About

We aim to harness the big data and the AI revolution to develop innovative technological approaches to assess nature recovery in both fine detail and at large spatial scales.

We are in the midst of an exponential proliferation of data about our environment from sources as varied as a new generation of satellite sensors, social media posts and time-lapse cameras with image recognition. In concert with this data richness, there is immense potential to use AI/machine learning to fuse, interpret and correlate data to work at both fine spatial resolution and large spatial scale in the midst of significant complexity.

For the first time we have the potential to map and model ecological connectivity across whole countries, map different farming approaches or infrastructure in fine detail and track the connectivity of biodiversity associated with different farming landscapes. We will advance state-of-the-art AI approaches to combine different sources of data, including drones, satellite, survey data and social media, that are robust to a range of environmental scenarios, data noise and model reliability.

We will initially focus on three machine learning tasks to address the information requirements of our programmes:

  • The exploitation of existing machine learning technology and subsequent identification and filling of methodological gaps
  • Novel methods for data interpretation
  • Requirements for hardware, imagery and human intervention for cost-efficient, scalable data analytics.

Projects

Theme outputs

    Jesús Aguirre-Gutiérrez, Sandra Díaz, Sami W. Rifai et al. (2025). Tropical forests in the Americas are changing too slowly to track climate change. Nature.

    Species are expected to shift their ranges as the climate changes, but shifts may not occur fast enough, especially for immobile species such as plants. Two papers in this issue assess the degree to which plant species are tracking climate change in the American tropics, where data availability has constrained inference. Ramírez-Barahona et al. show that in Mesoamerican cloud forests, climate change and deforestation together have led to a mean upward shift in species ranges since 1979, mainly due to contracting lower range edges. In tropical forests across the Americas, Aguirre-Gutiérrez et al. found that tree traits are not shifting fast enough to track climate change based on trait-climate relationships, with smaller shifts in montane forests

    Publications
    LCNR supported
    • Remote sensing
    • Scale and Technology
    • Society

    Jesús Aguirre-Gutiérrez, Sami W. Rifai, Xiongjie Deng, Hans ter Steege, Eleanor Thomson, Yadvinder Malhi, et al. (2025). Canopy functional trait variation across Earth’s tropical forests. Nature.

    Tropical forest canopies are the biosphere’s most concentrated atmospheric interface for carbon, water and energy. However, in most Earth System Models, the diverse and heterogeneous tropical forest biome is represented as a largely uniform ecosystem with either a singular or a small number of fixed canopy ecophysiological properties. This situation arises, in part, from a lack of understanding about how and why the functional properties of tropical forest canopies vary geographically4. Here, by combining field-collected data from more than 1,800 vegetation plots and tree traits with satellite remote-sensing, terrain, climate and soil data, we predict variation across 13 morphological, structural and chemical functional traits of trees, and use this to compute and map the functional diversity of tropical forests. Our findings reveal that the tropical Americas, Africa and Asia tend to occupy different portions of the total functional trait space available across tropical forests. Tropical American forests are predicted to have 40% greater functional richness than tropical African and Asian forests. Meanwhile, African forests have the highest functional divergence—32% and 7% higher than that of tropical American and Asian forests, respectively. An uncertainty analysis highlights priority regions for further data collection, which would refine and improve these maps. Our predictions represent a ground-based and remotely enabled global analysis of how and why the functional traits of tropical forest canopies vary across space.

    Publications
    LCNR supported
    • Remote sensing
    • Scale and Technology
    • Society
See all outputs for this theme