The heat content (H) of the ocean upper layer may be calculated directly by integrating the water temperature over depth. Its anomaly may be estimated by the sea surface height anomaly (SSHA) that is associated with the thermal expansion effect of the water column. In the tropical eastern Pacific Ocean, we may use the Tropical Atmosphere and Ocean (TAO) temperature array to estimate the heat content anomaly with root-mean-square deviation (RMSD) of 23x107(J/m2 ), or use the SSHA of TOPEXIPOSEIDON (TP) to estimate with RMSD of 20 X 107~30 x107 (J/m2 ). High correlation between these estimates was found in the earlier studies. This assures that we may estimate the heat content anomaly of the ocean upper layer with altimeter data. Because the mixed layer and thermocline dominate the temperature profile, H is mostly determined by the sea surface temperature (SST) and the mixed layer depth (h) if the shape of temperature profile in the thermocline does not change much. With this assumption, we use the estimated SST from Geostationary Operational Environmental Satellite (GOES) and the computed SSHA from TP to derive the temporal change of mixed layer depth h. The quality of this method in estimating h is assessed with the in situ measured h by the TAO array of temperature sensors. It was found that using GOES SST and TP' SSHA to estimate h is much better than nocast data, i.e. the long term mean h. It is also found that the accuracy of GOES SST varies with region, therefore the RMSD of h has regional dependence.

Multi-temporal imageries and cadastral GIS datasets were combined to interpret the rice paddy distributions, by applying the following means: (1) NDVI intervals approach on basis of multi-temporal SPOT images, (2) the Bayesian classifier based on spectral reflectance curve measured from different growth stages, (3) the statistic probability classifier, (4) the object oriented fuzzy classifier. Results of last three classification approaches were coupled with a prediction of probability. A selected threshold was applied to the results to obtain a dichotomy category. These results were compared with visual interpreting data from aerial photos for assessing accuracy. It " shows that the overall accuracies and k ̂ index for the four means are 92.22% & 0.83, 96.27% & 0.92,91.89% & 0.82 and 90.97% & 0.81, respectively. Similar approach is applied to a contrast test site for the rice paddy in the first season. The results of the above-mentioned classifiers show overall accuracies and k ̂ index of 94.27% & 0.83, 95.43% & 0.87 and 92.75% & 0.79 that deducted from the object oriented fuzzy classifier. For the rice paddy in the second season, results of the above-mentioned three classifiers show overall accuracies and k ̂ index of 93.99% & 0.79, 95.33% & 0.84 and 93.47% & 0.77. There are the same classified results that applied to Tainan and Pindong County for rice paddy interpretation. It is concluded that the Bayesian classifier not only can generate probability of each class, but also give better classified results after thresholding process.

The previously proposed Least-squares Model-image Fitting (LSMIF) algorithm for Model-based building extraction (MBBE) is an iterative solution using Newton's method, which needs good initial approximations of the unknown parameters. This thesis, therefore, tailored a Genetic Algorithms (GA) to be a new model-image fitting method for MBBE, which does not need good approximations. We focus on the design of fitness function and the assessment of fitting results. In the approach, the initial population is generated randomly in the predefined parameter domain. Consequently, this population of primitives are transformed to object coordinate system with pose and shape parameters and projected to photo coordinate system with the known exterior orientation parameters. We propose a fitness function, which take the number of edge pixels in the buffer, the length of projected edge line into consideration as well as the sum of the perpendicular distances from the edge pixels to the projected edge lines. After a series of reproduction, crossover, and mutation operators, the solution is found by keeping the evolved model with highest fitness value in each generation. Our experiments show that GA can correctly and globally find the near-optimal solution.

The purpose of this study is to establish a muti-scales forest-land classification decision support system under the ecosystem management hierarchical principle. The mathematic programming method is used to assist the fuzzy logic approach to rationalize the criteria of classification in the knowledge-base, and this system can help the manager to make decisions in the different management levels. Meanwhile, this system is used to estimate the potential area of Taiwania. The conclusions can be draw as the followings: 1. Forest-land classification decision support system can help us to program the forest land efficiently. This system provides the decision reference to the managers in different levels and integrates the management model among different hierarchical levels. 2. Fuzzy logic is more correspond with the environment characteristic in theory, therefore the reasonable classification result can be obtained by the combination of mathematic programming and fuzzy logic criteria in the knowledge-base.3. The assessment of Taiwania potential area with forest-land classification decision support system is approximate to the actual Taiwania distribution. It shows that the system is useful to site selection.

Due to the huge demand and the importance of 3-D city models, an effective solution for geometrical building modeling is indispensable. From the application point of view, an interactive system for geometrical building modeling based on a semi-automatic approach is presented. A true-orthophoto rectification scheme based on the generated building models is also proposed. The data used in the proposed system are 3-D visible roof-edges that have been manually measured from a stereo-model, or the 3-D building outlines from a digital topographic map. The core technology is called the "Split-Merge-Shape" (SMS) method. The SPLIT and MERGE steps are the two main procedures for topology reconstruction from non-related roof-edges. The SHAPE step uses the available roof-edge height information to define an appropriate rooftop. With the SMS method, the topographic mapping of buildings, and geometric building modeling, can be seamlessly integrated into a unified procedure. In a conventional ortho-rectification procedure using a digital terrain model, the correction of relief displacement due to buildings is not considered. In order to restore the most complete information and to correct the relief displacement of buildings, a rigorous true-orthophoto rectification scheme is~proposed. In the proposed scheme, 3-D building models, multi-view aerial images, digital terrain models, and the sun's orientation are all utilized. Terrain and building relief displacement corrections as well as hidden area compensation, seamless mosaicking, and shadowed area enhancement are integrated into one procedure. In this paper, the SMS method is presented, including an evaluation of its feasibility, robustness, efficiency and accuracy. Finally, the generated building models are applied to true-orthophoto generation and 3-D GIS querying.

Forest protection and resources conservation are the major goals of modem forest ecosystem management (FEM). In order to promote the sustainable FEM, a suitable mechanism for forestland classification and a map of wildlife resources would be the important basis for delineating the boundary of resources conservation. Unfortunately, the complicated native environment makes the surveying of all the individuals wildlife not feasible. The quantitative index of both Shannon's and Simpson's diversity are hence not available for depicting the spatially variation of wildlife biodiversity. This study applied the species diversity as the basis of biodiversity quantification and developed EPD and EPR indices for representing the wildlife's diversity. We then use the geostatistical technique to derive the spatial distribution of species diversity of the amphibians, reptilians, birds, and mammals wildlife. The geographical pattern of diversity indices of those vertebrate wildlife is also explored and the biodiversity map could then be applied for regionalizing the forestland for sustainable FEM. Results demonstrates that EPH is an identifiable index of species diversity.