo waveform of the sensor reveals all the informa-

K. IOKI ET AL.
tion the laser pulse collected during its trip to the surface (Hug
et al., 2004). Persson et al. (2005) compared the point clouds
from the conventional discrete-return sensor and the point
clouds extracted from the waveform sensor in several forest
stands. They demonstrated that additionally extracted points
from waveform were obtained from reflections in the tree
crowns or in the understory. Thus, the vertical point distribution
from waveform returns is expected to provide more detailed
three-dimensional information of stands.
The objective of the present study was to investigate the use
of leaf-on ALS data acquired by full-waveform sensor in pre-
dicting the vertical distribution of vegetation cover within dif-
ferent strata of a warm temperate forest. Vegetation coverage in
this study, defined as the proportion of the area covered with
vegetation consisting of foliage or other woody materials, was
estimated at four vertical stratum ranges (>12 m, 8 - 12 m, 4 - 8
m, and 0 - 4 m) from the proportions of pulse returns in each
height range. The technique presented in this study, which we
call the Attribute Based Weighting (ABW) method, is based on
the attributes of laser returns and each of the developed indices
is weighted according to the numbers of returns. This method
was used to create indices of vegetation cover applicable to
heterogeneous broad-leaved and coniferous forests.
Materials and Methods
Study Area
The study site is located in the mountains of Shugakuin Im-
perial Villa, Kyoto city, in western Japan (35˚03 ′N, 135˚48′E),
covering an area of 53 ha (Figure 1). The area is characterized
by mountainous forests and a complex topography, which var-
ies in elevation from 100 to 343 m above sea level (a.s.l). The
landscape of the forest consists of a mosaic of relatively small
stands of various tree species. The forest types within the study
area include coniferous plantation, broad-leaved, and mixed
stands with diverse vegetation structures. The dominant conif-
erous species are Japanese cedar (Cryptomeria japonica D.
Don), Japanese cypress (Chamaecyparis obtuse Sieb. et Zucc.),
and Japanese fir (Abies firma Sieb. et Zucc.). More than 30 spe-
cies of broad-leaved trees are present, including Quercus spp.
(Quercus serrata Murray, Quercus variabilis Bl., and Quercus-
glauca Thunb.) and Japanese maple (Acer palmatum Thunb.).
The composition of broad-leaved stands is typical of forests of
the warm-temperate zone in Japan, comprising a mixed com-
munity of many broad-leaved evergreen species in the upper
Shuga kuin
Imperial VillaBroad-leaved
Coniferous
JAPAN
N
Kyoto
0 125 250 500
m
Figure 1.
Study area and distribution of sample plots.
canopy, succeeded by a complex vertical profile containing
several distinct tree layers. In general, two or three tree layers
can be identified in addition to the shrub layer (Shidei, 1974).
Ground Refe rence Data
Vegetation coverage was assessed in collected in 17 broad-
leaved and 12 coniferous sample plots (each 10 × 10 m) during
the summer of 2008. The most widely known of the measuring
approaches is that of Braun-Blanquet scale (1932) which de-
fines the coverage and abundance of plant communities. It is
often applied with four vegetation layers, trees (B), shrubs (S),
herbs (K), and mosses (M). Based on this concept, the propor-
tion of vegetation cover was determined at six stratum ranges
within each plot (>12 m, 8 - 12 m, 4 - 8 m, 2 - 4 m, 1 - 2 m, and
0 - 1 m). The field measurements for the 0 - 1 m, 1 - 2 m, and 2
- 4 m strata were subsequently combined and reported as one
stratum (0 - 4 m). The locations of all the plots were determined
by the differential global positioning system (DGPS) (GPS
Pathfinder ProX; Trimble Navigation Ltd., USA) and by com-
pass survey (Tracon LS-25 surveying compass; Ushitaka Mfg.
Co., Ltd., Japan).
ALS Data Col l ectio n
ALS data were acquired from a helicopter (Nakanihon Air
Service Co., Ltd.) on 23 July 2008 using a Riegl LMS-Q560
laser scanner. Average flying altitude was 300 m a.g.l., result-
ing in a footprint diameter of approximately 15 cm. Specifica-
tions of the ALS data are listed in Table 1.
The ALS survey provided digitized echo signals, which were
then converted into laser point clouds using RiAnalyze software
(RieglLMS GmbH, Austria). RiAnalyze applies a full wave-
form analysis to digitized echo signals and transforms geomet-
ric data (i.e., range and scan angles) into Cartesian coordinates
(RieglLMS GmbH, 2009). Using RiAnalyze software, only
local maxima with absolute amplitudes exceeding threshold
values are taken into account; by applying Gaussian pulse esti-
mation, point clouds with four different threshold values were
available. Average echo densities were between 72 returns/m2
at the highest threshold value and 85 returns/m2 at the lowest
threshold value. Although we analyzed the dataset using dif-
ferent threshold values, the results of the different trials were
not significantly different (as determined by Tukey’s HSD test);
therefore, the dataset obtained with the highest threshold value
which is routinely used in practical operations was selected for
this study, because less noise was present in the data.
Table 1.
Specifications of the ALS data.
Parameter Value
Laser wavelength 1550 nm
Flight altitude 300 m
Flight speed 80 km/h
Laser pulse frequency 150 kHz
Scan frequency 80 Hz
Divergence 0.5 mrad
Footprint 0.15 m
Swath distance 346 m
Scan angle ±30˚
90

