The recent decades decline of the multi-year sea ice pack show that the emerging Arctic may be beyond the point of recovery by only reducing the GHG emissions and urges for additional mitigation measures. Arctic Ice Project (AIP) has proposed a technology that increases the brightness of the sea ice enhancing the sea ice albedo and thus reducing the sea ice melt. Climate modeling by Climformatics, Inc. has shown promising results for this technology to slow down the accelerated rates of Arctic ice melt due to Global warming. The primary goal of this project is to develop ML models of the relationship of the Arctic surface albedo and the sea ice pack characteristics such as sea ice thickness and area, using results from climate model simulations and observations. We will also investigate the sensitivity of these models to the Arctic cloud cover at different vertical levels (low, mid- and high-clouds). This will help to determine where and when AIP surface albedo enhancement will have the largest impact for the Arctic ice pack restoration. The students will use advanced methods of machine learning and data science to analyze climate model output data sets and observational data sets (satellite or in-situ) for albedo and cloud fraction.
PROJECT TIMELINE: This research project is continuation from previous semesters interns work. We have developed ML models tested in single location. Currently, we are looking for scaling up the models to the entire Arctic Basin (week 2-3) which will involve python coding and preferably parallelizing the codes. Next we will visualize, interpret and analyze the results (week 4-5), making any adjustments if needed to refine the ML models. Part of the work will be to make literature search and overview of the most recent publications on the topic (week 1). The goal is to finalize and submit a paper draft by the end of the semester (week 6-10). The last two weeks (9&10) we will also prepare a final presentation for the DS Discovery program.
PROJECT WORKFLOW: We already have two team members on the projects continuing the work from previous semesters and we are looking for one or two new members to help with the workload. We will have regular weekly team meetings and meetings per request. We also encourage you to meet among yourself as a team.
DELIVERABLES: The main deliverable of the project are: ML models representing the relationship of the Arctic sea ice cover and the surface albedo and with predictive capability to show how the ice cover will change when we enhance the surface albedo artificially. We also expect to finalize and submit the manuscript which we are working on.
PREFERRED APPLICANT SKILLSET:
- Python - Beginner/Intermediate/Advanced
- EDA - Beginner/Intermediate/Advanced
- Data Visualization - Beginner/Intermediate/Advanced
- Geospatial Data Analysis - Beginner/Intermediate/Advanced
- Machine/Deep Learning - Beginner/Intermediate/Advanced
- Cloud Computing - Beginner/Intermediate/Advanced
PREFERRED ADDITIONAL SKILLSET: Linux/Shell. Parallel supercomputing
ABOUT THE DISCOVERY PROGRAM:
The Discovery Research Program aims to empower undergraduate researchers by connecting them to data science-driven research projects. Discovery encourages applicants from all majors and backgrounds. The majority of our available projects leverage tools and methodologies in Data Science for research inquiry in a broad range of topic areas.
Depending on the project, students can expect to spend anywhere between 6-15 hours per week. While backgrounds in STEM are all useful, we believe that diverse perspectives and experiences are the most essential qualities for successful research.
For additional information about the Discovery Research Program, visit:
