Abstract:

East China lies in the subtropical monsoon climatic zone and is dominated by subtropical evergreen broad-leaved forests, a unique vegetation type on earth mainly distributed in East Asia with the largest and most concentrated distribution in China. It is important to be able to monitor and estimate forest biomass and production, regional carbon storage, and global climate change impacts of these important vegetation types. In this paper, we used coarse resolution remote sensing data to identify the vegetation types in East China and develop a map of the spatial distribution of vegetation types in this region. Nineteen maximum normalized difference vegetation index (NDVI) composite images (Acquisition time span 7 months from February through August), which were derived from 10-days of National Oceanographic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR) channel 1 and channel 2 observations, and an unsupervised classification method, the ISODATA algorithm, was employed to identify the vegetation types. The image was processed using principal component analysis (PCA) to reduce the dimensions of the dataset resulting in a total of 28 spectral clusters of land cover of which 2 clusters were urban/bare soil and water. The 26 remaining spectral clusters were merged into 6 vegetation types: evergreen broad-leaved forest, coniferous forest, bamboo forest, shrub-grass, aquatic vegetation and agricultural vegetation using the Chinese vegetation taxonomy system. The spatial distribution and areal extent calculated for the coniferous forests, shrub-grass, evergreen broad-leaved forests and agricultural vegetation compared well with the Vegetation Atlas of China at a 1∶1 000 000 scale. The spatial accuracy for coniferous forests, shrub-grass, evergreen broad-leaved forests and agricultural vegetation was 79.2%, 91.3%, 68.2% and 95.9%, respectively, and the area accuracy was 92.1%, 95.9%, 63.8% and 90.5%, respectively. The spatial and area accuracy of the bamboo forest was 28.7% and 96.5%, the spatial accuracy of aquatic vegetation was 69.6%, but there is great error in its area accuracy because image acquisition did not cover the full year. Our research demonstrated the feasibility of using NOAA-AVHRR to identify the different vegetation types in the subtropical evergreen broad-leaved forest zone in East China. The spatial location of the 6 identified vegetation types coincided with the actual geographical distribution of the actual vegetation types in East China.

Key words: Vegetation classification, Remote sensing, NDVI, Principal component analysis, East China