A global 500 m dataset of tree crown morphology to advance vegetation modeling and management

Tree crown architecture varies widely across the globe and strongly influences ecosystem functioning, biodiversity, the carbon cycle, and land–atmosphere interactions. However, its global variability and spatial patterns remain poorly quantified due to sparse field and LiDAR measurements, limiting large-scale ecological and modeling applications. Here, we present a globally consistent 500 m tree crown structure dataset, including crown depth (CD), crown width (CW), and aspect ratio (AR, defined as CD/CW) across eight plant functional types (PFTs). The dataset was generated by linking tree-level crown observations, including 218,942 CW records and 6,109 CD measurements, with tree height and bioclimatic predictors. PFT-specific Random Forest models were used to characterize crown structure–height relationships and then applied globally to produce wall-to-wall maps. These models outperform the best-fitting allometric equations, increasing R 2 from 0.26 to 0.84 and reducing RMSE from 2.43 m to 1.14 m for temperate PFTs. The resulting map reveals distinct PFT-specific patterns, with large crowns in tropical rainforests (mean AR = 1.3 ± 0.4), deep, narrow crowns in boreal PFTs (mean AR = 2.2 ± 0.6), and wide, shallow crowns in temperate regions (mean AR = 0.95 ± 0.4). AR showed weak dependence on tree height, suggesting that it captures ecologically meaningful variation in crown form beyond overall tree size. We also provide uncertainty estimates for CD and CW to support cautious interpretation and downstream use. This dataset provides a novel gridded reference for understanding plant physiological processes along environmental gradients, improving biomass estimation, and supporting more realistic representation of crown geometry in Earth system models.