PhD Position in Dynamics-Aware Downscaling of Extremes over Arctic Glaciers

PhD Position in Dynamics-Aware Downscaling of Extremes over Arctic Glaciers

Extreme Arctic glacier melt is driven by multi-scale compound drivers. We will leverage machine learning methods to bridge the gap between drivers at coarse model resolutions and impacts captured by high-resolution observations.

Job description

Arctic glacier melt drives regional water availability, glacial outburst flood risks, freshwater input to the ocean, and global sea-level rise. A growing body of evidence shows that a substantial portion of glacier runoff results not from gradual warming, but from short-lived, intense melt and precipitation events. These events are fueled by global-scale atmospheric drivers and amplified by local topographic conditions. However, current global climate models (GCMs) lack the spatial resolution to capture these processes, while high-resolution regional models remain too computationally expensive for large ensemble use. Frequently, even higher resolutions do not capture dynamic drivers. This mismatch limits the utility of climate projections for predicting localized glacier change.

This project aims to close this gap by developing a machine learning-based downscaling framework that links coarse resolution (0.25°-1°) reanalysis and climate model outputs to fine-scale (∼100 m) estimates of surface melt and precipitation based on observations across Arctic glaciers, particularly the drivers and impacts of extreme events.

As a PhD student at TU Delft, you will work on Arctic science at the nexus of models, AI and spaceborne remote sensing. You will first identify large-scale drivers of compound extremes in models and observations, then build an emulator using advanced AI methods, such as convolutional neural networks and diffusion models, to estimate high-resolution melt and precipitation fields. This emulator will be used to drive surface mass balance and glacier models, substantially expanding the impact of climate simulations.

This work will quantify the contribution of extreme events to glacier runoff and mass loss, improve predictions of future glacier change under multiple climate pathways, and provide open-source tools for broader application. This project will advance scientific understanding of cryosphere-atmosphere interactions and support societal adaptation through improved projections of Arctic glacier behavior under climate change.

Your project will be conducted within the Geoscience and Remote Sensing Department at TU Delft with multiple collaborators on-site, interacting with a vibrant community on the TU Delft campus. You will also have regular contact with collaborators at Vrije Universiteit Brussels (involving one extended trip for the PhD candidate) as well as the National Center for Atmospheric Research (NCAR) in the US. Our group at TU Delft is dedicated to building a collaborative, supportive environment which will help you flourish both personally and professionally.

Job requirements

• You hold an MSc in Earth science, environmental science, data science, physics, mathematics or computer science, with practical machine learning/artificial intelligence courses and relevant project and thesis experience.
• You have a keen interest in or experience working in the domain of the cryosphere (the part of the world covered in ice), and interact with the larger science community.
• You are intrinsically excited about the prospect of creative collaborative science.
• You are motivated to develop expertise independently with multiple tools, including observations and models, AI and statistical methods.
• You have a good command of written and spoken English, as scientific work will be conducted in English.

Faculty of Civil Engineering and Geosciences

The Faculty of Civil Engineering & Geosciences (CEG) is committed to outstanding international research and education in the field of civil engineering, applied earth sciences, traffic and transport, water technology, and delta technology. Our research feeds into our educational programmes and covers societal challenges such as climate change, energy transition, resource availability, urbanisation and clean water. Our research projects are conducted in close cooperation with a wide range of research institutions. CEG is convinced of the importance of open science and supports its scientists in integrating open science in their research practice. The Faculty of CEG comprises 28 research groups in the following seven departments: Materials Mechanics Management & Design, Engineering Structures, Geoscience and Engineering, Geoscience and Remote Sensing, Transport & Planning, Hydraulic Engineering and Water Management.