Next generation forest dynamics modelling using remote sensing data

Globally, human-induced climate change and biodiversity loss threaten ecosystem function and the services the biosphere provides for humans. Forests are carbon-dense ecosystems and are home to the majority of terrestrial biodiversity, so are crucial tools to mitigate adverse impacts. Indeed, many countries, including many in Europe, have ambitious policies to restore and replant forests to restore carbon and habitats. However, forests are themselves threatened by climate change and biodiversity loss, so understanding and predicting their future in the face of global change is a priority.

In order to understand how forests are changing, and how they will change in the future, we need large monitoring networks collecting data, to embrace new measurement techniques, to fuse data from multiple sources, and to create robust, data-driven, predictive models. Traditional forest data is severely limited in both its spatiotemporal coverage and what it can measure, and whilst existing ecological models are tailored to such data, these focus on the small scale and cannot predict the future of forests at large enough scales to help understand the impacts of climate change. New approaches are needed.

This fellowship will use cutting-edge remote sensing data and modern data science techniques to generate new understanding of current and future forest functioning. Active and passive remote sensors, including terrestrial and drone laser scanning and structure from motion photogrammetry, are able to capture the full three-dimensional structure of a forest to sub-cm scale within three-dimensional point clouds. This fellowship will collect and collate such data from tens of thousands of trees across hundreds of forest plots in Europe, creating a massive new dataset of tree and forest structure. Such data are extremely complex to analyse, and the project will use specially developed and tailored deep learning techniques to extract ecological information from noisy point clouds. Some plots that have already been measured will be re-measured, to capture three dimensional tree growth and forest structural change.

The fellowship will analyse these data to determine how trees and forests are structured across Europe, and how their three-dimensional structure affects and is affected by their productivity, carbon storage, and the diversity of both the trees and other species living in forests. New insights into how biodiversity is related to three-dimensional structure will bring help develop approaches to co-monitoring biodiversity and biomass, crucial for demonstrating the value of ecosystems towards tackling both climate change and biodiversity loss.

Using newly developed software, the fellowship will scale local, single-measurement plot-scale information to continental scale and continuous monitoring by fusing ground and Earth Observation (satellite) data. Using deep learning to link the structural and diversity information from hundreds of thousands of plot locations across Europe with the spectral properties measured by satellite sensors, the fellowship will bring new understanding on how forests are structured and how they are changing across Europe. Finally, using findings from all parts of the fellowship, a new modelling framework which can predict ecological change on the ground at local scale but which can ingest satellite data will be developed. This data-driven approach will enable robust and updatable predictions of climate change impacts on forest diversity and dynamics across Europe. It will be constructed to be flexible to incorporate future data streams, so informing inform climate change mitigation policy across the continent.