emperature
11 dominant
fic surface
e and time
ne relative
o surface
> effective
h reference
us feature.
specific
que. The
| enriched
of local
tion which
M., 1998.
spatial non-
. M., 2002.
nalysis of
Sons Ltd,
f mortality
nvironment
opment in
nce as a
g regional
graphically
Systems, 4,
, 1976. A
w tundra at
heat island.
108, pp.1-
sostatistical
niology of
resented at
the Spatial
>, Dunedin,
eat Islands.
resentation.
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B8, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
GROUND-BASED CLOUD OBSERVATION FOR SATELLITE-BASED CLOUD
DISCRIMINATION AND ITS VALIDATION
M. Yamashita *, M. Yoshimura °
* Survey College of Kinki, 1-5-9 Yata, Higashi-sumiyoshi-ku, Osaka, 546-0023, Japan -yamashita@kinsoku.ac.jp
? PASCO Corporation, Research and Development Center, 2-8-11 Higashiyama, Meguro-ku, Tokyo, 153-0043, Japan —
mairtu1698@pasco.co.jp
Commission VIII, WG VIII/3
KEY WORDS: Atmosphere, Weather, Development, Application, Radiation, Observations
ABSTRACT:
Solar radiation is a source of energy for ground surfaces. Clouds control energy budget on both local and global scales.
We propose the simple ground-based observation method for various sky conditions instead of the existing visual observation.
In the paper, we mainly describe how to discriminate clouds area on whole sky images, how to overlay the cloud areas to the
satellite-based images and the test experiment. As for the discrimination of sky conditions, we use Sky Index (SI) and Brightness
Index (BI) calculated from whole sky imageries. SI shows the extent of the blueness and grayscale and BI indicates the extent of the
brightness. Sun, cloud and blue sky areas are divided by SI and BI threshold. As an extension of ground-based observation to
satellite-based observation, clouds area discriminated by ground-based observation overlay to satellite images. Through this
experiment, the utilities of ground-based cloud observation are discussed.
The proposed ground-based cloud observation method and its satellite-based extension should be connected to satellite products with
high quality by implementation of sky and clouds observation covering a wide area.
1. INTRODUCTION
Clouds control energy budget on both local and global scales.
Generally, it is well known that lower clouds cool the surface
and upper clouds have the effects to warm the surface (Clement
et al, 2009). Cumulus clouds especially in atmospheric
boundary layer play crucial role in modulating the exchange of
radiation, heat and moisture (Freedman et al. 2001). The size
and the lifespan of cumulus clouds are small and short,
therefore it is difficult to observe and monitor them spatially
and temporally.
Satellite-based observation and its products related to clouds
properties have been useful to estimate some physical values
such incident radiation/ Photosynthetic Active Radiation (PAR)
and Fraction of PAR etc., and to be advanced various high
quality satellite products. However, there are still some
uncertainness such as cloud motion, cloud coverage, cloud type
and these effects by satellite-based observation. Therefore, the
efficient ground-based cloud observation is necessary to
validate satellite-based observation such existing Terra & Aqua/
Moderate Resolution Imaging Spectroradiometer (MODIS) and
under planning Global Change Observation Mission - Climate
(GCOM-C) / Second generation GLobal Imager (SGLI) etc.
Recently, the ground-based cloud observation systems have
been developed by using whole sky camera in world widely
instead of the existing visual observation (Calbo and Sabburg,
2008; Yamashita and Yoshimura, 2008; Heinle et al., 2010).
The purpose of this study is to discriminate the cloud and
overlay the cloud areas with satellite-based images by the
efficient and simple ground-based cloud observation using
commercial based digital camera and fish eye lens in order to
validate the satellite-based observation. In the paper, we mainly
describe how to discriminate the cloud area on whole sky
Images, how to overlay the cloud areas to the satellite-based
images and the test experiment. Through this experiment, the
utilities of ground-based cloud observation are discussed.
2. GROUND-BASED CLOUD OBSERVATION
2.1 Camera system
Automatic-capturing Digital Fisheye Camera (ADFC) is the
instruments to take and accumulate the whole sky imageries.
ADFC consists of digital camera (Nikon, Coolpix4500) with
fisheye lens (Nikon, Fisheye Converter FC-E8), waterproof
hard case and remote controlled cable connected to PC
(Tsuchida et al, 2005).
In order to decrease the influence for CCD saturation caused by
the strong sun light, the neutral density filter with 10%
transparency (FUJIFILM, ND1.0) puts between the digital
camera and the fisheye lens. The camera is set up to take
imageries at every two-minute interval by fixed exposure of the
aperture as F2.6 and shutter speed as 1/500s. The image size has
2204pixels by 1704lines with RGB colors and JPEG (1/4
compressed) format instead of RAW or Tiff format to
accumulate the large volume of image files. In this study, we
already have confirmed that the almost no influences between
JPEG (1/4 compressed) and RAW/Tiff formats after image
processing. Also, digital number (DN) of RGB channels can be
used instead of radiance by fixing aperture and exposure.
The head of digital camera of ADFC is fixed to north direction
and levelled to horizon.
2.2 Discrimination of cloud, blue sky and sun
The sky conditions consist of each state of sun, clouds and blue
sky. In order to discriminate the sky conditions from whole sky
imageries, we use two indices which show the levels of the
blueness/greyscale and the brightness calculated from the
imageries. Here, the method how to discriminate the sky
condition is described. Figure 1 shows the flowchart to classify
cloud and blue sky areas from the whole sky imagery taken at
the time (7).
