Abstract: Based on remote sensing technology,GIS software,and SPSS data analysis method,the thesis selects the factors of forest health assessment and summarizes their variation rules in remote sensing spectral features and vegetation index,through correlation analysis,principal component analysis and the use of Landsat 5 TM data.The result shows that remote sensing spectral features and vegetation index are of high relevance to forest health,which can reflect forest health condition to a certain degree,especially remote sensing factors TM2,TM3,TM4,TM7 and vegetation index DVI,ARVI,PVI.Then,by making use of those variation rules,this paper analyzes the health conditions of Guangzhou’s forests via supervised classification,and according to the actual situation of the region,raises solutions to the restoration of Guangzhou’s forest health.

Key words: remote sensing, geographic information system, forest health, principal component analysis, recovery measures of forest health