王景旭在Remote Sensing发表基于高光谱和LiDAR点云的黄河流域湿地植被LAI精细反演的研究成果
叶面积指数(LAI)是评估湿地生态系统功能的核心参数,其精准获取对湿地保护、生态监测及功能评估具有关键意义。黄河三角洲湿地作为典型的生态脆弱区与重要生态屏障,植被动态变化直接影响区域生态稳定,但单一遥感数据源反演精度不足,难以满足高精度监测需求。因此,探索多源遥感数据融合的LAI反演技术,成为湿地生态遥感领域的重要研究方向。
该研究以黄河三角洲滨海湿地为研究区,创新性融合无人机载高光谱与激光雷达(LiDAR)数据,构建“光谱+结构”双维度植被信息获取体系。研究提取38种植被指数与12个点云特征,通过皮尔逊相关分析与SHAP方法集成筛选关键特征,结合RF、XGBoost、CatBoost三种机器学习算法构建LAI反演模型。结果显示,多源特征融合模型精度显著提升,其中随机森林模型表现最优(R²=0.968,RMSE=0.125),同时揭示了湿地植被LAI的空间异质性分布规律。
该研究成果近期发表于遥感领域期刊Remote Sensing,联合培养研究生单晨强为论文第一作者,王景旭副研究员为通讯作者。由河南省科学院地理研究所与河南理工大学、中国石油大学(华东)的研究人员共同合作完成。
此项研究得到国家自然科学基金(4230144)、河南省科技研发计划联合基金(225200810057)、河南省科技项目(252102321108)、河南省科学院创新团队项目(20230107)的共同支持。
论文信息:
Shan, C., Cai, T., Wang, J., Ma, Y., Du, J., Jia, X., Yang, X., Guo, F., Li, H., & Qiu, S. (2026). Refined Leaf Area Index Retrieval in Yellow River Delta Coastal Wetlands: UAV-Borne Hyperspectral and LiDAR Data Fusion and SHAP–Correlation-Integrated Machine Learning. Remote Sensing, 18(1), 40.
论文链接:https://doi.org/10.3390/rs18010040

Figure 1. Location of the study area in the Yellow River Delta wetlands: (A,B) Study area location within China and Dongying city boundaries (green dots); (C) In situ LAI sampling points (orange-yellow dots) with UAV hyperspectral RGB composite background; (D,E) Representative on-site vegetation imagery of the study area.

Figure 2. Technical framework of this study.

Figure 3. SHAP analysis of fused features: (a1,a2) illustrate the relative importance of fused features in LAI prediction, quantified by mean absolute SHAP values. (b–m) depict the SHAP partial dependence plots of the selected fused features.

Figure 4. Accuracy analysis of LAI estimation based on fused features.

Figure 5. Spatial distribution of LAI: (A) Overall LAI spatial pattern; (B) LAI profile with an upward coastal-to-inland trend; (C,F) Suaeda salsa-dominated community (low LAI) and its RGB image; (D,G) Suaeda salsa-Tamarix chinensis mixed community (transitional zone, medium LAI) and its RGB image; (E,H) Tamarix chinensis-Phragmites australis mixed community (high LAI) and its RGB image.
撰稿:王景旭
初审:任 杰
复审:杜 军
终审:鲁 鹏