K. IOKI ET AL.
Each return was classified into one of four echo attributes: 1)
first—the first return of a multiple return; 2) last—the last re-
turn of a multiple return; 3) intermediate—intermediate returns
between the first return and the last return of a multiple return;
and 4) only—a single return. Figure 2 shows examples of each
type of vegetation return.
tation return was calculated by subtracting the altitude of the
DTM from the altitude of the vegetation return. Returns below
0.5 m were excluded from further analysis to eliminate the
effects of misclassifications of vegetation and ground returns.
Model Development
Our approach, ABW method is based on echo attributes of
laser returns. First, using the points classified as vegetation
returns and their echo attributes (first, last, intermediate, and
only), we calculated two parameters for each 1 m stratum
(Figure 3): 1) The first parameter (FO) is related to the first and
singular echo attributes; for the stratum i = 1, FO is
Data Analysis
To generate a DTM, first, ALS data consisting of all returns
were classified by Terrascan software (Terrasolid Inc., Finland)
to distinguish between “ground” and “vegetation” returns. The
procedure was performed by an embedded triangulated irregu-
lar network (TIN) algorithm developed by Axelsson (2000).
Table 2 shows parameter settings of the ground classification.
Second, a DTM raster with a 50 cm grid was created, also using
the TIN algorithm. The remaining data were classified using the
DTM, and the points above the DTM were classified as vegeta-
tion returns. Finally, the height above the ground of each vege-

F1 O1
F
1
1
FO
kO
m
k
nn
nn




k
(1)
and for other strata (i ≥ 2), FO is
16 - 17 m
15 - 16 m
14 - 15 m
13 - 14 m
12 - 13 m
11 - 12 m
10 - 11 m
9 - 10 m
8 - 9 m
7 - 8 m
6 - 7 m
5 - 6 m
4 - 5 m
3 - 4 m
2 - 3 m
1 - 2 m
0 - 1 m
Only
First
Last
Intermediate
0 400 800 1200
frequency
First
Only
Last
Intermediate
DTM
DTM
height height
First
Only
Last
Intermediate
Only
First
Last
Intermediate
16 - 17 m
15 - 16 m
14 - 15 m
13 - 14 m
12 - 13 m
11 - 12 m
10 - 11 m
9 - 10 m
8 - 9 m
7 - 8 m
6 - 7 m
5 - 6 m
4 - 5 m
3 - 4 m
2 - 3 m
1 - 2 m
0 - 1 m
0 400 800 1200
frequency
Figure 2.
Examples of broad-leaved (top) and coniferous (bottom) vegetation plots. Each row shows a photograph of the plot, a 3-D profile of classified echo
ttributes of vegetation returns, and a histogram of ALS-derived vegetation return frequencies based on height above the ground. a

K. IOKI ET AL.
Table 2.
Parameter settings used for the classification of ground returns.
Ground Classification Parameter Setting
Max Building Size 60 m
Iteration Angle 12˚
Iteration Distance 1.5 m
Terrain Angle 88˚
VCl
1
VCl
2
VCl
m–12
VCl
m–11
VCl
m–8
VCl
m–7
VCl
m–4
VCl
m–3
VCl
m
CUMVCI 12
CUMVCI 8
CUMVCI 4
CUMVCI 0
12 m
8 m
4 m
0 m
Figure 3.
Schematic showing the calculation of the cumulative vegetation cover-
age index (CUMVCI) for different stratum ranges. VCI: vegetation cov-
erage index.

