Multi-source Remote Sensing Images Of Desertified Land Information Extraction Research

The land resource is the base of human survival, along with the world's population sharp growth and the change of global climate and environment, the harm of land desertification especially the land sandification is becoming more and more evident. The combination of remote sensing and GIS is a indispensable and important method to monitor the sandy land, it not only provides the timely and quantitative analysis of current sandy land situation, but also provides the dynamic monitoring and quick evaluation from time and space, establishes the scientific basis for further control and forecast of sandy land. The study of extracting sandy land information in different methods and different sensor images, we can get the comparatively satisfied method of extracting sandy land information in different sensor images.At present, CBERS-2,Landsat-5 TM and SPOT images are widely applied to the sandy land study in our country. This paper researches the image and spectral characteristic, digs the spectral knowledge of different kinds of sandy land in these images; build the sandy land extractable model. It shows that: the overall extracted presion is more than 80%, and the presion from TM image is the highest, CBERS-2 image is lowest. This conclusion indicated that there is no positive relationship between the classified results and the image space resolution.Spectral Angle Mapper is also used to extract the different kinds of sandy land information from CBERS-2,Landsat-5 TM,SPOT-5,TM and SPOT-5 fusion image and the overall presion is more than 85%. Among the results, the fusion image extractable presion is the highest, SPOT-5 image take second place, TM image is the lowest. It shows that there is a positive relationship between the classified results and the image space resolution——the higher the space resolution, the higher the extractable presion.Compare and analyse the extractable presion of different methods and different sensor images respectively, gained the conclusion: the Spectral Angle Mapper is more convenient and accurate than the spectral knowledge method. Thereinto, the extracted presion of TM and SPOT-5 fusion image in this method is the highest, SPOT-5 take second place and CBERS-2 is also higher than other combinations. Think over the actual application, CBERS-2 data has been provided freely in our country now; the price of TM data is lower than SPOT-5 data. So, the sandy land monitoring in large range such as the city or county level upwards, CBERS-2 can achieve the presion; in some key monitored space, SPOT-5 would get better effect; and the extractable presion will be higher if fuse the multi-spectral image and high-resolution image in permitted economic condition.