Landuse/cover Change Monitor And Landscape Pattern Analysis Based On Remote Sensing Data And Landscape Index

With the rapid increasing of world poulation and economic development in the last century, urbanization has become faster than ever before. Urban expansion not only influences local environment by changing its land cover, but also induces some social problems.Recently, urban land use and cover change (LUCC) monitor and urban landscape pattern analysis becomes a hotspot in study of urban geography and environment. Study of urban LUCC can help urban managers to make reasonable decision that protect the local environment and support the sustainable development.Changsha city was chosen as a study area. A hybrid classification method was explored to classify imagery and produced four time series landuse maps respectively. The structure of landuse/cover change and its impact on local environmente were analyzed using landscape metics quantitatively. The main conclusions are as follows:1. A hybrid landuse/cover classification method.In order to improve the accuracy of landuse/cover classification, this paper designed a hybrid landuse/cover classification method which involved three steps: unsupervised classification, patch based analysis, and manual interpret based on classification knowledge. The classification result showed that the average total classification accuracy of four landuse map is more than 89.6%, which is higher than ML classification approach by 12.6%.2. Landuse/cover classification based on remote sensing imagery.In order to acquire the landuse/cover information of each time in Changsha area, we employed a hybrid classification approach to classify the imagery into landuse map of five landuse/cover types(water forest, arableland, city, unused land). The average total classification accuracy and Kappa coefficient of four landuse maps is higher than 87% and 0.8, which can support the next progress of landuse/cover change detection. 3. Landsue/cover change monitor.The post-classification compared change detection method was used to extract the land use/cover change information from 1973 to 2006. The result showed that from 1973 to 1993, urban area increased from 51.6 to 124.6 km about 7.1% per year, and arableland decreased form 347 to 328 km about 10.9 % per year. From 1993 to 1998, urban area increased from 124.6 to 186.7 km about 10.9% per year, arableland decreased form 347 km to 328 km about 10.9 % per year, water and forest both decreased, and unused land increased about two times in this period. From 1998 to 2006, urban area replaced arableland land becoming the largest landscape in the study area, urban area inreased 120 km, about 14.5 % per year. Forest increased about 34.6 km and arableland decreased 133.6 km.4. Analysis on landscape structure and chenges of landscape pattern.Class-level landscape pattern index showed that the shape of urban experience a course of simpleness-complexity-simpleness. This result indicated that urban expansion inceased rapidly and forest becomes more and more fragmental.landscape-level pattern change analysis showed that from 1973 to 2006, index of SHDI, SHEI, LPI increased a little, and MPS decreased, which indicated that landscape of the study area become more and more fragmetal, urban expasion played a negative role in this progress. Therefore, it is essential to protect other land resources.