FO
1
FO FO
11
FO ii
imi
kkkk
kk
nn
nn nn




 

, (2)
where nF is the number of first echoes, nO, is the number of only
echoes, i is the number of the target stratum (counted at 1-m
intervals from the top stratum), and m is the total number of
strata. As both the first and singular echo attributes are the first
objects detected, we assumed that FO is related to the overstory
canopy density; 2) The second parameter (IL) is related to the
intermediate and last echo attributes; when i ≥ 2, IL is
I
-1
F
1
IL i
ii
k
k
nn
n



Li
, (3)
where nI is the number of intermediate echoes and nL is the
number of last echoes. In the model, the intermediate and last
echo attributes for the first stratum were ignored, because IL
represents the proportion of the intermediate and last echoes
relative to the total number of first echoes obtained from the
previous stratum. The intermediate and last echoes are subse-
quent detections that follow the first detection of objects.
Therefore, we assumed that IL is related to the density of the
sub-canopy or understory.
Second, the vegetation coverage index (VCI) of each stratum
was calculated as the average of FO and IL, weighted according
to the magnitude of the denominator of each parameter. For i ≥
2, we have



11
FO FO
11
11
FO FOF
11
1
F
1
11
FO FOF
11
VCI FO
IL
ii
kkkk
kk
imi
kkkk
kk
i
k
k
mi i
kk kkk
kk k
nn nn
nn nnn
n
nn nnn







 
 


 




 



1
1
i
k
k



 
(4)
and for i = 1, we have
VCI1 = FO1. (5)
The cumulative sum of the vegetation coverage index
(CUMVCI) from the hth to the ith stratum (for h < i) was cal-
culated as

1
11
CUMVCIVCIVCI1 VCIk
j i
hih j
jh k

 

 




. (6)
This formula is analogous to the cumulative effects of semi-
transparent sheets of light.
Finally, CUMVCI for each stratum was obtained for com-
parison with field measurements. Above 12 m,
CUMVCI 12 = 1CUMVCIm–12; (7)
from 8 to 12 m,
CUMVCI 8 = m–11CUMVCIm–8; (8)
from 4 to 8 m,
CUMVCI 4 = m–7CUMVCIm–4; (9)
and from 0 to 4 m,
CUMVCI 0 = m–3CUMVCIm. (10)
Statistical Analysis
Calculated CUMVCI 12, CUMVCI 8, CUMVCI 4, and
CUMVCI 0 by the ABW method were compared with field
measurements of the corresponding stratum ranges. For each
stratum range, we performed a linear regression (yielding the
slope, the intercept, and their 95% confidence intervals (CI))
and calculated the root mean square error (RMSE). The predic-
tion error was calculated as the difference between field meas-
urements and values predicted from ALS data. All analyses
were performed with R software, v. 2.12.0 (R Development
Core Team, Austria).
Results
The CUMVCI of all stratum ranges were compared with
field measurements. Results of the linear regression analyses
are summarized in Tables 3 and 4 and Figures 4 and 5. In
broad-leaved stands, for the stratum range >12 m, the RMSE
was 0.164, corresponding to 37% of the mean observed value.
For the stratum range, 8 - 12 m, the RMSE was 0.230, corre-
sponding to 41% of the mean observed value. For the stratum
range, 4 - 8 m, the RMSE was 0.393, corresponding to 56% of
the mean observed value. The lowest stratum range, 0 - 4 m,
showed the highest RMSE (0.562), corresponding to 85% of
the mean observed value.
ALS model underestimation of the vegetation index was ob-
served in cases where points plotted below the 1:1relationship
line (Figure 4). The slopes of regression lines were close to 1
92

