An upscaling framework for estimating regional-scale fractional cover of Chinese fir (FCCF) by integrating UAV and satellite images

作者：Houxi Zhang, Jianwei Cai, Minghui Cai, Xunlong Chen, Yiming Sun, Kun Chen, Shijiang Cao, J. Ren · 发表于：International Journal of Digital Earth · 年份：2025 · DOI：10.1080/17538947.2025.2510570 · 被引用次数：4 · 研究领域：Remote Sensing in Agriculture、Remote Sensing and LiDAR Applications、Forest ecology and management

Accurate estimation of fractional cover of Chinese fir (Cunninghamialanceolata (Lamb.) Hook.) (FCCF) is crucial for forest management and carbon sequestration assessment. Traditional field inventory is time-consuming, labor-intensive, prone to sampling bias, and often mismatched with satellite image resolutions, hindering accurate regional FCCF mapping. In this study, we propose an upscaling framework for regional estimation of FCCF integrating unmanned aerial vehicle (UAV) and satellite imagery. High-precision classification labels were generated from UAV visible imagery using three methods: object-based image analysis with random forest (OBIA-RF), maximum likelihood, and minimum distance classifiers. The resulting binary classification maps were then aggregated to different spatial resolutions and linked with Sentinel-2A (10 m and 20 m) and Landsat 8 OLI (30 m) data to develop FCCF estimation models. We evaluated the performance of four models, including linear regression (LR), RF, extreme gradient boosting (XGBoost), and light gradient boosting machine (LightGBM). SHapley Additive exPlanations (SHAP) values were employed to quantify the influence of vegetation indices (VIs) derived from satellite images on model accuracy. Results showed that OBIA-RF achieved the highest accuracy for Chinese fir classification based on UAV images, with an average overall accuracy and recall of 0.919 and 0.909, respectively. At the satellite scale, the LightGBM, RF, and XGBoost models had th...