Desertification monitoring in Naiman-Qi, Inner Mongolia, China in the 1960's was conducted using satellite imagery of reconnaissance satellite of USA, CORONA. Imagery observed on 12th September 1961 and on 30th November 1970 were used in this study. The map developed by Chinese Academy of Science was utilized for supervised classification. The result clarified that in the north west part of the study area, desertification had markedly processed between 1961 and the 1990's. Except the north west area and river terrace along the Jiorai River where desertification almost stopped after the 1980's by artificial forestation. Therefore no much new desertified area had emerged in the northern part of Naiman-Qi. Remedies of desertification recommended to concentrate into north west area with enough care taken to river terrace along the Jiorai River after the series of studies completed by this study.

View full abstract

Recently, reduction of the construction cost and insufficiency of skilled hands are becoming serious problems on the construction plant from the point of view of the work efficiency. With this motive, the authors have been developing an ERGOMA system for improvement of working environment or evaluation of the work efficiency on the construction plant using sequential images. However, visualization or animation procedures are needed for easy understanding of human motion, for improvement of working environment or evaluation of the work efficiency on an ERGOMA system.
This paper describes visualization of human motion using animation techniques, and investigates working load on the waist on the construction plant using 3D data of the waist and the head (helmet) obtained from video image sequences.

View full abstract

Vegetation cover types on Changbai mountain, a natural biosphere reserve (2000km²) in northeast China, were derived by using multisensor satellite imagery combined with Landsat TM and JERS1 UPS. DEM data were used for improving classification accuracy. Cover types were classified into 20 groups. Bands 4 and 5 of Landsat TM image acquired on July 18, 1997 and band 1 of JERS1 UPS image acquired on Feb. 4, 1997 were fused to a false color image for the final output, and maximum likelihood supervised classification was performed. Data fused either with multitemporal (OPS) or with multisensor (TM+OPS) showed a high accuracy of identification, comparing to individual images. The overall accuracy of classification of individual images of OPS presented less than 40%, and TM less than 70%, while the fused data set provided an accuracy higher than 73% which was raised to 83% by post classification including filtering and verification with DEM.
There were 5 vegetation zones on the mountain, from the base to the peak, deciduous forest zone, mixed forest zone, conifer forest zone, birch forest zone, and tundra zone. Spruce-fir conifer forest was the most dominant (nearly 50%) vegetation type, followed with mixed forest (10%) and larch forest (8%) . Classification accuracy was not only determined by data fusion from different sensors of different resolutions, but was also affected by image composite from different seasons. In winter images, the contrast between cover types in life-form level, e.g. evergreen and summer-green, or forest and meadow, is enhanced, while more detailed information of spectral characteristics of plant communities can be extracted from summer images. For boreal vegetation, autumn imagery is considered useful for discriminating cover types, because the leaf color in this season is significantly diversified. It is concluded that, for vegetation consisting of conifer and deciduous species, multitemporal imagery fused from phenologically different data is meaningful for creating vegetation maps with high accuracy.

View full abstract

An algorithm to extract drainage networks by using digital elevation model (DEM) was proposed. It has three characteristic points, i.e., pre-specification of inland water areas, shortest path for outlet in flat areas, and restriction of filling level for depression areas. The algorithm was applied to DEM with 50m grid size covering Tohoku district of Japan, which has about 35 million inland grid points. The extracted drainage networks were relevantly compatible to those described in 1/25, 000 topographical maps, and also included a lot of finer drainages in rugged areas. The processing time by using HP9000/C3000 workstation system, however, was about 25 minutes.

View full abstract

The method for selection of appropriate band combinations for improvement of Sea Surface Temperature (SST) estimation accuracy is proposed. It is found that the proposed method for selection of appropriate band combinations for maximizing SST estimation accuracy using radiative transfer code, MODTRAN-3.7 in this case is useful to investigate a most appropriate SST retrieval algorithm as well as choosing an additional band to the existing band combinations. In comparison between the band combination of the existing NOAA/AVHRR and the results from this study, it is found that 60.2 to 98.4 [%] of improvement on SST estimation accuracy can be achieved for the proposed band combinations. Also it is found that 12 [μm] band would be better to shift to the longer wavelength. Meanwhile, the center wavelength of 3.7 [μm] band would be better to shift to the longer wavelength.

View full abstract