The inversion of forest aboveground biomass

Aboveground biomass (AGB) is a vital indicator for studying the carbon sink in forest ecosystems. Semi-arid forests harbor substantial carbon storage but received little attention as high spatial-temporal heterogeneity. This study assessed the performance of different data sources (annual monthly time-series radar: Sentinel-1 (S1), annual monthly time-series optical: Sentinel-2 (S2), and single-temporal airborne LiDAR) and seven prediction approaches to map AGB in the semi-arid forests at the border between Gansu and Qinghai provinces in China. Five experiments were conducted using different data configurations from SAR backscatter, multispectral reflectance, LiDAR point cloud, and their derivatives (polarimetric combination indices, texture information, vegetation indices, biophysical features, and tree height- and canopy-related indices). Results showed that compared to S1 (R2:0.24 - 0.45 and RMSE:47.36 - 56.51 Mg/ha), S2 acquired better prediction (R2:0.62 - 0.75 and RMSE:30.08 - 38.83 Mg/ha), although their integration further improved the results (R2:0.65 - 0.78 and RMSE:28.68 – 35.92 Mg/ha). The addition of single-temporal LiDAR highlighted the structural importance in semi-arid forests. The best mapping accuracy was achieved by XGBoost combining metrics from S2 and S1 time series and LiDAR-based canopy height information (R2:0.87, RMSE:21.63 Mg/ha, and RMSEr:14.45%). Images obtained during the dry season were effective for AGB prediction. Sequential forward selection was used to determine the most contributing predictors. Our methodology advocates for an economical, extensive, and precise AGB retrieval tailored for semi-arid forests.

地上生物量（AGB）是研究森林生态系统碳汇的关键指标。半干旱森林拥有大量的碳储存能力，但由于其空间和时间上的高度异质性，因而备受忽视。本研究评估了不同数据源（年度月度时间序列雷达：Sentinel-1（S1），年度月度时间序列光学：Sentinel-2（S2），以及单时相航空激光雷达）和七种预测方法在绘制中国甘肃省与青海省交界处半干旱森林AGB分布图时的性能。通过SAR后向散射、多光谱反射率、激光雷达点云及其衍生指标（极化组合指数、纹理信息、植被指数、生物物理特征以及树木高度和冠层相关指数）的不同数据配置，进行了五次实验。结果显示，与S1（R²：0.24 - 0.45，RMSE：47.36 - 56.51 Mg/ha）相比，S2获取了更佳的预测结果（R²：0.62 - 0.75，RMSE：30.08 - 38.83 Mg/ha），尽管两者的整合进一步提升了结果（R²：0.65 - 0.78，RMSE：28.68 – 35.92 Mg/ha）。单时相LiDAR的加入凸显了在半干旱森林中结构的重要性。通过XGBoost算法结合S2和S1时间序列数据以及基于LiDAR的冠层高度信息，实现了最佳的映射精度（R²：0.87，RMSE：21.63 Mg/ha，RMSEr：14.45%）。干旱季节获取的图像对于AGB预测有效。序列前向选择被用于确定最具贡献的预测因子。我们的方法推崇一种经济、广泛且精确的AGB检索技术，该技术专为半干旱森林量身定制。