1985-2015年5KM全球森林生物量5年产品

森林生物量是全球气候观测系统认定的基本气候变量之一，在全球碳循环和气候变化相关领域起着重要作用。本项目旨在利用多源遥感数据实现全球森林生物量制图。首先，探讨了高级遥感产品特别是GLASS 数据产品在森林生物量制图中的性能；然后利用植被覆盖度和反照率等GLASS数据产品和集成学习算法生成1985-2015年全球森林生物量产品（GLASS森林生物量）。
基于地面实测点、激光雷达数据和部分高分辨率生物量数据编译生成的森林生物量基准数据，采取交叉验证的方式对GLASS森林生物量数据进行精度评价，结果显示R2为0.83，均方根误差为34.45 Mg/ha，生物量估算精度在低值区和高值区仍需要进一步提升。与其他全球覆盖森林生物量数据集进行对比，发现GLASS森林生物量在空间分布上较为合理。

Forest biomass is one of the fundamental climate variables recognized by the Global Climate Observing System (GCOS), and plays a critical role in global carbon cycle and climate change-related research. This study aims to produce global forest biomass products using multi-source remote sensing data. First, the performance of advanced remote sensing products, particularly GLASS data products, for forest biomass mapping was evaluated. Subsequently, a global forest biomass product (GLASS Forest Biomass) spanning the period 1985–2015 was generated using GLASS data products including vegetation coverage and albedo, combined with ensemble learning algorithms.
A forest biomass reference dataset was compiled from ground truth plots, LiDAR data, and selected high-resolution biomass data. Accuracy assessment of the GLASS Forest Biomass product was conducted via cross-validation, yielding a coefficient of determination (R²) of 0.83 and a root mean square error (RMSE) of 34.45 Mg/ha. The estimation accuracy of forest biomass still requires further improvement in both low and high biomass value regions. Compared with other global-scale forest biomass datasets, the spatial distribution of GLASS Forest Biomass is relatively reasonable.

搜集汇总

背景与挑战

背景概述

该数据集是1985年至2015年间的全球森林地上生物量5年产品，空间分辨率为5公里，采用GLASS遥感数据产品和集成学习算法生成，主要用于支持全球碳循环和气候变化研究。数据集通过交叉验证评估，R²达到0.83，均方根误差为34.45 Mg/ha，显示较高的准确性，但在生物量极值区域仍需进一步优化。