Abstract: Taken ten orbit MODIS images in 2006 as data source, normalized difference vegetation index NDVI chosen on the basis of a analysis of reflected spectrum characteristics on vegetation by MODIS data, and the results analyzed of the MODIS-NDVI variational curve of three different ground types, such as shelter forest, drought land and residential area in Fujian coastal regions, the littoral shelter forest information of Fujian was extracted. The training areas based on the characteristics of the MODIS-NDVI variational curve of three different ground types had been chosen, and the littoral shelter forest information of Fujian have been distilled by supervised classification. Taking the remote sensing image of CBERS-2 of Fujian in 2005 as reference, the error results of shelter forest monitored using MODIS data had been corrected. The overall classification accuracy of littoral shelter forest area in Fujian was 79.67%, and kappa index was 0.7625. The results showed that the human-computer interactive interpretation method to monitor the littoral shelter forest information of Fujian by using MODIS data was valid. The area of littoral shelter forest of Fujian in 2006 was about 336000 ha according to remote sensing monitoring.

Key words: Remote sensing, Remote sensing, Shelter forest, Monitoring, MODIS, Fujian coast