Biomazon: A Multimodal Dataset for 3D Forest Structure and Biomass Modeling in the Amazon Basin

2026-06-03

AI Summary

This study addresses the limitation of existing forest remote sensing approaches, which typically predict canopy height or aboveground biomass independently, thereby neglecting holistic modeling of vertical structure and physical consistency. We present the first 20-meter multimodal benchmark dataset tailored for the Amazon basin, integrating GEDI vertical profiles with aboveground biomass labels and diverse remote sensing data—including Sentinel-1/2, ALOS-2 PALSAR-2, Copernicus DEM, Dynamic World land use/land cover, and AlphaEarth embeddings—under a unified spatial partitioning and evaluation protocol. Our model employs a shared encoder–decoder architecture with task-specific heads to enable structurally consistent, joint three-dimensional forest prediction. Systematic ablation studies quantify the contributions of individual modalities, model scale, and fusion strategies, demonstrating that our benchmark results match or exceed those of current gridded products.

Abstract

Accurate, spatially explicit characterization of tropical forest structure is essential for carbon accounting and ecosystem monitoring, yet most ML pipelines predict canopy-top height proxies (e.g., RH95/RH98) or AGBD as separate scalar targets, rather than learning the forest vertical structure as an ordered profile. The community lacks a ML-ready multimodal benchmark for predicting the entire GEDI RH profile jointly with AGBD, or for evaluating methods that enforce physically consistent ordering across RH percentiles. We address this with Biomazon, a 20 m multimodal benchmark dataset over the Amazon Basin that pairs GEDI RH and AGBD targets with multi-sensor predictors (Sentinel-1/2, ALOS-2 PALSAR-2, Copernicus DEM, Dynamic World LULC, and AlphaEarth embeddings) under standardized spatial splits and evaluation protocols. Using a shared encoder-decoder with task-specific heads as a baseline framework, we conduct a comprehensive ablation study of (i) backbone/model scale, (ii) modality contributions, and (iii) the use of auxiliary embeddings under standalone and fusion settings, and we report both single-target and joint-target results to quantify tradeoffs under a unified training protocol. Finally, we contextualize baseline performance through regionally aligned comparisons against existing gridded products, including GEDI L4D RH10-RH98 and AGBD, at matching temporal scale. Biomazon, together with the accompanying protocols and baseline results, establishes a reference benchmark for future work on structurally consistent RH-profile prediction and structure-biomass modeling in tropical forests.

Problem

• forest vertical structure
• biomass modeling
• GEDI RH profile
• multimodal dataset
• tropical forests

Innovation

• multimodal benchmark
• vertical forest structure
• joint RH-AGBD modeling
• physically consistent ordering
• Amazon Basin