K. IOKI ET AL.
Table 3.
Results of linear regression analyses for each stratum range in broad-leaved stands. Regressions are shown in Figure 4.
Estimate p-value Lower 95% Upper 95%
Above 12 m Slope 1.043 1.12 × 10−6*** 0.771 1.315
Intercept 0.049 4.94 × 10−1 −0.099 0.197
8 - 12 m Slope 0.606 6.78 × 10−3** 0.196 1.015
Intercept 0.172 1.73 × 10−1 −0.083 0.427
4 - 8 m Slope 0.631 2.97 × 10−1 −0.607 1.870
Intercept −0.017 9.68 × 10−1 −0.901 0.866
0 - 4 m Slope 0.127 5.87 × 10−1 −0.357 0.611
Intercept 0.038 8.10 × 10−1 −0.289 0.364
p-value: ***< 0.001, **< 0.01, *< 0.05.
Table 4.
Results of linear regression analyses for each stratum range in coniferous stands. Regressions are shown in Figure 5.
Estimate p-value Lower 95% Upper 95%
Above 12 m Slope 0.713 4.07 × 10−3** 0.290 1.137
Intercept 0.307 6.37 × 10−2 −0.017 0.632
8 - 12 m Slope 0.982 2.19 × 10−5*** 0.692 1.272
Intercept −0.043 5.15 × 10−1 −0.184 0.098
4 - 8 m Slope 0.195 2.94 × 10−2* 0.026 0.364
Intercept 0.024 4.63 × 10−1 −0.047 0.095
0 - 4 m Slope 0.045 3.96 × 10−1 −0.067 0.158
Intercept 0.018 4.58 × 10−1 −0.033 0.069
p-value: ***< 0.001, **< 0.01, *< 0.05.
(a) (b)
(c) (d)
Figure 4.
Observed vegetation coverage versus predicted values of the cumulative vegetation coverage index
(CUMVCI) for stratum ranges in broad-leaved plots: (a) above 12 m; (b) 8 - 12 m; (c) 4 - 8 m; and
(d) 0 - 4 m. Regression lines are solid; dashed lines show 1:1 relationships.

K. IOKI ET AL.
(a) (b)
(c) (d)
Figure 5.
Observed vegetation coverage versus predicted values of the cumulative vegetation coverage index
(CUMVCI) for stratum ranges in coniferous plots: (a) above 12 m; (b) 8 - 12 m; (c) 4 - 8 m; and (d) 0
- 4 m. Regression lines are solid; dashed lines show 1:1 relationships.
in upper stratum ranges, but decreased for stratum ranges closer
to the ground. The slopes (±95% CI) for the stratum ranges >12
m, 8 - 12 m, and 4 - 8 m were, respectively, 1.043 ± 0.148,
0.606 ± 0.409, and 0.631 ± 1.238; intercepts (±95% CI) for
these stratum ranges were, respectively, 0.049 ± 0.148, 0.172 ±
0.255, and −0.017 ± 0.884. There were no significant differ-
ences between observed and predicted values of the vegetation
index, as the 95% CIs included slopes of 1 and intercepts of 0.
However, for the stratum range 0 - 4 m, the 95% CI for the
slope did not contain the value 1 (Table 3).
In coniferous stands, for the stratum range >12 m, the RMSE
was 0.122, corresponding to 16% of the mean observed value.
For the 8 - 12 m stratum range, the RMSE was 0.133, corre-
sponding to 34% of the mean observed value. RMSE values in
the stratum ranges 4 - 8 m and 0 - 4 m were particularly high,
0.322 and 0.419, respectively, corresponding to 99% and 106%
of the mean observed values, respectively (Figure 5). The
slopes (±95% CI) for the stratum ranges >12 m and 8 - 12 m
were, contain the slope of 1; intercepts (±95% CI) for these
stratum ranges were, respectively, −0.043 ± 0.141 and 0.024 ±
0.070, both of which contained the intercept of 0. However, for
the remaining stratum ranges (4 - 8 m and 0 - 4 m), slopes of
the regression lines were significantly different than 1.
Figure 6 shows the mean percentage of available vegetation
returns and the RMSE for each stratum range. In coniferous
stands, available vegetation returns decrease dramatically from
the stratum range >12 m to the 4 - 8 m range; thus, only a small
percentage of available returns are available for the lower two
stratum ranges. In broad-leaved stands, the percentage of avail-
able vegetation returns decreases more gradually than in conif-
erous stands.
Discussion
We investigated the use of ALS data (proportion of returns
distinguished by attributes: first, last, intermediate, and singular)
for estimating stratified vegetation cover. The model was ap-
plied to broad-leaved and coniferous forest stands. In the case
of broad-leaved stands, the model gave reproducible estimates
of vegetation coverage within the upper stratum ranges (>12 m
and 8 - 12 m); in these ranges, RMSE values were 0.164 and
0.230, respectively, corresponding percentage of mean ob-
served values remaining less than 50%. For the stratum range 4 -
8 m, although about half the points plotted close to the 1:1 line,
a number of points clustered below this line, indicating that the
ALS model under estimated vegetation coverage (Figure 4(c)).
In coniferous stands, the estimation accuracy was acceptable in
the first and second stratum ranges (>12 m and 8 - 12 m); how-
ever, it dropped severely in the lower stratum ranges. RMSE
values increased from 0.133 in the stratum range 8 - 12 m (cor-
responding to 34% of the mean observed value) to 0.322 in the
94

