Tracking Climate Change with Satellites and Artificial Intelligence (2023-2024)

Background

Climate change is beginning to impact infrastructure, transportation systems, energy, food, water supplies and human health, with Africa and Asia being two of the most vulnerable regions to the impacts of climate change. However, these impacts are often difficult to quantify, especially in regions where ground-based economic surveys occur infrequently.

For climate adaptation and resilience planning, information is needed on vulnerable infrastructure such as transmission lines, water systems, food supply chains and transportation networks as well as climate hazard exposure to severe weather, fires, floods and the consequent damage. To make the best possible plans and track progress and impacts, infrastructure data needs to be quantified, tracked and communicated. 
Thus, the goal of this project is to democratize access to data relevant to climate change mitigation and adaptation planning as well as the underlying models that can be used to acquire those data.

Project Description

Building on the work of previous teams, this team will create a “foundation model” to enhance climate change mitigation and adaptation planning that can be used to monitor a broad range of climate change contributing factors and impacts. This advancement will provide evidence to inform climate mitigation and adaptation strategies and hasten those strategies’ evaluation and implementation by key stakeholders and decision-makers.

A foundation model is a model (usually a deep neural network) that has been trained on a large and diverse set of data, after which it can be adapted to a variety of related inference tasks. Historically, the development of foundation models has required access to massive quantities of labeled imagery. However, recent research has shown that effective foundation models can be developed without any hand-labeled imagery using self-supervised learning. 

Therefore, the project will aim to develop a foundation model specifically for remote sensing imagery using self-supervised learning techniques. Team members will investigate the efficacy of such a model in monitoring a wide range of variables relevant to the causes and consequences of climate change. 

The project’s long-term vision is to: 

  • Gather more evidence more quickly by using remotely sensed data, without reliance on expensive ground-based economic surveys
  • Facilitate more equitable and broadly applicable climate data gathering
  • Enable a wide range of custom queries regarding climate data needs for anywhere on Earth through the integration of a natural language processing encoder

Each of these elements will contribute to the goal of directly informing climate change mitigation and strategic decision making.

Anticipated Outputs

Paper(s) submitted to a relevant journal or conference; new dataset for self-supervised learning of geospatial data; open-source codebase; online tool for text-based search of key climate and energy-relevant infrastructure 

Student Opportunities

Ideally, this team will include 2 graduate students and 6 undergraduate students. Interested students may come from disciplines such as engineering, computer science, economics, public policy, environmental science, mathematics and statistics. Students with experience in quantitative methods are preferred. Experience with programming is required, and experience with Python programming is especially beneficial.

Students will have the opportunity to learn about the multiple facets of climate change, engage in team-based problem solving, apply a formalized research process, use advanced data analytics tools, learn the fundamentals of machine learning and further their verbal and written communication skills. Graduate students will gain valuable project management and leadership experience.

There will be a related Climate+ project for Summer 2023; there is a separate application process for students who are interested in this optional component.

A graduate student will be selected to serve as project manager.

Timing

Summer 2023 – Spring 2024

  • Summer 2023 (optional): Complete Climate+ project
  • Fall 2023: Complete intensive short course on data science, machine learning, climate change, energy systems and research techniques; begin executing research
  • Spring 2024: Continue and conclude research; create a website, final presentation and paper for submission to a relevant conference or journal

Crediting

Academic credit available for fall and spring semesters; summer funding available

See related Climate+ summer project, Tracking Climate Change Causes & Impacts With Satellites and AI (2023), and earlier related team, Tracking Climate Change With Satellites and Artificial Intelligence (2022-2023).

 

Image: Kangerlussuaq Glacier, Greenland, by European Space Agency, licensed under CC BY-SA 2.0

Satellite view of glacier.

Team Leaders

  • Kyle Bradbury, Pratt School of Engineering-Electrical & Computer Engineering|Energy Initiative
  • Jordan Malof, Computer Science Department, University of Montana

/graduate Team Members

  • Darui Lu, Electrical/Computer Engg-MS
  • Song Oh, Interdisciplinary Data Science - Masters

/undergraduate Team Members

  • Irene Biju
  • Muaz Bin Kashif, Computer Science (BS)
  • Kushagra Ghosh, Computer Science (BS)
  • Malini Kamlani
  • Xinshi Ma, Economics (BS)
  • Morgan Pruchniewski, Statistical Science (BS)
  • Anushka Srinivasan

/yfaculty/staff Team Members

  • Leslie Collins, Pratt School of Engineering-Electrical & Computer Engineering
  • Marc Jeuland, Sanford School of Public Policy
  • Luana Lima, Nicholas School of the Environment-Environmental Sciences and Policy
  • Robyn Meeks, Sanford School of Public Policy