By assuming that the ground reflectance is constant, variations of satellite signal may be attributed to variations of the atmospheric optical properties. Based upon this, Tanre et al., 1988 has defined the structure function for deriving aerosol optical depth from TM images. Holben (1992) also has gotten the similar results using AVHRR data. Basically, the structure function is defined by single direction only. It can't represent the complete properties of the surface structure. To over­come this, a multi-directional structure function, which should include more information than the single directional from the surface structure, is used in this paper to improve the structure function method for SPOT data. In general, the error of aerosol optical depth determination could be enlarged by the different pattern of structure function distribution which is caused by the characteristics of local land­covers, terrain and the high resolution property of SPOT data. In order to reduce such error, it is necessary to determine an optimum number of distance to truncate the different pattern of structure function distribution. After these improvements, the results of this study are reasonable when compared with the Dense Dark Vegetation method's results.

Mountainous relief can affect the growth and distribution of vegetation. This must be considered when using remote sensing techniques to detect forest vegetation. This study use factorial experiment method to examine the effects of relief factors and land cover on TM data. It aims to upgrade the efficiency of forest monitoring by use of the satellite image. Relief factors used in this study are aspect, slope, surface roughness, sky view factor (SVF), terrain configuration factor (TCF), and direct light ratio (DLR). Results reveal that there are significant relationships between these relief factors, in which the spectral information explained by the aspect was different from the others. The infrared band (TM4) data is most sensitive to aspect variation in all of the TM bands. Few TM data variation explained by SVF, DLR, and TCF in the case of land cover, aspect, surface roughness, and slope factors exist. So, these three factors could be neglect in related forest studies with remote sensing. The significant interactions between land cover, aspect, surface roughness, and slope factors have proven that vegetation could have very different spectral value in different conditions of such factors. That is why low classifi­cation accuracy occurs in most of the forest studies with satellite image in Taiwan. Surface roughness is present for the relief information in a pixel and is an important factor, which could affect the variation of TM data. Therefore, it is necessary to generate a better preci­sion of DTM data. After that a more perfect experiment could conduct to get a better understanding of mountainous conditions. Based on the pre-list findings, we select a lot of training samples to accomplish maxi­mum likelihood classification. The overall accuracy is about 90 percent. The spectral charac­teristics of conifer and broad-leaf trees are very similar, but the green and the first mid-infrared band data (TM2 and TM5) still could distinguish them. Red, infrared, and the second mid-infrared channel data could extract tea field, grass, and farmland. The spectral characteristics of bamboo and bare land are very distinctive from other cover type in all reflective TM bands.

This study compares two fractal interpolation schemes based on fractional Brownain motion (fBm) with two conventional interpolation schemes: bilinear interpolation and cubic convolution. Numerical experiments are performed with one pair of digitized stereo photographs. The original image is reduces to a lower spatial resolution for interpolation. The statistic indices such as the root mean square error, average error, maximum devia­tion, and correlation coefficient derived by comparing interpolated and the reference images are applied for the evaluation. Besides, an image matching based on least squares is also utilized for comparison. All quality measures indicate that the conventional interpolation schemes are better than the fractal-based schemes.

More than two thirds of Taiwan Island are covered by rugged terrain. The most difficult part of constructing a sound national highway system in Taiwan is to manage the difficulties of crossing the mountainous terrain, where geological hazard-prone areas are expected but existing geological information are still lacking. Therefore, remotely-sensed data have been employed very extensively since 1989 to give information on geo-hazards. It was also found that the usage of air-photos taken 40 years ago, can be readily identified to give indications of geological sensitive areas such as abandoned river courses and ancient ponds in the level and hilly terrain, where now are covered by recent deposits and are highly cultivated. It is, therefore, concluded that when conventional aerial photographs are employed together with satellite images, the best benefit by using remote sensing techniques can be gained. In mountainous terrain, images can be used to locate potential land­slides; and, in hilly or level terrain, images can be used to identify weak and sensitive geological units. This information is valuable for assessing alternative highway corridors.

The first Glossary of Surveying and Mapping in Chinese is being compiled by National Institute for Compilation and Translation. The Glossary Committee consists of 40 profes­sors and experts. There are approximately one thousand terms of photogrammetry and remote sensing included in the glossary. The terms listed below are to be discussed for refinement of their definitions. These terms are scanning、orientation、digit、pass point、tie point、DTM、DEM、image、scale point、stereomate、data、continuous strip photograph、holography、affine plotting、and etc.