Tree-Ring Database

Understanding the magnitude and cause of variation in tree growth is central to successful sustainable forest management in terms of informing key policy issues and precisely estimating present and future changes in available wood fiber and biomass resources. The Canadian Forest Service (CFS), along with several partners, has built a tree ring database called the CFS-TrenD, which allows the assessments of individual tree, tree population and forest ecosystem vulnerabilities to drought stress or other changing forest disturbance regimes (e.g. insects, diseases, fire).

Project objectives

This database is extremely large and has been cleaned to the best of researcher's abilities, it is not absent of errors. Specifically, the database was built with data from approximately 40 contributors, and so there are errors data labeling and consistency. The current process of quality control requires extensive manual effort, even with the use of advanced computing and programming techniques. The process for a machine learning (ML) project is normally two-fold: data cleaning and preparation, and then building the ML models for a specific purpose (e.g. predicting a result). This project is taking a novel approach by applying machine learning to the data cleaning and preparation part of the process. In summary, artificial intelligence (AI) techniques are being applied to compare different samples, build intelligent algorithms that determine when the data is faulty and correct as required.

The objectives are to:

• Develop an online tool that makes use of AI for tree-ring data collection
• Optimise workflows for data cleaning (anomaly detection), cross-dating (pattern recognition) and data correction
• Facilitate the inclusion of the tree-ring database project into the TreeSource website developed by the Canadian Wood Fibre Centre
• Implement data storage, and data visualization tools for online analysis in the TreeSource website
• Enable new research (e.g. meta-analyses) to better understand the drivers, patterns and implications of changes in tree growth
• Facilitate new research partnerships and contribute to stakeholder needs for sustainable forest management