PhD Studentship: Machine Learning and Earth Observation for identifying drivers and barriers to post-fire recovery of forests

This large-scale ecological project investigates the barriers and drivers of post-fire forest recovery. With climate change and the spread of forest fires to new areas, it is important to investigate the conditions that support forest recovery after a fire.

The study areas can be defined using global forest fire maps from 2000–2023, provided by Hansen et al. The relaunch of GEDI provides repeated global LiDAR measurements related to biomass. Sentinel missions offer increased temporal resolution in both multispectral (for vegetation indices) and SAR data (for moisture and structure). Regarding climate-related data, Landsat and Sentinel-3 provide land surface temperature. BOKU offers daily precipitation measurements across Europe. Land-related parameters are available through JULES.

The DR would co-develop the research objectives and select the methods to be implemented with supervisory support. Some ideas to discuss include integrating repeat GEDI LiDAR surveys with time-series multispectral and/or SAR data to improve biomass recovery estimations, measuring biases between GEDI and EO time-series estimations, developing customised hybrid neural networks (e.g., CNN-LSTM for capturing both spatial and temporal patterns) to understand the effects of varying climatic conditions on post-fire recovery, and evaluating spatial recovery patterns across different elevations, slopes, and forest types.

Advert information