| The processing of spatial pattern of traditional land use change model is always constrained in taking the canton as a basic unit and is difficult to reflect the dynamic change principle of land use change in detail. In the process of studding land use change of Qianjiang city in Hubei Province, a landscape pattern analysis is introduced into this study for exploring of the land use change in Qianjiang city. Additionally, the result of traditional land use change study is often stored into an independent and isolated database system so as to hardly be reused and produce a little benefit as well as share the information in public. This study tries to change this embarrass situation and firstly use the fashionable RDBMS to store the land use information that contains large number of spatial and nonspatial information. Meanwhile, synthetically using the automatically classification by computer and interpreting by expert experience and integrating the remote sensing and GIS technologies, the accuracy and efficiency of land use classification is improved greatly. The main research results are shown as followings:1. For this study, the TM images of the year 1986 and 2000 are selected as the data source of the extracting the landscape patches. Without the high resolution images of SPOT, this study tries to use a resolution merge with the pan band (the eighth band of TM 2000) instead of the pan band of SPOT imagines that are commonly used in other studies. Based on this, the image resolution is conducted by means of principal component, and implemented with a supervised classification to acquire the land use map of the year 2000 later. The research results indicate that the accuracy of land use classification reaches 90.64%, which is met with the requirement of land use classification2. In the process of studying the whole land use and landscape of the Qianjiang City, a threshold method is adopted with a limit value of 45 for image translation of the water area, while for image translation of the residential area a model based on the spectral knowledge is selected. All the results will update the classification by overlapping and intersecting based on the spatial analysis of topology. Especially, the remote sensing imagery processing software ERDAS 8.5 is used with its spatial analysis modular. Through Criteria Function of ERDAS Imagine, this research teststhe threshold value repeatedly to gain a fine result with an accuracy coming to 91.5%, and it is imported to the ARC/INFO database to modify the classification by the command of UPDATE. Thus, it is proved that a method taking advantage of ERDAS and ARC/INFO to integrate the classification automatically by computer and other good ways as well as interpreting by expert experience can overcome the low accuracy of classification by computer.3. Using above-gained results of image translation, the INFO tables in ArcGIS are exported to database software and an area statistics is conducted by grouping land use code and other criteria. After that, the land use change matrix and relevant landscape patter index are obtained to analyze the factor that result in the land use change of specific types. In this way an effective approach for the real-time and dynamic monitoring of the land use change and controlling the land use change is formed.4. With the land use transformation matrix and some landscape pattern index selected in this research, some significant results are shown as followers: (1)This study sample plot is The Qianjiang City with large number of farms, and the patches is regular other than fragmented in the county. Comparing with the water and forestland, the water land is changing with trivial stews and irregular aquatic water land, and the water area is increased largely with a value of 12.08% and the water area is fragmented. (2)The forestland mainly is distributed at the northern and ambient river branches. With the faster and faster urbanization in past fourteen years some the construction land is increased 33% in the area. (3)With the two period ima...
|
PDF Full Text Request