Climate change is bringing more frequent heatwaves, prolonged periods of drought, and other extreme weather events. All of this has a major impact on forest health. The tree species we plant today will shape the landscape for decades to come. Seeking to answer the question of which tree species will thrive under future conditions is Markéta Hošmánková, a student at the Faculty of Information Technology (FIT CTU), through her project within the Research Summer at FIT (VýLeT) program. This program supports student involvement in scientific research already during their studies and enables them to work on their own projects with the potential for scientific publication.
Markéta Hošmánková’s project, titled “Evaluating the suitability of tree species using machine learning,” aims to create an AI model capable of predicting how tree height will develop depending on soil and climatic conditions.
“Tree growth is an important indicator of its vitality and ability to adapt to the environment. The model predicts height increase over five years based on soil and climatic conditions. Such a model can further be used to create autoregressive predictions of long-term tree growth,” Markéta explains.
The foundation of the project is an extensive dataset of measurements of average forest stand heights at five-year intervals across the Czech Republic from 1961 to 2020. The data includes information on tree height and age. Markéta enriched the dataset with detailed location characteristics such as soil properties and climatic indicators (temperature, precipitation, drought occurrence, and solar radiation). Based on these data, AI models were created for the four most common tree species in the Czech Republic: Norway spruce, Scots pine, European beech, and English oak.
As part of the research, traditional forestry models—nonlinear models of height depending on stand age and environmental parameters used by Czech foresters to predict tree growth—were compared with modern machine learning approaches. The best results for most tree species were achieved by a “random forest” model, which most accurately predicted their growth.
“It turned out that climate variables play a crucial role in prediction. Models that did not incorporate them were significantly less accurate,” Markéta adds.
An important feature of the proposed solution is its ability to generate long-term predictions using an autoregressive approach. The model builds on its previous estimates step by step, thereby simulating the development of tree growth over time. Despite the natural accumulation of errors, it maintains relatively high accuracy, making it possible to realistically estimate future forest development under different climate scenarios.
“It has been shown that machine learning can help foresters decide which trees to plant, but the predictions should not be taken literally. Some tree species grow more slowly even under favorable conditions, so simply comparing growth rates of several species does not allow us to determine which is more resilient or suitable. The model mainly serves as a tool that complements expert knowledge and helps professionals make more informed decisions,” Markéta notes. “I plan to continue this research as part of my bachelor’s thesis,” she concludes.
“Participating in VýLeT gave me the opportunity to apply my knowledge to solving a real-world problem. At the same time, through producing an output in the form of a scientific article, it taught me how to present both the process and results of my research in a professional and systematic way. I would definitely recommend this experience to other students—it is very beneficial and undoubtedly useful not only for final theses,” Markéta evaluates the VýLeT program.
Through the VýLeT program, the faculty supports its students in science and research every year. Students work on an independent research task in cooperation with a mentor and contribute to preparing a scientific article for a journal or a paper for a scientific conference. Successful participants can receive an extraordinary scholarship or a financial reward of up to CZK 35,000 for their work.
