ArctExa: Towards Exascale Computing for Monitoring Arctic Ice Melt - PhD

Lancaster University

Lancaster, UK 🇬🇧

Project institution: Lancaster University

Project supervisor(s): Prof Mal McMillan (Lancaster University), Dr Dave McKay (University of Edinburgh), Dr Jenny Maddalena (Lancaster University) and Dr Israel Martinez Hernandez (Lancaster University)

Overview and Background

This project offers the exciting opportunity to be at the forefront of research to exploit the potential of exascale computing, for the purposes of satellite monitoring of Earth’s polar regions, at scale.

The Arctic is one of the most rapidly warming regions on Earth, with ongoing melting of the Greenland Ice Sheet and Arctic ice caps making a significant contribution to global sea level rise. As Earth’s climate continues to warm throughout the 21st Century, ice melt in the Arctic is expected to accelerate, leading to large-scale social and economic disruption.

Satellites provide a unique tool for monitoring the impact of climate change upon the Arctic, and are key to tracking the ongoing contribution that ice masses make to sea level rise. With recent increases in data volumes, computing power and the use of data science, comes huge potential to rapidly advance our ability to monitor and predict changes across this vast and inaccessible region. However, currently this potential is not fully realised.

This project will place you at the forefront of this research, working to advance our current capabilities towards exascale computing, through a combination of state-of-the-art satellite datasets, high performance compute, and innovative data science methods. You will be supported by a multidisciplinary supervisory team of statisticians, computer scientists and environmental scientists, with opportunities to contribute to projects run by the European Space Agency. Specifically, this project aims to develop new large-scale estimates of surface meltwater fluxes from all Arctic ice sheets and ice caps into the ocean and, in doing so, better constrain their contribution to sea level rise over the past two decades.

Methodology and Objectives

Project Aim: This project aims to utilise new streams of satellite data, alongside advanced statistical algorithms and compute, to transform our ability to monitor glacier melt at the pan-Arctic scale. More specifically, the successful candidate will develop new estimates of ice cap and ice sheet melt using high-volume, high-resolution datasets from the latest NASA and ESA satellite altimeters. These will be used to determine the first large-scale estimates of meltwater run-off into the Arctic Ocean.

Methods Used: This project will build upon recent proof-of-concept work developing Kalman Smoothing Data Assimilation techniques to create and analyse a unique record of ice melt. The focus of this PhD will be to apply these methods to the latest high-volume satellite altimetry datasets, and to do so at a massive scale. To fully exploit these big data streams and to do so at the pan-Arctic scale, will necessitate the use of Graphical Processing Units (GPU’s) on High Performance Computing (HPC) clusters. As such, developing the code to work on this high-level computing architecture will be a key element of the project. Within the first year of the PhD, the successful candidate will have the opportunity to explore 2 teaser projects, one of which will then be taken forward into subsequent years.

Teaser Project 1: High Resolution Measurements of Greenland Ice Melt over the past 15 years

This teaser project will develop novel estimates of Greenland ice melt over the past 15 years, based upon state-of-the-art CryoSat-2 swath altimetry satellite data. Specifically, a Kalman Smoothing approach, which has recently been tested within our group at several small-scale sites, will be further developed and deployed at scale, with the aim of mapping elevation changes across the entire ice sheet at high resolution. To achieve this, will require the current prototype code to be refactored and then deployed for the first time on GPU-enabled systems. Depending on the progress made, there will also be the opportunity to integrate other data streams, for example to include complementary measurements from the Sentinel-3 high-resolution Synthetic Aperture Radar altimeters.

Teaser Project 2: Towards pan-Arctic Monitoring of Ice Melt

The second teaser project will make use of the same Kalman Smoothing approach introduced above. This will ensure close synergy and complimentary between both of the first year teaser projects, thereby ensuring that the student reaps maximum gain from the development of their technical skills around this subject. Here, the student will deploy, for the first time, the Kalman Smoothing approach to monitor a small, and highly sensitive, Arctic ice cap, such as Austfonna on the Svalbard Archipelago. These smaller ice caps represent a more challenging target for satellite-based monitoring, and so alongside CryoSat-2 the student will also test the use of complementary ICESat-2 photon counting altimetry within the Kalman framework. Because of the high data volumes and the longer-term ambition to operate at the pan-Arctic scale, this project will again work to deploy the chain on GPUs.

In later years of the PhD, depending upon the student’s interests, there will be the opportunity either to extend this work to integrate output from Regional Climate Model simulations, to build more sophisticated machine learning elements into the processing chains, or to utilise other diverse streams of high-volume data, such as ultra-high resolution Digital Elevation Models or historical satellite missions.

Informal enquiries are welcome; please contact Prof Mal McMillan.

References and Further Reading

  1. Antarctica’s ice is melting 5 times faster than in the 90s
  2. Climate change: Satellite fix safeguards Antarctic data
  3. Greenland lost a staggering 1 trillion tons of ice in just four years
  4. CPOM
  5. CEEDS

5 days remaining

Apply by 17 February, 2025

POSITION TYPE

ORGANIZATION TYPE

EXPERIENCE-LEVEL

DEGREE REQUIRED

You ad could be here!