Scott Hosking is an Environmental Data Scientist at the British Antarctic Survey (BAS) which is part of the Natural Environment Research Council (NERC). He is a Co-Leader of the BAS Artificial Intelligence Lab (BAS AI Lab) and has over 15 years’ experience in atmospheric and climate research, including his PhD research with the University of Cambridge. He continues to work closely with the University of Cambridge and is a Co-Director of the UKRI Centre for Doctoral Training (CDT) in the Application of AI for the study of Environmental Risks, and is a Co-founder and Board member of the Cambridge Centre for Climate Science (CCfCS).
His research involves the application of machine learning on vast and various environmental datasets including the output of global and regional climate models, satellite sensors, and in situ surface measurements. The primary aims of Scott’s work are to identify and understand the key mechanisms that drive year-to-year climate variability over vulnerable regions, the changes in frequency and strength of extreme weather events, and to reduce uncertainties in future climate predictions.
- Developing a Digital Twin Earth, Digital Twinning for the natural environment (e.g., Ocean, Cryosphere, Forests)
- Application of AI for climate risks, and the prediction of high-impact weather events
- Intelligent post-processing of climate data, including bias correction and downscaling
- Simplifying data analysis pipelines for the application of AI, including Pangeo.io (a cloud platform for Big Data geoscience)
- Probabilistic machine learning, providing robust uncertainty estimates for business and environmental policy decision making
- Combining Little Data and Big Data
- Scalable data science methods for large N-dimensional spatiotemporal datasets
- Flexible approaches for irregularly sampled and fragmented datasets,
- Causal inference: understanding the drivers of environmental change