The melting of the permafrost is one of the biggest yet least-talked-about consequences of global warming. The permafrost is a layer of frozen soil, located in areas near the arctic, that stays frozen for several years. There are many concerns with permafrost melting, but one of the biggest issues is related to its effect on the climate. There is an estimated three times more organic (and digestible) material stored in the permafrost than there is in all the forests left on the planet. As permafrost melts, more materials in the permafrost are decomposed. This leads to more carbon dioxide being released into the atmosphere, warming up the globe. The increase in temperature then speeds up the thawing process, thus creating a positive feedback loop.
This project’s goal is to bring more awareness to the melting of the permafrost by creating an interactive UI where users can input dates and see the forecasted changes in active layer thickness.
For the project, the team started by searching for available data on permafrost from open-source websites. Due to the incompleteness of many datasets, the team settled on a dataset from Samoylov, Russia that only contains the dates of measurement and the active layer thickness in centimeters. The models used for the project include time series models, random forest regression, and XGBoost. Out of all these models, XGBoost performed the best and is now the model deployed on the team’s website.
Overall, the team learned a lot about modeling, data analysis, and how to create a website. In the future, they hope to find better ways to improve their current model as well as further analyze how other factors other than time could affect the permafrost thawing rate.
Project by: Jesica Sutandi, Venkata Durga Sai Leela Arun Putcha, Luanne Yang