| As widely grown of rubber plantation in Xishuangbanna in recent decades, it brings a lot of problems while it promotes economic development, such as, there is a significant reduction of the tropical monsoon forest, the arid, soil erosion, and soil degradation in the rubber trees planted regions.In this paper, the study area is Xishuangbanna, based on the object-oriented method, four periods of remote sensing image were used as source data, the land cover information of multiple times’ of the study area were got from the classification of land cover and change detection, then got the temporal and spatial distribution of land cover in the study area. On analysis of the changed area and the data of geographic background of rubber, the main factors were find out which affected the distribution of rubber trees. The results of this research as the following aspects:(1) In this article, the method of Object-oriented classification were introduced, the concept of multiscale segmentation techniques, principles, the weight of remote sensing image band and the selection of parameters for multi-scale segmentation were Elaborated. It was also introduced the method of classification in this article, fuzzy expert classification and nearest neighbor classification (class supervised classification).(2) The test used the TM remote sensing images as source data in, first, divided the images in multiscale segmentation techniques, got the appropriate segmentation scale for study area, then, obtained land cover classification of study area; Third, the test extracted the distribution of rubber plantation, according to the growing season and non-growing season images NDVI, EVI difference and texture characteristics of rubber plantation. The end evaluate the accuracy of the classification results, the classification accuracy is90.2%.(3) The test obtained land cover information based on the object-oriented ideological and got multi-image information changes with the use of change detection change vector analysis. The overall accuracy of the final change detection is relatively high, and achieved satisfactory results for change detection, the object-oriented change detection methods turns to be effective.(4) The characteristic of study areas’ land cover spatial and temporal distribution was got, which from the overlay analysis of four land cover remote sensing data and the DEM, SLOPE data in study area. The main types of land cover are broad-leaved forests and rubber plantations in study area. On the whole, Distribution of rubber plantation is at an altitude of1300m or less, slope between6-25°where human activities are more frequent gentle slope, slope area, and in these area of rubber plantation accounted for more than78.8%of the total area, and showed a rising trend.(5) In the end, the test analysis the variation and trend of rubber plantation in nearly20years of the study area, on the basis of overlay analysis by three rubber plantation change data (1990-2000,2000-2005,2005-2010) and DEM, SLOPE, temperature, precipitation, topography, soil data in the study area, meanwhile, analyzes the key geographic factors affecting the expansion of the rubber plantations in study area. It turns out that the major expansion of rubber plantation area are small fluctuations in the lower elevations of the mountains and the undulating in middle elevation of the mountains, in these region of which soil types are mainly Acrisols and Ferralisols, altitude are below1150m, slope are between6-25°, temperature are between22-23℃, precipitation are between1310-1470mm. Finally it get the weight of impact factors for rubber plantation change, DEM influence weight is0.22, SLOPE is0.10, precipitation is0.17, temperature is0.17, geomorphological is0.14, soil is0.18. The most important impact factor for the change of rubber plantation is elevation. |
PDF Full Text Request