A Satellite-Unmanned Aerial Vehicle (UAV) fusion framework for spatially explicit mapping of new culm yield in Moso bamboo forests

Moso bamboo ( Phyllostachys edulis ) is a critical ecological and economic resource in subtropical Asia. However, monitoring its yield—specifically the successfully recruited new culms that determine economic returns—remains a major challenge due to the decoupling of canopy signals from the concealed, belowground shoot growth. This study addresses the critical observational gap between canopy signals and subsurface recruitment by developing and validating a multi-scale, satellite-Unmanned Aerial Vehicle (UAV) fusion framework for the spatially explicit mapping of new bamboo culm yield. Initially, Sentinel-2 time series imagery was analyzed to capture biennial spectral signatures, thereby delineating on-year stands (potential recruitment zones) with high accuracy (overall accuracy >90%). Subsequently, within these zones, UAV multispectral data and a Refined Local Maximum algorithm precisely quantified individual mature bamboo density (R² = 0.81). A predictive model integrating the satellite-derived phenological mask (identifying on-year stands) with UAV-derived stand density was then developed and validated against independent field data. The model predicted the spatial distribution of new culms at a regional scale (R² = 0.72, RMSE = 1.8 culms per 100 m²). This study provides a scalable and mechanism-guided methodology for estimating bamboo recruitment yield, offering a transformative shift in management from reliance on visible cues towards predictive, quantitative planning. The framework is transferable to other clonal plant ecosystems with concealed belowground resource dynamics. • First satellite-UAV fusion framework for mapping new bamboo culm yield. • UAV-derived stand density is the dominant predictor of yield. • Sentinel-2 time series analysis enables accurate delineation of on-year and off-year stands.