 This paper proposes a new approach to estimating carbon stock in mozo-bamboo forests. It uses a multi-scale, object-based method which differs from traditional pixel-based approaches. The results show that the multi-scale object hierarchy accurately analyzes the correlation between carbon storage and the microstructure of the bamboo forest. Furthermore, it has a high accuracy of 0.74 when used to predict carbon storage. This article was authored by Yulong LV, Ning Han, and Hua Chengdu.