A dynamic ecosystem is compose of animal, plant, and environment. In order to keep the biodiversity and make the strategy of management, it is essential for understanding vegetation succession. In a dynamic ecosystem, there are many disturbed factors, so to know where and how they work is the most important thing for understanding vegetation succession. In this study, disturbed factors of vegetation society are decided by references and survey data, land cover types at Nanjenshan ecological reserve were interpreted from aerial photos of 1976, 1989, and 1997. Those classified digital data were used to derive Shannon diversity index for mining the dynamics change of land­scape ecosystem in different disturbed factor. T-test was applied to detect the landscape differentiability between every two periods. Results of this study show that the moisture, northeaster monsoon, road establishment, afforestation, and human exploitation are actually the disturbed factors of vegetation society. The Nanjenshan ecological reserve is effective in tending to a stable high-density forest vegetation society, but the administrative department should make proper starages for the distubed factors individually.

Vegetation ecological indices are useful for understanding the ecological diversity, physical environment, and stand structure of vegetation. This research utilized global positioning system to measure the coordinates of the base points of permanent plots designated in nature reserves. Furthermore, the locations of sample trees within the plots, slope, aspect, and several physical environmental factors of the plots were measured. Geographic information system was applied to establish a geographic database for vegetative resources using field data. The absolute coordinates of plots and sample trees were calculated by using coordinates transformation, and the thematic maps of plots and sample trees were generated automatically from the data. In addition, a vegetation ecological indices analysis model was implemented to derive a number of ecological indices of vegetation, which includes index of dominance, species richness index, index of diversity, evenness index, and important value index (IVI). The results show that GIS is very effective in managing geographic database of vegetation and improving the data quality. Using GIS, the analyst can dynamically select sample plots and rapidly derive various ecological indices of the vegetation. The results of vegetation ecology analysis are useful for understanding the environmental changes and the processes of ecological succession within nature reserves.

This study applied the geostatistically ordinary Kriging algorithm to predict the spatially rainfall distribution of Chiayi forest area. It aims to know the rainfal1 amount that is annually kept by the forest area for purposes of forestland classification. Quantity of annually gross holdback (AGH) was defined by sub­tracting the discharge quantity from the rainfall. The data used in this study were supplied by the Water Resources Agency, Ministry of Economic Affairs; and were analyzed by the GIS software ArcGIS. Results showed that the Chaiyi forest area was wide located on several watersheds, such as Tseng-Wen Chi, Tseng­Wen Dam, Lao-Nung Chi, Ching-Chui Chi, Chi-San Chi, Cheng-You-Lan Chi, Pa-Chang Chi, and Chi-Shui Chi. In 1999, the mean annually rainfall of these watershed are more than 160 mm and the gross ratio of rainfall keeping of each working circle of Chaiyi forest area, i.e., Alishan, Yushan, Dapu, and Yuchin, is 38.61%, 21.25%, 31.70%, and 65.81% respectively. The AGH by the forest area and watershed is varying in a waveform corresponding with the serial month changing. Most of the positively AGH is happened from May to August, after that AGH is smoothly come down to negative until next May.

This study took the Liukuei area of Kaohsiung County as the study area. Data on historic land uses, as well as the environmental and socioeconomic conditions of the study area were collected to develop a GIS database, and logit models were used to examine the possible factors contributing to the observed forestland changes. Comparisons of the land use conditions of Liukuei area in 1977 and 1994 showed that the major change during the period was the recovery of forestlands from farmlands. However, some of the forestlands were also converted into farmlands during the same time. Results of the logit models showed that forestlands located at lower elevation, without steep slope, and in the proximity of roads were more likely to be converted into farmlands. The probabilities were further increased at villages having more agri­cultural families. On the other hand, farmlands located at higher elevation, with steeper slope, and at villages having fewer farmers were more likely to be abandoned or reforested.

A drought database has been established based on accumulated data of the Kao-Ping Meteorological Station and drought information published by the agricultural administrative agencies. A time-series analysis model, ARIMA, was used in rainfall time series analysis of the data. Two commonly seen crops of the area, sweet potatoes and sugar cane were selected for study of ref1ection spectrum variation due to the prevailing weather condition, more specifically by draught. A preliminary result showed that the drought indexes calculated from NOAA satellite images had responded to the water deficiency started 3 months ago. The drought index and seasonal variation of precipitation are highly correlated. When the drought started, NOAA crops image sensitively ref1ect different degree of drought danger. A controlled experimentation had also proved that characteristics of spectrum crop leaves ref1ected had significant differences between different moisture conditions. Drought cycles by years and by months derived from past meteoro­logical data in this study may be a valuable reference to the future drought monitoring system.

Forest fire can cause serious damage to the ecological environment and human lives and property. Once the fire does not be put off in time, the trees are destroyed and the land loses its cover, this affects soil & water conservation. In order to obtain the information about forest fire more efficiently and correctly, this research is based on NOAA-AVHRR to derivative the NDVI (Normalized Difference Vegetation Index) and from the variation of NDVI reveals the impact of forest fire on vegetation. Auxiliary the digital number on thermal-IR band, the high probability location of forest fire can be delimited. The result can help the manager in obtaining forest fire information and thus hopefully forest fire can be put off as soon as possible.