Multi-source Information Integration Based Remote Sensing Information Retrieval in North Hebei Province

North Hebei province lies in the ecotone from agriculture to animal husbandry and forest and is eco-fragile transitional region. Serous land degradation, especially grassland degradation occurs in this area, because of overgrazing and over-reclamation. There are also many disasters, such as drought, dust storm, forest fire, and so on. This area has characteristics of ecological degradation and economic poverty. Soil erosion and dust storm has important impacts on eco-safety of Beijing and Tianjin city. This paper used multi-source information integration method to retrieve land use/cover information using Landsat TM data as information source. Multi-source information integration method used in this paper included multi-source remote sensing data fusion method, visual interpretation and field survey and validation method. Multi-source remote sensing data fusion methods included panchromatic and multispectral data fusion, normalized difference vegetation index, tasseled cap transformation, remote sensing image and DEM information fusion, remote sensing information and GIS database fusion and supervised classification. The result shows that total classification accuracy in 2000 was 84.62% and total Kappa was 0.8026. Arable land increased from 34.3 percent in 1987 to 35.3 percent in 2000, forest land decreased from 38.0 percent in 1987 to 37.4 percent in 2000, and grassland decreased from 24.3 percent in 1987 to 23.9 percent in 2000. Arable land, forest land and grassland maintained 96.6 percent of total land use/cover types both in these two years. * Corresponding author 1.INTRODUCTION Land is an important component of global ecosystem; land use/cover change (LUCC) is one of the important research fields in global change. Research on land use/cover change can promote research process of global environment change and sustainable development. Remote sensing technology has characteristics of dynamic, macroscopical, real-time and multitemporal, traditional land use/cover survey technology is incomparable to remote sensing. There are many methods to retrieve land use/cover information, Liu Yalan used segmentation-based classification approaches for different spatial images (Liu et.al., 2002), Others used knowledge-based expert classification (Gao et. al., 2000a; Ton J, et al.,1991a; Wharton S W.A., 1987a ), improved artificial neural network classification (Jia, 2000a; Xiong, et. al., 2000a) fuzzy logic classification, spatial data mining (Di et. al., 2000a), and so on. This paper used multi-source information integration method to retrieve land use/cover information using Landsat TM data as information source. Multi-source information integration method used in this paper included multi-source remote sensing data fusion methods, visual interpretation and field check method. Multi-source remote sensing data fusion methods included panchromatic and multispectral data fusion, normalized difference vegetation index, tasseled cap transformation, remote sensing image and DEM information fusion, remote sensing information and GIS database fusion and supervised classification.