International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
The accuracy of tree top measurement by LIDAR depends on
the accuracy of DSM and DEM. Therefore, we compared the
DSM and DEM by the resolution, and discussed the factors that
give influence on the accuracy of tree height measurement.
Table 5 shows the correlation coefficient between the high-
resolution data and low-resolution data of DSM points group
and DEM points group at the study sites. As a result, the
correlation coefficient of DSM was 0.834 while that of DEM
was 0.995, and therefore, DSM had slightly lower correlation
coefficient. As the correlation coefficient of DEM was very
high, it is considered that there is no large difference in the
process of generating DEM from raw data by the difference of
resolution.
When the comparison was made by tree types, it was species
and density of trees that showed large difference between DSM
and DEM. By the species of tree, the correlation coefficient of
DSM showed low value (0.372) in case of the coniferous trees,
and by the difference of density, that of DSM was showed a
slightly lower value (0.744). This is a tendency coincident with
the tree height measuring characteristics described before, and
it was suggested that the characteristics of DSM was reflected
on the characteristics of tree height measurement.
ET N ec
DSM DEM
Species b 25,918 0.836 0.996
s C 2,806 0.372 0.818
Stratum: 4,620 0.928 0.930
u 24,104 0.813 0.994
Density d 14,342 0.870 0.997
14,382 0.744 0.827
All 28,724 0.834 0.995
N: number of LIDAR points, C.C: correlation coefficient, T.T:
tree types (b=broad-leaf, c=coniferous, m=multi-storied forest,
u=uniform forest, h: high, |: low, d=dense, s=sparse)
Table 5. Correlation coefficient of DSM and DEM of
different resolution
0000000000000000000088
— n — Meters
200 0 200 400 600
Figure 5. Difference of DSM between high-resolution data
and low-resolution data
Figure 6. Picture of site A
516
Figure 5 shows the difference between DSM of high-resolution
data and that of low-resolution data. The areas where the high-
resolution data show lower value are indicated in blue color and
where the high-resolution data show higher value are indicated
in red color. In general, it is known from this figure that there
are many more areas where the high-resolution data show lower
value. Especially, large estrangement was observed between
both DSMs at site A which is sparse Zolkova trees (Figure 6).
5. CONCLUSION
In this study, we measured the tree height at the urban forests
using LIDAR data of different resolutions, and discussed the
relation between error and tree types and the measuring
characteristics by the difference of resolution. As a result, it has
become clear that LIDAR tends to measure the tree height
lower than actual height as the characteristics that does not
depend upon the resolution, and this tendency is conspicuous
especially in the case of coniferous trees, and also that the
accurate measurement is difficult in the case of low trees in
multi-storied forest because the laser is intercepted by the tall
tree story. It has also become clear that the accuracy of tree top
measurement by the resolution is greatly influenced by the
species and density of trees and high-resolution data which
have higher probability of tree top irradiation provide the higher
accuracy in the case of coniferous trees while the low-
resolution data which has low transmission rate through
branches show the higher measuring accuracy in case of low
density forest. These measuring characteristics of tree top are
reflected by DSM, and it was recognized that the measuring
accuracy by LIDAR data does not necessarily depend solely on
the resolution.
ACKNOWLEDGEMENT
We hereby express our hearty thanks to General Manager, Mr.
Toshio Sekiguchi, and Superintendent, Mr. Yoshihiro Yada, of
Koganei Park Management Office, Incorporated Foundation
Tokyo Metropolitan Park Association.
REFERENCES
Friedlaender, H. and Koch, B., 2000. First experience in the
application of laser scanner data for the assessment of vertical
and horizontal forest structures. [International Archives of
Photogrammetry and Remote Sensing, 33(B7).
Handa, M. and Teshirogi, J., 2002. Urban forest for global
warming prevention - Towards sustainable community -
Urban Green Technology, 47, pp.17-21.
Naesset, E., 1997. Determination of mean tree height of forest
stands using airborne laser scanner data. /SPRS Journal of
Photogrammetry and Remote Sensing, 52, pp.49-56.
Setojima, M., Akamatsu, Y., Funahashi, M., Imai, Y. and
Amano, M., 2002. Measurement of forest area by airborne laser
scanner and its applicability. Journal of the Japan Society of
Photogrammetry and Remote Sensing, 41(2), pp.15-26.
Yamagata, Y., Oguma, H., Sekine, H. and Tsuchida, S., 2002.
The role of remote sensing for monitoring the carbon sink
activities. Journal of the Remote Sensing Society of Japan,
22(5), pp.494-509.
