PhD Scholarship
Developing Multimodal Foundation Models for Volcano Monitoring Using Satellite Radar
- Visual Information Laboratory, School of Computer Science, University of Bristol
- Scholarship Details: Home fee rate
- Stipend: Royal Society rate (£22,797 for 2027/28 , subject to annual adjustment every April)
- Duration: 4 years
- Eligibility: Home (UK) and EU citizens who have confirmation of UK settlement or presettlement status under the EU Settlement Scheme. If you are an overseas student, please contact me for potential fee waiver.
- Start date: September 2027
- Prof. Pui Anantrasirichai, School of Computer Science
- Prof. Juliet Biggs, School of Earth Sciences
Project description
Volcanic eruptions reshape landscapes through lava flows, ash deposits and the growth or collapse of volcanoes. Satellite radar can observe these changes through clouds and darkness, when optical images are unavailable. However, radar signals are difficult to interpret because surface roughness, terrain and ground movement can produce overlapping effects. This creates a machine learning challenge in distinguishing meaningful volcanic changes from noise and other sources of variation.
This four-year PhD will develop multimodal AI foundation models (FMs) that combine radar and optical satellite data to analyse volcanic activity. As part of the Magma4D programme (led by Prof Juliet Biggs in the School of Earth Sciences), the student will investigate how a shared model can learn from complementary radar measurements. These include backscatter, describing reflected signal strength; coherence, measuring signal similarity between observations; and interferometric SAR, providing information about ground movement. The project will explore how approaches such as MAE, DINOv3 and CLIP can support representation learning and transfer between volcanoes with limited labelled data.
The student will investigate Transformer architectures with temporal attention or memory mechanisms to learn how volcanic surfaces change across image sequences. Combining radar representations with optical imagery will provide complementary information about volcanic deposits. Lightweight prediction heads, attached to the shared model, will support selected tasks such as mapping lava flows, detecting newly affected areas and removing noise from ground-movement measurements. Evaluation will assess accuracy, uncertainty, computational efficiency and performance on previously unseen volcanoes.
The project offers training in satellite data analysis, computer vision, deep learning and geospatial programming. Research can develop around the student’s interests, including self-supervised learning, multimodal fusion and efficient model adaptation. Working across computer science and volcanology, the student will develop reusable methods that help scientists interpret satellite observations and monitor volcanic change more reliably over time.
The student will be hosted by the Visual Information Laboratories, a leading research group in image processing and computer vision, and will have access to Isambard-AI’s supercomputing facilities. The student will also collaborate with volcanologists and other scientists through COMET, the UK Centre for the Observation and Modelling of Earthquakes, Volcanoes and Tectonics.
Entry requirements
Applicants must hold or achieve a minimum of a master’s degree (or international equivalent) in a relevant discipline. Applicants without a master’s qualification may be considered in exceptional cases, provided they hold a first-class undergraduate degree. Please note that acceptance will also depend on evidence of readiness to pursue a research degree. More details can be found here.
How to apply
- All candidates should submit a full CV and covering letter to Prof. Pui Anantrasirichai.
- The opportunity remains open until filled.