The research projects that are carried out in the CDT utilise diverse and often untapped environmental data sets and are based upon two key themes:
- Environmental data classification, integration & analysis - using machine learning to process data to provide actionable information. AI will underpin next generation data-analytics systems that can, e.g., process data from diverse sources (including sensors on the ground, in the air or in space) and classify them into categories that humans can understand. The data can be optimally combined to generate, e.g., key indices to track progress against sustainable development goals or information to ensure conservation efforts and resources are deployed efficiently and cost-effectively.
- Environmental modelling - developing new computer models for environmental problems using data-based approaches. Smart post-processing of model output using data-driven approaches can generate bespoke results, e.g., bias-correction, downscaling and optimal weighting of ensemble climate model output to generate high-resolution decision-relevant information. AI approaches can also be used to form reduced models or to develop new empirical parameterisations for models.
These themes also touch on widely-applicable emerging methodologies (e.g. provenance, data and model curation), and serve as context in which different modelling methodologies can be compared.
The primary application areas are in:
- Built environment and energy
- Carbon capture and carbon credits
- Climate & air quality
- Ecology and biodiversity
- Geophysical and environmental modelling
- Natural resources and forests
- Space, oceans and polar science