The principle of GPS surpported aerial triangulation is to use GPS to determine the coordinates of the camera exposure centers and then using them as controls. The amount of ground control points required for the aerotriangulation adjustment can thus be reduced. In the past few years, considerable researches and developments of GPS supported aerial trian­gulation have taken place in some countries. But in Taiwan, there is still no any practical experience in this field. This paper presents the first test of applying GPS supported aerial triangulation in Taiwan. The main goals of the test are to study the actual available accu­racy on check points under different control patterns, and to try to find out a proper opera­tional method in Taiwan. Due to limitations on budget, we could not purchase any existing system from well known companies. Instead, we assembled the whole system from existing photo-navigation equipments with minimum new purchasing of some necessary equipments. Based on ground check points, it is found out, that under proper control and proper weight­ing of the GPS determined camera projection centers, the GPS supported aerial triangula­tion can full fill the accuracy requirement of large scale photogrammetry (1/5000). During the experiment, we have also found out some problems which need to be taken care of in high precision applications. These problems will also be analyzed in this paper

Satellite data from Landsat Thematic Mapper (TM) and NOAA Advenced Very High Resolution Radiometer (AVHRR) were used to derive the sea surface temperature (SST) of coastal waters and thermal effluent from Hsinta Power Plant. Ground truth tempera­tures measured simultaneously from a ship as Landsat passed were used to compare with the satellite SST. Two methods were used to estimate the atmospheric correction of TM­derived SST. Local radiosonde measurements were used in Lowtran7 program to estimate the atmospheric effects, sea surface radiances and the atmospheric transmittances. The mean deviation of satellite-estimated SST from the ship-measured SST is 0.94℃ if the atmospheric correction process is applied, otherwise it will be over 7℃ . Secondly, a scheme combining NOAA-AVHRR and Landsat-TM was used to derive SST. The advan­tage of this scheme is that AVHRR-SST has included the atmospheric correction. The mean deviation between the ground truth and satellite-derived SST from this scheme is 1.01℃ for the thermal plume of Hsinta power plant, and 0.62℃ for Tseng-Wen estuarine waters, respectively.

In this study, the Experimental Forest of National Taiwan University (EFNTU) was selected as the study site. First, the subwatersheds within the EFNTU were delineated automatically, then the entire area was classified into economical forest land and non­economical forest land using multivariate statistical methods. In the economical forest land, the geographic information system was applied to select suitable planting sites for Taiwania. The results indicated that: 1. Multivariate statistical methods, such as factor analysis, clustering analysis, and discriminant analysis can be used to categorize economical forest land and non-economical forest land effectively. The protected forest for water conservation within the economical forest land was delineated, and the suitable planting sites were selected. 2. Within the economical forest land, the suitable planting sites for Taiwania were selected. The largest proportion (28.09%) of suitable planting sites was found in compartment 2, and compartment 22 had the second largest proportion (20.37%). When compared in subwatersheds, unit 75 (located in compartment 2) had the largest proportion (64.32%), and unit 78 (located in compartment 22) followed with (59%). Traditionally, the Taiwania plantation was selected based on field experience. In contrast to the traditional measure, this study utilized the environmental factors to select suitable planting sites for Taiwania. This method was found to be helpful for forest resources manager in planning plantation.

This study selected the Salisen region at Hsinyi of Nantou County as a study area, and utilized the geographic information system, vegetation analysis method and multivariate statistics, to discuss how to build an ecological information of vegetation in natural forest. In this study, the structure of plant society was explored and derived the models from the ecological vegetation parameters and geographic factors. Then these model was used to simulate the distribution maps of vegetation parameters for the study area. The results indi­cated as follow: 1. The vegetation types of study area were divided into Abies zone, Tsuga-Picea zone, Quercus zone and Machilus-Castanopsis zone by detrended correspondence analysis and table rearrangement. Besides, the B, C values of Weibull probability density function and Shannon's diversity index for each vegetation zone were calculated. It showed that the natural regeneration of each vegetation zone was fine and its stand structure was stabile. The diversity of broad-leaved forest was more abundant than coniferous forest. 2. The distribution pattern of vegetation zone, diversity index, and stand structure were simulated by discriminant analysis. The accuracy of classification was over than 70 percent. The vegetation distribution was affected by elevation, slope, moisture gradient class and direct light skyspace. The diversity was affected by elevation, aspect, moisture gradient class and geology. The aspect, slope, whole light skyspace and geology, influenced the B value of Weibull probability density function. The slope affected the C value of Weibull probability density function.