|
|
|
Digital Photogrammetry
|
Global Image Composite of ADEOS/OCTS GAC Data
2.3 Image Composite
The composite process was carried out separately over land regions and ocean regions (see Figure 4). The land/ocean mask prepared in advance determined the separation of both regions. For the land, the MVC would be applied only if channel 7 (near infrared channel) was not saturated. Among the spectral channels of the OCTS data, channel 5 corresponds to the chlorophyll absorption channel (red color region). Generally, the plants containing chlorophyll have smaller reflectance than other ground cover or clouds. When the channel 7 was saturated, the process was carried out by selecting the pixels with the minimum value in channel 5.
Such a minimum value composite method by using channel 5 was also applied to the ocean regions. But, the method could not be applied to the Polar region, because some pixels were illuminated by small Solar radiation and had very small Digital Number (DN) owing to the big Solar zenith angle. To prevent such a case, one of the thermal infrared channels (channel 11) was used. While both the ice/snow and the cloud are cold, the cloud will be colder than the ice/snow. When the channel 7 was saturated in the Polar land region, the pixels with maximum value in channel 11 would be selected. In case that the channel 7 was not saturated, the pixels seem to be illuminated by small Solar radiation. The pixels observed at the minimum solar zenith angle was selected in such a case.
The period for image composite is very important for the analyses with use of them. If the period is too short, it will very difficult to extract cloud free pixels. On the other hands, if it is too long, the results will be less sensitive to the rapid changes of vegetation. Basically, the 10 days composite and the 30 days composite were adopted in this work. The OCTS was operational only about eight months after the initial checkout period. In three months of those 8 months, the OCTS was operated without the tilt function. All observed data were processed for the image composite. After all, 20 images for 10 days composite and 8 images for 30 days composite were generated as shown in Table 3.
3. Conclusions
- The system for the
precise geometric correction of ADEOS/OCTS was
developed. The algorithm of the geometric
correction is based on the principle of
collinearity condition.
- The new algorithm for
generating a cloud free composite from OCTS data
was developed. It was applicable to the
saturated pixels in near infrared channels.
- 20 images for 10 days composite and 8 images for 30 days composite covering all observation period (8 months) were generated. It is very expected that these composite images will be used in some land environmental applications.
Acknowledgment
This work was partially supported by the 'GCMAPS program' conducted by Scientific Technology Agency of Japan.
References
- Hashimoto, T., 1997.
Precise Geometric Correction of ADEOS/OCTS
Imagery, Journal of JSPRS, Vol.36, No.5,
pp.42-51
- Hashimoto, T., 1998.
The Estimation of Motion and Attitude of ADEOS
Satellite Utilizing the Principle of Exterior
Orientation, Journal of JSPRS, Vol.37, No.6,
pp.4-13
- NASDA, 2000, home page
on ADEOS, on
http://www.eorc.nasda.go.jp/
- NOAA, 1997, NOAA GVI
GUIDE, on
http://www2.ncdc.noaa.gov/docs/gviug/
- Wessel, P., 1999,
GMT-The Generic Mapping Tools, on
http://www.soest.hawaii.edu/gmt
Figure 4 Selection of Pixel for Image Composite
Table 3 Composite images to be generated
|
|
|
|
|
|
|
