Overview
Across the city of Liverpool mortality differs by up to 15 years – this is mainly based on deprivation. Our research has shown that every neighbourhood has different levels of air pollution – yet models still only predict at city scale. This groundbreaking work will aim to develop ML based hyper resolution air quality models of the city of Liverpool. Such work will help with inform local government for specific traffic calming measures and even help us target the most poor to have a significant positive impact of their lives.
About this opportunity
This project will focus on developing new techniques for integration of diverse environmental datasets into foundational AI models for a first use case of air quality modelling. In port cities like Liverpool, air pollution models are needed to provide local decision-makers with information to support local urban planning initiatives, such as re-rerouting traffic or creating pedestrian-friendly zones. Such models require integration of low resolution country-scale pollution observations and forecasts with local sensor observations, which can also be complemented by fluid dynamics simulations. This project aims to develop a super-high-resolution spatiotemporal air quality model, leveraging the latest advances in AI, such as geospatial foundation models, multi-modal fusion, and physics-
-informed machine learning. Datasets will include pollution forecasts from the UK Met Office and localised air quality measurements, along with the Liverpool Metaverse: a high resolution digital elevation model of Liverpool and its port.
The project provides the opportunity to work with multi-disciplinary teams in academia and industry. The supervisory team at the University covers explainable AI, air quality monitoring and visualization, modelling in Computational Fluid Dynamics (CFD), multivariate statistical analysis, and policy on port city air quality, while the team at IBM Research will bring world-leading expertise in the development and deployment of AI on geospatial data with the development of their Prithvi family of geospatial foundation models for earth observation and weather/climate data.
Applicants must have, or expect to obtain, a Masters (or equivalent) degree in Physics, Maths, Computer Science, or related subject. Previous experience in analysis of geospatial weather or pollution-related data at scale is an advantage, but not essential. Applications should include a Cover Letter, detailing your experience and motivation for PhD studies (300 words max), and a CV with contact details of two referees.
Who is this opportunity for?
People who want to learn and be future leaders in an exciting field.
Further reading
Sokhi et al. (2022). Advances in air quality research – current and emerging challenges, Atmos. Chem. Phys., 22, 4615–4703, https://doi.org/10.5194/acp-22-4615-2022.