K. IOKI ET AL.
(a)
(b)
Figure 6.
Mean percentage of available vegetation returns and the root mean
square error (RMSE) in each stratum range, for (a) broad-leaved stands
and (b) coniferous stands.
stratum range 4 - 8 m (corresponding to 99% of the mean ob-
served value). In the regression analysis, the CIs for the slope
and the intercept did not contain the values of 1 and 0, respec-
tively, in the lower stratum ranges (4 - 8 m and 0 - 4 m); this
implies that the number of available returns was insufficient to
accurately estimate vegetation coverage in lower stratum ranges
(Figure 7).
It is likely that the underestimation was caused by a defi-
ciency of return pulses. Figure 7 shows the relationship be-
tween errors and the ratio of the cumulative sum of the vegeta-
tion returns to all vegetation returns for the second (8 - 12 m)
and the third (4 - 8 m) stratum ranges. The errors were distrib-
uted rather evenly until approximately 90% of the total return
had been incorporated into the model in both broad-leaved and
coniferous stands. However, results show consistent underesti-
mation after approximately 95% of the total returns had been
used. This indicates that there are not sufficient available re-
turns for the model, which resulted in the failure of the estima-
tion.
The difference in pulse return distributions between broad-
Figure 7.
Errors in the 8 - 12 m and 4 - 8 m stratum range plotted against cumula-
tive percentage of vegetation returns above 4 m in height.
leaved and coniferous stands observed in this study can be ex-
plained by the canopy shape and foliage distribution. Previous
studies pointed out that crown shape may affect the height dis-
tributions of ALS data (Nelson, 1997; Næsset, 1997, 2004; Holm-
gren & Persson, 2004; Moffiet et al., 2005) and pulse penetra-
tion of the upper canopy depends on density, reflectivity and
orientation of leaves and canopy closure (Gaveau & Hill, 2003,
Clark et al., 2004). Moffiet et al. (2005) described differences
in laser penetration between species of the broad-leaved poplar
box and the coniferous cypress pine; while poplar box exhibits
larger gaps in crown foliage, the needle density of cypress pine
provides less opportunity for a significant second reflection. In
this study, vegetation coverage in the highest strata of conifer-
ous stands is relatively dense; this pattern is a result of planta-
tion without thinning, so that stands developed dense foliage
canopy surfaces. This was probably a reason why there were
less pulse returns for the lower layers in the coniferous stands.
Furthermore, Persson et al. (2005) compared the point clouds
that are extracted using the waveform signal with the point
clouds recorded from the conventional discrete-return ALS
system in pine, spruce and deciduous stands. The results re-
vealed that the increase in number of points extracted from
waveform compared to points delivered by the discrete-return
system was at 57% for deciduous stand, whereas pine and
spruce stands recorded at 18% and 30% respectively. This may
indicate that full-waveform data can be used to give a better
description of the vertical vegetation distribution in broad-
leaved forests. More training data sets with more variations
need to be collected in different conditions (e.g., species, age)
for further analysis.
Conclusion
This paper proposed a new technique, the ABW method, for
reproducing the vertical distribution of vegetation cover using
leaf-on ALS data of warm temperate forests. This approach is
unique in its ability to predict and quantify vegetation cover as
a function of vertical stratum range. Notably, the models gave
reasonable estimates of vegetation cover until approximately

