Keywords: 
• forest ecology /
• moso bamboo forest /
• remote sensing /
• dynamic monitoring /
• carbon stock

Abstract: Maximum likelihood classification method was used to extract moso bamboo（Phyllostachys pubescens） forest from multitemporal Landsat-5 Thematic Mapper （TM） images. The dynamic change of bamboo forest areas in Anji County，Zhejiang Province in the past 30 years was conducted. Meanwhile，total aboveground carbon stock of bamboo forest was estimated. The results showed that （1） overall classification accuracy for each TM image was over 85%，and Kappa coefficient for moso bamboo forest ranged from 0.80 to 0.95. The relationship between moso bamboo forest area estimated from remote sensing and the forest inventory data was significant with a R2 value of 0.981；（2） during 1986 - 2008，bamboo forest areas were increased with highest rate of 86% in Xiaofeng and lowest rate of 14% in Tianhuangping，except Kuntong with decreased rate of 8.89%；（3） the increase of moso bamboo forest area in Anji County during the past 30 years was at the expense of conifer forest，broadleaf forest and farmland；（4） according to classification results and aboveground carbon density of moso bamboo （i.e.，20.297 Mghm－2），the aboveground carbon stock was 1.106 Tg in 1986，1.213 Tg in 1991，1.327 Tg in 1998，1.413 Tg. in 2004 and 1.466 Tg in 2008，respectively.［Ch，2 fig. 4 tab. 31 ref.］