GreenHyperSpectra: A multi-source hyperspectral dataset for global vegetation trait prediction

2025-07-09

AI Summary

Conventional field sampling struggles to capture large-scale spatial variability in vegetation functional traits (e.g., leaf carbon content, specific leaf area), while hyperspectral prediction suffers from severe domain shift across sensors and ecosystems and is hampered by scarce labeled data. Method: We introduce the first global, multi-source hyperspectral pretraining dataset explicitly designed for semi-supervised and self-supervised learning, enabling cross-domain generalization evaluation. We propose a contrastive learning–based multi-output regression framework that optimizes spectral representations to enhance robustness in low-data regimes. Contribution/Results: Our method significantly outperforms fully supervised baselines both in-distribution and out-of-distribution, improving label efficiency by over 40%. The code and dataset are publicly released.

Technology Category

• Machine Learning: Semi-Supervised Learning
• Computer Vision: Multi-modal Vision
• Search and Optimization: Learning to Search

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Plant traits such as leaf carbon content and leaf mass are essential variables in the study of biodiversity and climate change. However, conventional field sampling cannot feasibly cover trait variation at ecologically meaningful spatial scales. Machine learning represents a valuable solution for plant trait prediction across ecosystems, leveraging hyperspectral data from remote sensing. Nevertheless, trait prediction from hyperspectral data is challenged by label scarcity and substantial domain shifts (eg across sensors, ecological distributions), requiring robust cross-domain methods. Here, we present GreenHyperSpectra, a pretraining dataset encompassing real-world cross-sensor and cross-ecosystem samples designed to benchmark trait prediction with semi- and self-supervised methods. We adopt an evaluation framework encompassing in-distribution and out-of-distribution scenarios. We successfully leverage GreenHyperSpectra to pretrain label-efficient multi-output regression models that outperform the state-of-the-art supervised baseline. Our empirical analyses demonstrate substantial improvements in learning spectral representations for trait prediction, establishing a comprehensive methodological framework to catalyze research at the intersection of representation learning and plant functional traits assessment. All code and data are available at: https://github.com/echerif18/HyspectraSSL.

Problem

• Predict global vegetation traits using hyperspectral data
• Address label scarcity and domain shifts in trait prediction
• Develop robust cross-domain methods for plant trait assessment

Innovation

• Multi-source hyperspectral dataset for vegetation traits
• Semi- and self-supervised pretraining methods
• Label-efficient multi-output regression models