About the project
Background
Dendroclimatology, the science of extracting past climate records from the distinct properties of annual tree rings, plays a critical role in climate research. The discipline has been widely applied in temperate and boreal regions, where strong annual cycles in temperature and light availability promote clear annual ring formation in trees. In tropical regions, however, its application has remained limited because tree-ring formation in many tropical species is often complex and less distinct due to low seasonal climatic variation and high intra-seasonal climate fluctuations.
As a result, long-term paleoclimate records in tropical countries, particularly in Africa, remain scarce, making African climate science among the least developed globally. Yet, tree rings represent one of the most widely available climate-proxy data sources across the continent, highlighting the urgent need for improved methods and technologies for detecting and characterizing tropical tree rings.
Hyperspectral imaging (HSI) offers a promising solution for this challenge. HSI enables detailed characterization of physical and chemical properties of materials at the molecular level by capturing large amounts of spatially resolved spectral information. Combined with deep learning (DL), which can automatically identify complex patterns and features in large image datasets, HSI has strong potential to transform tropical dendroclimatology research.
Objectives
The overall objective of the ATHIL project is to develop and test a novel tropical dendroclimatology approach by integrating hyperspectral imaging (HSI), deep learning (DL), and advanced multivariate data analysis to improve the detection and characterization of climate proxy signals in tropical tree rings.
To achieve this objective, the project will:
- Develop and optimize hyperspectral imaging approaches for tree-ring analysis by exploring different HSI systems, sample preparation methods, instrumental configurations, and light conditions suitable for detecting and characterizing tree rings from various wood samples.
- Apply deep learning and advanced data-processing techniques to automatically extract meaningful spectral signatures and complex patterns from hyperspectral datasets.
- Assess the potential of HSI for detecting and differentiating individual tree rings in selected tropical tree species.
- Evaluate the ability of HSI to characterize physical tree-ring properties linked to climate variability.
- Investigate whether HSI can identify, classify, and quantify the chemical composition of tree rings as potential climate proxy indicators.
- Test the applicability of portable hyperspectral systems for field-based analysis of wooden samples.
Participants
Partners
Publications
Akinyemi, O.O., Butt, M.H.F., Keinänen, M. and Rannestad, M.M., 2025, November. HYFORMER: A Vision Transformer AI Model for Identifying Tropical Tree Species Using Hyperspectral Images of Wood. In 2025 15th Workshop on Hyperspectral Imaging and Signal Processing: Evolution in Remote Sensing (WHISPERS) (pp. 1-6). IEEE.