基于人工智能和卫星遥感的人造灌木林碳汇估算及监测研究

Research on Estimation and Monitoring of Carbon Sinks in Artificial Shrublands Based on Artificial Intelligence and Satellite Remote Sensing

摘要: 在碳中和背景下，传统灌木林碳汇监测方法因效率低、精度不足，难以满足大范围动态评估需求。本研究提出一种基于随机森林算法与多源遥感数据融合的灌木碳汇估算模型，突破灌木碳汇量化技术瓶颈。通过整合Sentinel-1/2、激光雷达等多源数据，结合光谱纹理时序特征优选及动态参数校正算法，构建高精度灌木识别系统，分类精度超过90%。结果表明，该方法显著提升灌木碳汇识别精度，有效降低传统线性模型的监测偏差，支持分钟级大范围动态解析，并验证卫星遥感监测在DEA模型下的完全经济有效性。研究成果为灌木林碳汇精准核算及规模化项目管理提供高效、低成本技术方案，助力“双碳”目标实现。

Abstract: Against the background of carbon neutrality, traditional monitoring methods for shrubland carbon sinks are difficult to meet the needs of large-scale dynamic assessment due to low efficiency and insufficient accuracy. This study proposes a shrub carbon sink estimation model based on the fusion of random forest algorithm and multi-source remote sensing data, breaking through the technical bottleneck of shrub carbon sink quantification. By integrating multi-source data such as Sentinel-1/2 and LiDAR, combined with optimal selection of spectral texture time-series features and dynamic parameter correction algorithm, a high-precision shrub identification system was constructed with a classification accuracy of over 90%. The results show that this method significantly improves the accuracy of shrub carbon sink identification, effectively reduces the monitoring deviation of traditional linear models, supports minute-level large-scale dynamic analysis, and verifies the complete economic efficiency of satellite remote sensing monitoring under the DEA model. The research results provide an efficient and low-cost technical scheme for the accurate accounting of shrubland carbon sinks and large-scale project management, contributing to the achievement of the “dual carbon” goals.