Research On Land Use And Land Cover Change Detection Method In The Modern Yellow River Delta Based On Remote Sensing

Land use and land cover change (LUCC) becomes the main reason which causes the environment change. The Modern Yellow River Delta is an important base for petroleum and commissariat in our country. The conflict between economy development and ecosystem protection is prominent. Therefore, the research on LUCC in this area is significant for guiding its sustainable development. At the same time, the development of remote sensing makes it possible to detect changes fleetly and automatically. Now change detection methods such as image arithmetic differencing, image ratio differencing, vegetation index differencing, PCA analysis, change vector analysis, post-classification comparison method, and so on, are widely used. Their ability and accuracy are limited for change-detection.This thesis mainly researches on change detection methods with Landsat TM images in the Modern Yellow River Delta. Firstly, it establishes the LUCC classification system. Then it tries the methods mentioned above to detect changes of study area and analyses their advantages and disadvantages respectively. After that, the paper proposes an improved spectral angle mapping (SAM) classification method for quantificational information. It is to add the front 3 bands of K-L conversion, 3 bands of K-T conversion (brightness, greenness and wetness) and its normalized difference vegetation index information to the original image of 6 bands. The total of bands is up to 13. Then interpreter symbols of RS image are established, which are the basis of choosing referenced samples to create referenced spectral profiles. After obtaining a series of spectral angle maps by SAM classification with ERDAS, the thesis proposes a method without threshold to determine the class of objects. Moreover it puts forward subarea classification method to reduce the confusion between downtown resident district and salina or sand with high salt. And then accuracy assessment is done to evaluate the classification. Finally, LUCC image of study area is created by algebraic operation on the former and the latter classification images. And the transfer matrix of LUCC is calculated to analyze the changes shortly.The experiment confirms that the method enhances spectral difference of objects and dispenses with choosing threshold by manpower. Therefore, classification accuracy is advanced and process efficiency is improved. The accuracy assessment indicates it has good effect.