Changing climatic circumstances have a significant impact on forests: besides higher temperatures, more intensive and frequent storms and drought spells are major challenges for forest conservation and management in the future. European Forests, on the other hand, have been a significant carbon sink within the last decades, allowing for natural carbon dioxide removal from the atmosphere and so aiding in the mitigation of climate change and its consequences. Tree growth is a key indicator of tree health and carbon sequestration. Growth dynamics and underlying physiological processes are, however, of great complexity and challenging for traditional statistical modelling approaches. Moreover, they do not allow considering instantaneous changes in growth due to climatic extremes (i.e. drought, heat) and the changes of phenological patterns (i.e. length vegetation season) which can nowadays being monitored with the latest measurement equipment.For that reason and since data availability has increased substantially, AI technologies have become increasingly popular in recent years.

The proposed project aims at a unique modelling approach for tree growth. Machine learning is enabled by the availability of detailed long-time datasets on ecological and climatic parameters through an extensive forest monitoring network in Austria respectively all over Europe. Within the network, growth data is available on different time scales allowing studying different aspects of growth dynamics: physiological tree processes and interactions between growth and climate variables can be analysed using hourly single tree growth data; stand characteristics gathered every couple of years on the other hand provide information on long-term tree development and carbon sequestration. Another important part of the project is the integration of airborne and terrestrial laser scanning in order to obtain detailed information on the studied tree stands. In addition to that, satellite derived parameters will be actively incorporated into the developed models. This combined approach allows developing a model that can map both high- and low-frequency sampled features for growth modelling. Furthermore, using AI with an explainable error analysis module, growth parameters can be identified, aiding researchers in developing and planning monitoring setups for different research questions.

AI4Trees will thus research, develop and combine technologies that would allow us to understand and explain tree growth, an essential component for optimizing carbon segregation, biodiversity and climate adaptation in forest ecosystems, empowering response to minimize potentially harmful consequences for modern societies in line with the UN Sustainable Development Goals.