Object-Based Deep Learning for Probabilistic Convective-Core Nowcasting from Satellite Data
M. Rakotomanga, D. J. Parker, N. Ben Rached, S. R. Anderson, C. Klein, and S. C. Wells
Quarterly Journal of the Royal Meteorological Society
Featured Research
Object-based deep learning for storm nowcasting
My research focuses on machine learning for satellite-based convective storm nowcasting across Africa. I develop object-based approaches that represent convective systems through their physical and spatial properties rather than treating satellite observations only as images.
A pan-African live version is available on the UKCEH Nowcasting Portal ↗
Publications
Selected academic work
Massive Star Clusters and Clumps in the Collisional Ring Galaxy Arp 147
Z. Randriamanakoto, M. Rakototafika, B. Mongwane, P. Väisänen, and M. Rakotomanga
Monthly Notices of the Royal Astronomical Society
Experience
Research &
technical
experience
2023–present
PhD Researcher
University of Leeds · Machine learning for satellite-based storm nowcasting
2025–2026
Research & AI Nowcasting
University of Leeds · PANCAST operationalisation and pan-African satellite product evaluation
2025
NC-International Visiting Scientist
UK Centre for Ecology & Hydrology · Development of PANCAST
2024–2025
WISER-EWSA Nowcasting Testbed Coordinator
Operational nowcasting activities across southern Africa
2023
Data Science & AI Fellow
Max Planck Institute for Intelligent Systems · Machine learning for gravitational-wave inference
2022
Full Stack Developer Intern
Orange Digital Center Madagascar
Education
Academic
background
2023–2027
PhD in Applied Mathematics
University of Leeds
2022–2023
Pre-PhD Stay in Computer Science
University of Tübingen & Max Planck Institute for Intelligent Systems
2021–2022
MSc in Mathematical Sciences · Cum Laude
AIMS South Africa & Stellenbosch University
2017–2020
Master's in Physics and Chemistry · Magna Cum Laude
École Normale Supérieure & University of Antananarivo
Selected Talks
Presentations
2026
RMetS Annual Weather and Climate Conference
Exeter · Oral presentation · AI in Meteorology panellist
2026
European Nowcasting and Weather Forecasting Conference
Copenhagen · Oral presentation on deep learning for African convective storm nowcasting
2025
EUMETSAT Meteorological Satellite Conference
Lyon · Oral presentation on AI-based nowcasting within WISER-EWSA
Recognition
Awards &
recognition
2026
EPS Leeds Partnership Awards
Highly Commended · Community Impact and Innovation
2023
AIMS–Tübingen Fellowship
Fellowship Programme in Data Science & Artificial Intelligence
2021
Mastercard Foundation Scholar
Full scholarship for MSc studies at AIMS South Africa and Stellenbosch University
Skills
Technical
skills
Programming & Machine Learning
Python, PyTorch, TensorFlow, scikit-learn
Scientific Computing & Data
NumPy, pandas, xarray
High-Performance Computing
Large-scale and distributed model training and evaluation, SLURM
Software & Web
Git, JavaScript, Node.js, Express, React, MongoDB, PostgreSQL, HTML, CSS, LaTeX
Languages
English, French and Malagasy — fluent · German — limited proficiency
Leadership & Outreach
Beyond
research
Science sans Frontières
Science outreach and training in Madagascar.
Leeds Nowcasting Collective
Connecting nowcasting researchers and meteorological services.
Deep Learning IndabaX Madagascar
Leading the hackathon and poster sessions.
Contact