Global sea level rise has been accelerating at an alarming rate for the last couple of decades, endangering and displacing millions of people. The majority of this sea level rise can be attributed to ice mass loss processes. In particular, ice mass loss from the Greenland ice sheet (GrIS) is a major contributor to sea level rise. Every year, about half of the ice that is lost from Greenland, and added to the oceans, comes from melting ice from the surface of the ice sheet. A large driver of ice surface melting is solar radiation. The amount of solar radiation that is absorbed vs reflected is strongly controlled by the ice albedo, with dark surfaces having a lower albedo (and absorbing more solar radiation) than light surfaces.

The temporal and spatial complexity and variability of ice albedo on the GrIS are not well understood. This makes it very difficult to accurately represent this process in the climate models we use to predict GrIS ice mass loss, which can lead to inaccurate predictions of global sea level rise. A large part of the complexity of ice albedo stems from the presence of light-absorbing impurities (e.g. dust, black carbon, algae) on the ice surface that lower the albedo. We lack a complete understanding of the physical processes that control the distribution and evolution of these impurities.

We are therefore working towards training machine learning models (random forest, XGBoost, CNN-LSTM) to model the variability of ice albedo using climate model output and satellite data. The DSI Scholar would further develop this work by implementing diffusion-based techniques, creating a super-resolution framework to increase the resolution of the satellite data of ice albedo, or performing an attribution analysis to find drivers of ice albedo variability from atmospheric and ice surface processes.

This project is eligible for a stipend, with matching funds from the faculty advisor and the Data Science Institute. This is not a guarantee of payment, and the total amount is subject to available funding.

Faculty Advisor

  • Professor: Tedesco, Marco
  • Center/Lab: Lamont-Doherty Earth Observatory
  • Location: Lamont-Doherty Earth Observatory

Project Timeline

  • Earliest starting date: 4/3/2023
  • End date: 6/30/2023
  • Number of hours per week of research expected during Spring-Summer 2023: ~12

Candidate requirements

  • Skill sets: Python, machine learning
  • Student eligibility: freshman, sophomore, junior, senior, master’s
  • International students on F1 or J1 visa: eligible
  • Academic Credit Possible: Yes