deepT

Deep neural network algorithms to recognize species and their specific response patterns in stem growth data of TreeNet

deepT investigates times series of stem radius changes with deep neural network (DNN) algorithms in order to recognize tree species, growth patterns and tree water deficit-induced stem shrinkage characteristics. deepT tests different types of DNN algorithms, develops a Python environment for extensive data sets and quantifies the reliability of different approaches to recognize species-specific patterns and judge their potential for making predictions.

deepT is a collaboration bewtween EMPA (Mirko Lukovic, Cellulose & Wood Materials, Empa Dübendorf) and WSL.

 

Publications:

Reconstructing radial stem size changes of trees with machine learning
Mirko Luković, Roman Zweifel, Guillaume Thiry, Ce Zhang and Mark Schubert
Journal of the Royal Society Interface, 19, 20220349 (2022).