Mapping Large-Scale Bamboo Forest Based on Phenology and Morphology Features

Bamboo forest is a unique landscape that mainly composed of herbal plants. It has stronger capability to increase terrestrial carbon sinks than woody forests in the same environment, thus playing special role absorbing atmospheric CO2. Accurate and timely bamboo maps are necessary better understand quantify their contribution hydrological cycles. Previous studies have reported phenology pattern forests, i.e., on- off-year cycle, can be detected with time-series high spatial resolution remote sensing (RS) images. Nevertheless, this information not yet been applied large-scale mapping. In study, we innovatively incorporate newly designed features reflecting aforementioned cycles into typical end-to-end classification workflow, including two describing growing efficiency during green-up season difference between annual peak greenness. Additionally, horizonal morphology one tree height feature were employed, simultaneously. An experiment southeast China was carried out test method’s performance, which seven categories focused. A total 987 field samples used for training validation (70% 30%, respectively). The results show combining based on spectral bands vegetation indices patterns differentiate from other categories. Based these features, exhibit reasonable distribution satisfactory overall accuracy (0.89). area proportion 82 counties agrees statistics China’s Third National Land Survey, produced images commercial satellites human interpretation (correlation coefficient = 0.69, root mean squared error 5.1%). This study demonstrates new scheme incorporating helps map accurately while reducing sample size requirement.