K. IOKI ET AL.
96
95% of the total number of ALS returns had been incorporated
into the model. The results in this study show that the presented
method can be used for estimating the vertical distribution of
vegetation cover in the upper two or three stratum ranges in
broad-leaved stands, and in the upper two stratum ranges in
coniferous stands.Although CUMVCI was calculated for four
stratum ranges in this study, the approach can accommodate
different numbers of stratum ranges.Future research should
examine the applicability of the model to different forest types.
Acknowledgements
The authors wish to thank the Organization for Landscape
and Urban Green Technology Department, Japan for financial
support. We are also grateful to the Imperial Household Agency,
Kyoto Office, for assistance with ALS data acquisition and
field observations.
REFERENCES
Axelsson, P.E. (2000). DEM generation from laser scanner data using
adaptive TIN models. International Archives of the Photogrammetry
and Remote Sensing, 33, 110-117.
Braun-Blanquet, J. (1932). Plant sociology: The study of plant commu-
nities (English translation). New York: McGraw-Hill.
Brokaw, N. V. L., & Lent, R. A. (1999). Vertical structure. In I. Hunter,
& L. Malcom (Eds.), Maintaining biodiversity in forest ecosystems
(pp. 373-399). Cambridge: Cambridge University Press.
doi:10.1017/CBO9780511613029.013
Clark, M. L., Clark, D. B., & Roberts, D. A. (2004). Small-footprint
lidar estimation of sub-canopy elevation and tree height in a tropical
rain forest landscape. Remote Sensing of Environment, 91, 68-89.
doi:10.1016/j.rse.2004.02.008
Coops, N. C., Hilker, T., Wulder, M. A., St-Onge, B., Newnham, G.,
Siggins, A., & Trofymow, J. A. (2007). Estimating canopy structure
of Douglas-Fir forest stands from discrete-return LiDAR. Trees, 21,
295-310. doi:10.1007/s00468-006-0119-6
Gaveau, D. L. A., & Hill, R. A. (2003). Quantifying canopy height
underestimation by laser pulse penetration in small-footprint airborne
laser scanning data. Canadian Journal of Remote Sensing, 29, 650-
657. doi:10.5589/m03-023
Lefsky, M. A., Cohen, W. B., Acker, S. A., Parker, G. G., Spies, T. A.,
& Harding, D. (1999). Lidar remote sensing of the canopy structure
and biophysical properties of Douglas-Fir Western Hemlock Forests.
Remote Sensing of Environment, 70, 339-361.
doi:10.1016/S0034-4257(99)00052-8
Hashimoto, H., Imanishi, J., Hagiwara, A., Morimoto, Y., & Kitada, K.
(2004). Estimating forest structure indices for evaluation of forest
bird habitats by an airborne laser scanner. In M. Thies, B. Koch, H.
Spiecker, & H. Weinacker (Eds.), Laser scanners for forest and
land- scape assessment: Proceedings of the ISPRS Working Group
VIII/2, Freiburg, 3-6 October 2004, 254-258.
Hill, R. A., & Broughton, R. K. (2009). Mapping the understory of de-
ciduous woodland from leaf-on and leaf-off airborne LiDAR data: A
case study in lowland Britain. ISPRS Journal of Photogrammetry
and Remote Sensing, 64, 223-233.
doi:10.1016/j.isprsjprs.2008.12.004
Holmgren, J., & Persson, Å. (2004). Identifying species of individual
trees using airborne laser scanner. Remote Sensing of Environment,
90, 415-423. doi:10.1016/S0034-4257(03)00140-8
Hug, C., Ullrich, A., & Grimm, A. (2004). LITEMAPPER-5600: A wave-
form-digitizinglidar terrain and vegetation mapping system. Remote
