Gridded estimates of aboveground biomass by plant functional type within Alaska and Yukon, 1985-2020

This dataset includes 30 meter (m) resolution gridded estimates of live aboveground biomass (AGB) for five common plant functional types (PFTs; deciduous shrubs, evergreen shrubs, forbs, graminoids and lichens) within Alaska and northwest Canada. Estimates were produced for single years, every five years from 1985 to 2020. To model AGB we used a multi-scale approach, scaling from field harvest data and unmanned aerial vehicle (UAV)-based biomass predictions. Estimates were produced using linear mixed effects models with LASSO (Least Absolute Shrinkage and Selection Operator) regularization and were based on gridded climatological, topographic, phenological, PFT cover, and Landsat spectral predictors. A Monte Carlo approach with 100 iterations was used to propagate uncertainty. For each PFT and each year, this dataset provides the 2.5th percentile estimates (lower bound), 50th percentile estimates (best estimate) and 97.5th percentile estimates (upper bound) from the Monte Carlo analysis. These maps capture vegetation changes occurring within the Arctic/boreal region, including increasing shrub biomass and decreasing lichen and graminoid biomass. They also demonstrate the role of disturbances such as wildfire in shaping vegetation change trajectories.

本数据集包含阿拉斯加与加拿大西北部区域内，5种常见植物功能型（plant functional types, PFTs；即落叶灌木、常绿灌木、杂类草、禾本科草本植物与地衣）的30米分辨率地上活体生物量（aboveground biomass, AGB）栅格估算结果。该估算结果针对1985年至2020年间的单一年份及每5年间隔生成。为构建地上生物量估算模型，我们采用多尺度建模方法，以野外收获采样数据与无人机（unmanned aerial vehicle, UAV）获取的生物量预测结果为基础开展尺度上推。本次估算通过引入LASSO（最小绝对收缩与选择算子，Least Absolute Shrinkage and Selection Operator）正则化的线性混合效应模型完成，所用预测变量涵盖栅格化的气候、地形、物候、植物功能型覆盖度以及Landsat光谱数据。我们采用包含100次迭代的蒙特卡洛方法实现不确定性传递。针对每一种植物功能型与每一个年份，本数据集提供了蒙特卡洛分析得到的2.5百分位数估算值（下限）、50百分位数估算值（最优估算值）以及97.5百分位数估算值（上限）。该系列栅格图捕捉了北极/寒温带区域内的植被动态变化，包括灌木生物量的持续增长以及地衣与禾本科草本植物生物量的衰减，同时也揭示了野火等干扰事件在塑造植被变化轨迹中所发挥的关键作用。