A new strategy for improving the accuracy of forest aboveground biomass estimates in an alpine region based on multi-source remote sensing
A new strategy for improving the accuracy of forest aboveground biomass estimates in an alpine region based on multi-source remote sensing
复制标题
DOI：
10.1080/15481603.2022.2163574
复制
发表时间：
2023-12-31
影响因子：
6.7
通讯作者：
Li，Mingshi
中科院分区：
文献类型：
--
作者：
Zhang，Yali;Wang，Ni;Li，Mingshi
关于森林地区优势树种群和地上生物量（AGB）分布的明确空间信息对于制定有针对性的森林管理和生物多样性保护措施以及评估森林碳汇能力至关重要。缺乏持续更新的30米空间分辨率产品来绘制优势树种群。绝大多数基于遥感的 AGB 估算方法对于优势树种群或森林类型的准确度相对较低，不适合 AGB 建模。因此，本研究旨在开发一个考虑不同树种物候特征的综合框架，以提高森林优势树群的制图精度和相应的AGB估计。使用机器学习算法和物候参数创建了优势树种群的三十米分辨率地图。利用从光学和雷达图像中提取的特征以及物候特征以时间一致的方式构建AGB估计模型，以提高AGB估计精度并进行动态AGB监测。该方法准确表征了研究区优势树种群的动态分布。不考虑不同森林类型或物种的传统AGB模型的R2值为0.52，而考虑物候和森林类型的拟议模型的R2值为0.67。这一结果表明，结合物候和优势种的信息可以提高 AGB 估计的准确性。大部分地区AGB为30~55 t/ha，说明森林以幼林或中龄林为主，且研究期间AGB大于30 t/ha的面积比例有所增加，表明森林质量有所改善。此外，橡树AGB最高，表明应鼓励橡树造林，以增强未来森林生态系统的固碳能力。研究结果为研究人员和管理者了解森林发展和森林健康的趋势提供了新的见解，也为制定更合理的森林经营策略提供了技术信息和数据库。
Spatially explicit information on the distribution of dominant tree species groups and aboveground biomass (AGB) in forested areas is essential for developing targeted forest management and biodiversity conservation measures, as well as assessing forest carbon sequestration capacity. There is a shortage of continuously updated 30-m spatial resolution products for mapping dominant tree species groups. The vast majority of remote sensing-based AGB estimation approaches have relatively low accuracy for dominant tree species groups or forest types and are unsuitable for AGB modeling. Therefore, this study aims to develop an integrated framework that considers the phenological characteristics of different tree species to improve the mapping accuracies of forest dominant tree groups and corresponding AGB estimates. Thirty-meter resolution maps of dominant tree species groups were created using machine learning algorithms and phenological parameters. Features extracted from optical and radar images and phenological characteristics were used to construct AGB estimation models in a temporally consistent manner to improve the AGB estimation accuracy and perform dynamic AGB monitoring. The proposed method accurately characterized the dynamic distribution of the dominant tree species groups in the study area. The traditional AGB model that does not consider different forest types or species had an R2value of 0.52, whereas the proposed model that considers phenology and forest types had an R2value of 0.67. This result indicates that incorporating information on phenology and dominant species improves the accuracy of AGB estimations. The AGB in most regions was 30–55 t/ha, showing that the majority of the forests were young or middle-aged stands, and the areal percentage of AGB greater than 30 t/ha increased during the study period, suggesting an improvement in forest quality. Furthermore, the oak AGB was the highest, indicating that oak afforestation should be encouraged to enhance the carbon sequestration capacity of future forest ecosystems. The results provide new insights for researchers and managers to understand the trends of forest development and forest health, as well as technical information and a database for formulating more rational forest management strategies.