Tracking Climate Change With Satellites and Artificial Intelligence (2022-2023)

Background

Climate change is beginning to impact infrastructure, transportation, energy, food and water supplies, and human health across the globe, with Africa and Asia being two of the most vulnerable regions to the impacts of global warming. However, these impacts are often difficult to quantify, especially in parts of the Global South where ground-based economic surveys occur infrequently. In regions with incomplete information and knowledge gaps in the strategies needed to adapt to the impacts of climate change, some of the most vulnerable cities incur increased risk of disastrous impacts. 

Informing climate change mitigation and adaptation strategies requires a wide-ranging quantity of data. This requires information on energy infrastructure and access, population, income, demand growth trends, agriculture and land use as well as on vulnerable infrastructure such as transmission lines, water systems and food supply chains. To make the best possible plans and track policy progress and impacts, these data need to be regularly monitored.

Project Description

This project team will work to democratize access to climate change data and the strategies to acquire those data. Team members will develop and publicly release the first dataset to create a geospatial foundation model for enabling near real-time tracking of climate causes and impacts. This model will enhance climate change mitigation/adaptation monitoring and planning through developing robust features that can be used to monitor a broad range of contributing factors (e.g., energy infrastructure and use, agricultural activity) and impacts (e.g., economic impacts and human migration). This advancement will provide evidence for decision-makers to inform climate mitigation and adaptation strategies and hasten those strategies’ evaluation and implementation.

The team will work toward building the first foundation model specifically designed for remote sensing imagery for the purpose of extracting climate change relevant content at scale. A foundation model is 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 inference tasks.

Team members will train the models using self-supervised learning techniques to investigate the efficacy of this approach on monitoring a wide range of variables relevant to the causes and consequences of climate change. This foundation model will generate climate change evidence and data by using remote sensing, with less reliance on expensive ground-based economic surveys, and enable a wide range of custom queries regarding climate data needs for anywhere on Earth through the integration of a natural language processing encoder.

Anticipated Outputs

Training dataset of millions of scenes with extensive documentation; initial trained foundation model; paper for conference or journal submission; website describing project

Student Opportunities

Ideally, this project team will be comprised of 3 graduate students and 6 undergraduate students. Students can come from a variety of disciplines including engineering, computer science, economics, public policy, environmental science, mathematics and statistics. Students with some experience in quantitative methods are preferred. Experience with programming is required, and experience with Python is beneficial. Graduate students with excellent management skills and/or strong technical skills are preferred.

Participants will learn the fundamentals of machine learning and deep learning for computer vision, experimental research techniques as well as advanced data analytical tools such as Python. Students will gain skills in problem definition, literature review, research project design, execution, analysis and interpretation. They will also improve verbal and written communication skills, engage in team-based problem solving and take ownership of project management and organization. 

Participating graduate students will serve as project managers.

In Fall 2022, the team will meet on Mondays and Wednesdays from 12:00-1:15 p.m.

See the related Data+ project for Summer 2022; there is a separate application process for students who are interested in this optional component.

Timing

Fall 2022 – Spring 2023

  • Fall 2022: Take part in intensive short course on data science and machine learning, climate change, energy systems and research techniques; focus on goal and role development; begin research and training
  • Spring 2023: Continue executing project; conclude research; create website and final presentation; draft paper for submission to a relevant conference or journal

Crediting

Academic credit available for fall and spring semesters

See related Data+ project, Tracking Climate Change Causes and Impacts With Satellites and AI (2022), and earlier related team, Creating Artificial Worlds with AI to Improve Energy Access Data (2021-2022).

 

Image: Central Africa Appears to Be Completely On Fire, by NASA Goddard Space Flight Center, licensed under CC BY 2.0

Satellite view of Central African Republic.

Team Leaders

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

/yfaculty/staff Team Members

  • Leslie Collins, Pratt School of Engineering-Electrical & Computer Engineering
  • T. Robert Fetter, Nicholas Institute for Environmental Policy Solutions
  • 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