AGBD: A Global-scale Biomass Dataset

2024-06-07

arXiv.org

AI Summary

Existing global aboveground biomass (AGB) datasets face a fundamental trade-off between spatial resolution and geographical coverage: high-resolution local datasets lack representativeness, while coarse-resolution global products fail to capture ecological heterogeneity. To address this, we introduce the first machine-learning-ready, globally complete, 10-meter-resolution AGB benchmark dataset. It integrates GEDI lidar-derived ground-truth measurements with Sentinel-2 optical and ALOS-2 PALSAR-2 SAR imagery to jointly derive multi-dimensional features—including canopy height, topography, and land cover—as well as a seamless global AGB prediction map. Our work achieves, for the first time, unified, high-resolution, multi-biome, and multi-temporal AGB modeling at global scale. We propose an end-to-end, fully reproducible, open-source remote sensing biomass estimation framework. The released dataset supports one-line API access; its accompanying benchmark model demonstrates significantly superior cross-regional generalization compared to state-of-the-art methods, providing high-accuracy foundational support for climate change mitigation and biodiversity conservation.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Efficient ML / Green AI
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Accurate estimates of Above Ground Biomass (AGB) are essential in addressing two of humanity's biggest challenges: climate change and biodiversity loss. Existing datasets for AGB estimation from satellite imagery are limited. Either they focus on specific, local regions at high resolution, or they offer global coverage at low resolution. There is a need for a machine learning-ready, globally representative, high-resolution benchmark dataset. Our findings indicate significant variability in biomass estimates across different vegetation types, emphasizing the necessity for a dataset that accurately captures global diversity. To address these gaps, we introduce a comprehensive new dataset that is globally distributed, covers a range of vegetation types, and spans several years. This dataset combines AGB reference data from the GEDI mission with data from Sentinel-2 and PALSAR-2 imagery. Additionally, it includes pre-processed high-level features such as a dense canopy height map, an elevation map, and a land-cover classification map. We also produce a dense, high-resolution (10m) map of AGB predictions for the entire area covered by the dataset. Rigorously tested, our dataset is accompanied by several benchmark models and is publicly available. It can be easily accessed using a single line of code, offering a solid basis for efforts towards global AGB estimation. The GitHub repository github.com/ghjuliasialelli/AGBD serves as a one-stop shop for all code and data.

Problem

• Lack of global high-resolution biomass datasets
• Inconsistent biomass estimates across vegetation types
• Need for machine-learning-ready AGB benchmark data

Innovation

• Combines GEDI, Sentinel-2, and PALSAR-2 data
• Includes pre-processed high-level feature maps
• Provides high-resolution global AGB predictions