Study On The Oasis Distribution Extraction Based On TM Images

Oasis is the unique Landscape of arid, it is also the main space of human being production and living. In recent years, with the development of social and economic, ecological environment of oasis became degeneration, desertification became severe, the area of oasis declined, the spatial distribution of oases is changing all the time. Therefore, this study put forward a study of oasis spatial information extraction based on TM images.At first, this study expounds the concept of "oasis" which is not uniform yet, propose narrow concept of "oasis" and classification. Then vegetation and water body information are extracted by NDVI threshold and MIR/G index, space distribution of oasis is educed by porphyroclasts treatment. At last, this study chooses some sample checkpoints, verify the accuracy of method by high-resolution image and field survey. The main conclusions are as followed:1. On the basis of common understanding existing concept, this paper points out that oasis is made up of vegetation, water and residential area, but not only the simple composition. The suitable vegetation is what have high coverage, the suitable water and residential area are specific region that lying in the high-vegetation-coverage area. Specially, the predatory mining city, desert vegetation and semi-desert vegetation are not considered to be oasis, it is because that predatory mining city is difficult to embody the important characteristic of "its vegetation based and it has first productivity which significantly higher than the environment".2. At the aspect of determination of image's phase, this paper puts SPOT-NDVI data as reference, and analyzes vegetation phenological characteristic in agrarian area which at middle reach area of Heihe River. The results show that it is propitious to reflect the integrated spatial distribution of oasis. It's important to collect information of phenology in agrarian area at the time of field study and information finding. 3. The extraction process of information is divided into three main components, vegetation, water and residential area information extraction. Firstly, the information of vegetation is extracted on the basis of NDVI threshold segmentation; secondly, residential information is extracted and deleted based on the information of vegetation; thirdly, this study merges three types information maps together, generate the spatial distribution of oasis, the results shows that the accuracy of information extraction is 92.77%.