2004
ction
ay, it
1 the
3 r0f
t the
e can
, the
pixel
tio of
WS:
1gle
»nded
ynded
Index
rval’s
ight)
03
| 8:56
| (fig.4
As for
as 0.2-
visual
ue sky
These
refore,
jy the
| RGB
as one
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
isp g 0 0.2 0.4 0.6
Hemisphere RGB Image louds Detection
Sky Index EE TI
Figure 5. RGB, Sky Index and Clouds detection images
The clouds detection image was generated by the definition that
the values of less than 0.23 show clouds area. The grey scale
shows the brightness of clouds (the darker area shows the
brighter cloud). Comparing with both RGB and clouds
detection images, it shows the clear extraction of clouds area.
4.2 Cloud distribution and cloud cover
After generating the clouds detection images, the grid surfaces
of the zenith and the direction angles are overlaid in order to
visualize the cloud distribution. Figure 6 illustrates examples of
cloud distributions on hemisphere coordinates with sun orbit on
7* of February, 2003.
Hemisphere Coordinates
24 Directions / 10deg. Int.
of Zenith angles
c) 13:17
Figure 6. Cloud distributions on hemisphere coordinates
Furthermore, the results of cloud cover estimations were 0.55,
0.40 and 0.64 for a) 10:45, b) 11:54 and c) 13:17 in fig.6
respectively. According to the cloud distribution of fig. 6 a), the
detected clouds are distributed almost eastern half side and near
the horizon of western side, it looks cloud cover is more than
0.6 or 0.7, but estimated cloud cover is 0.55. This is represented
efficiently about the cosine correction of zenith angle. We also
can recognize different pattern of cloud and distributions from
Figure 6 visually.
829
5. CONCLUSION
Clouds have directly the strong impact to the solar radiation
environment. Accordingly, understanding the cloud condition is
indispensable for estimating global radiation consists of direct,
defuse and reflected radiations. We have observed cloud
conditions in whole sky in order to grasp cloud distribution an
estimate cloud covers in short time interval. In this paper, the
whole sky observation and the methods how to detect clouds
area from hemisphere imageries üsing commercial digital
camera and estimate cloud cover based on image analysis.
Results of this study will be useful for automatic estimation of
cloud cover and visualization of cloud distribution at the real
time.
As the continuous works, we will examine how cloud cover and
distributions influence to global radiation. Furthermore we will
set more than two observation sites in order to estimate cloud
volume and spatial distribution of clouds in regional area.
REFERENCES
Charles N. Long and Thomas P. Ackerman. 2000.
Identification of clear skies from broadband pyranometer
measuremsnts and calculation of downwelling shortwave clouds
effects. Journal of Geophysical Research, Vol.105, No. D12,
pp.15609-15626.
Gabriela Seiz, Emmanuel P. Baltsavias, and Armin Gruen.,
2002. Cloud Mapping from the Ground: Use of Photgrammetric
Methods. Photogrammetric Engineering & Remote Sensing,
Vol.68, No.9, pp.941-951.
Nakashizuka T., Lee H. S., Chong L., 2001. Studies on canopy
processes of a tropical rain forest in Lambir Hills National Park.
Proceedings of the International Symposium, Canopy Process
and Ecological Roles of Tropical Rain Forest, Miri, Malaysia,
pp.2-7.
Yoshimura M., Yamashita M., 2001. Multi-scale approach for
rainforest environment monitoring by remote sensing / GIS/
direct measurements — towards virtual field construction -.
Proceedings of the International Symposium, Canopy Processes
and Ecological Roles of Tropical Rain Forests, Miri, Malaysia,
pp96-101.
Oshita K., 2000. OP Fisheye-NIKKOR 10mm F5.6, Nikon Web
Magazine.
http://www.nikon.co.jp/main/jpn/society/nikkor/n06j.htm
(Japanese only)