Sensing and Spatial Information Sciences, 36, 24-29.
MacArthur, R. H., & MacArthur, J. W. (1961). On bird species diver-
sity. Ecology, 42, 594-598. doi:10.2307/1932254
Magnussen, S., & Boudewyn, P. (1998). Derivations of stand heights
from airborne laser scanner data with canopy-based quantile estima-
tors. Canadian Journal of Forest Research, 28, 1016-1031.
doi:10.1139/x98-078
Maltamo, M., Packalén, P., Yu, X., Eerikäinen, K., Hyyppä, J., & Pit-
känen, J. (2005). Identifying and quantifying structural characteris-
tics of heterogeneous boreal forests using laser scanner data. Forest
Ecology and Management, 216, 41-50.
doi:10.1016/j.foreco.2005.05.034
McElhinny, C., Gibbons, P., Brack, C., & Bauhus, J. (2005). Forest and
woodland stand structural complexity: Its definition and measure-
ment. Forest Ecology and Management, 218, 1-24.
doi:10.1016/j.foreco.2005.08.034
Miura, N., & Jones, S. D. (2010). Characterizing forest ecological struc-
ture using pulse types and heights of airborne laser scanning. Remote
Sensing of Environment, 11 4, 1069-1076.
doi:10.1016/j.rse.2009.12.017
Moffiet, T., Mengersen, K., Witte, C., King, R., & Denham, R. (2005).
Airborne laser scanning: Exploratory data analysis indicates potential
variables for classification of individual trees or forest stands ac-
cording to species. ISPRS Journal of Photogrammetry and Remote
Sensing, 59, 289-309. doi:10.1016/j.isprsjprs.2005.05.002
Nelson, R. (1997). Modelingforest canopy heights: The effects of can-
opy shape. Remote Sensing of Environment, 60, 327-334.
doi:10.1016/S0034-4257(96)00214-3
Nilsson, M. (1996). Estimation of tree weights and stand volume using
anairborne lidar system. Remote Sensing of Environment, 56, 1-7.
doi:10.1016/0034-4257(95)00224-3
Næsset, E. (2002). Predicting forest stand characteristics with airborne
scanning laser using a practical two-stage procedure and field data.
Remote Sensing of Environment, 80, 88-99.
doi:10.1016/S0034-4257(01)00290-5
Næsset, E. (2004). Practical large-scale forest stand inventory using a
small-footprint airborne scanning laser. Scandinavian Journal of
Forest Research, 19, 164-179. doi:10.1080/02827580310019257
Næsset, E., & Økland, T. (2002). Estimating tree height and tree crown-
properties using airborne scanning laser in a boreal nature reserve.
Remote Sensing of Environment, 79, 105-115.
doi:10.1016/S0034-4257(01)00243-7
Persson, Å., Söderman, U., Töpel, J., & Ahiberg, S. (2005). Visualiza-
tion and analysis of full-waveform airborne laser scanner data. Pro-
ceedings of ISPRS Workshop on Laser Scanning 2005, Enschede, 12-
14 September 2005, 103-108.
R Development Core Team (2010). R: A language and environment
forstatistical computing. Vienna: R Foundation for Statistical Com-
puting.
RIEGL LMS GmbH (2009). Full Waveform Analysis Software RiANA-
LYZE for RIEGL Airborne Laser Scanners LMS-Q560 and LMS-
Q680.
Shidei, T. (1974). Forest vegetation zones. In M. Numata (Eds.), The
flora and vegetation of Japan (pp. 87-124). Tokyo/Amsterdam: Ko-
dansha/Elsevier Scientific Publishing Company.
Soininen, A. (2003). Terra Scan User’s Guide.
Zimble, D. A., Evans, D. L., Carlson, G. C., Parker, R. C., Grado, S. C.,
& Gerard, P. D. (2003). Characterizing vertical forest structure using
small-footprint airborne LIDAR. Remote Sensing of Environment, 87,
171-182. doi:10.1016/S0034-4257(03)00139-1