Stage M2 : Monitoring tree phenology and dieback using aerial imagery in an urban forest

Keywords: Leaf phenology, aerial image analysis, functional traits, modeling, climate change

Supervisors: Thomas Caignard (tcaignard@gmail.com), Shannon Brandt (shannon.brandt95@gmail.com), Sylvain Delzon (sylvain.delzon@u-bordeaux.fr)

Location: UMR BIOGECO, Allée Geoffroy Saint-Hilaire, 33615 Pessac
Website: https://www.fondation.univ-bordeaux.fr/projet/forland

Context
Leaf phenology refers to the seasonal events of trees, such as budburst—the date leaves appear in spring—or senescence, the date they fall in autumn. These events are sensitive to climate variations, particularly temperature. As such, they serve as markers of climate change, with studies showing earlier budbursts over time due to rising temperatures (Vitasse et al., 2009). These fluctuations impact local and global ecosystems, influencing species interactions and broader biophysical and biogeochemical cycles. Tracking these events is critical for understanding the mechanisms affecting phenology and predicting forests’ responses to global changes.
Traditionally, leaf phenology is monitored through field observations, providing individual-level data. However, such methods are observer-dependent and time-consuming, especially when monitoring a large number of trees (Liu et al., 2021). Recently, imagery-based monitoring has been rapidly advancing, particularly due to progress in satellite observation and artificial intelligence (Schwartz et al., 2024). However, current satellite imagery lacks the resolution to track individual trees in mixed forests. Aerial drone imagery and photogrammetry techniques offer cm-level resolution, allowing individual tree monitoring across landscapes (Klosterman & Richardson, 2017). In parallel, the use of dendrometers allows precise measurement of the tree growing season.
The experimental forest of the University of Bordeaux is an urban forest located at the heart of the Bordeaux metropolitan area. Due to its proximity to the city, this forest faces various biotic and abiotic stresses that affect tree phenology and tree dieback. Since 2022, budburst and leaf senescence monitoring has been carried out on around sixty trees of different species using binoculars and drones, and wood phenology was measured with dendrometers. Their health is also assessed annually based on observations.

Objectives
The main objective of this project is to inter-calibrate aerial imagery data with field observations and wood phenology data to develop a model for expanding the monitoring of tree phenology and tree dieback to a larger number of individuals. The internship will combine field observations, data analysis, and mapping. The student will define indices to assess tree dieback, and monitor budburst based on the different methods. These results will help characterize the overall response of the urban forest to climatic and microclimatic variations. Eventually, these data will be used to calibrate and refine individual tree monitoring on a larger scale using satellite imagery. The student will have the opportunity to collaborate with and present results in both national and international laboratories.

Desired Profile:
Proficiency in image processing and analysis tools (preferably applied to drone-based aerial imagery); Familiarity with programming in R and statistical analysis skills, Knowledge in forest ecology

To apply:
Send a CV and a cover letter to the supervisors.