Predicting Glacier Ice Melt with Machine Learning to Address Climate Change

Subhash Kumar Choudhary, Shivani Kumari, Muzafar Mehraj Misgar

The accelerating decline in glacier mass due to climate
change presents a significant threat to global water
resources, sea levels, and ecosystem stability. This
research integrates machine learning techniques to
predict glacier ice melt patterns using historical mass
balance data. Leveraging the publicly available global
glacier mass balance dataset, the study investigates
temporal trends and employs regression models—
including Polynomial Regression Linear Regression,
Decision Tree, Random Forest, and Support Vector
Machine (SVM)—to forecast future glacier behavior.
Exploratory Data Analysis (EDA) reveals a strong
negative correlation (−0.96) between year and
cumulative mass balance, highlighting accelerated ice
loss over recent decades. Among the models, Random
Forest achieved the highest predictive accuracy (R² =
99.71%).(Working with a Two-Stage Ice Sheet
Model, n.d.)
followed by Decision Tree (R² = 99.57%), indicating
their robustness in capturing nonlinear glacier dynamics.
This machine learning framework serves as an effective
tool for evaluating glacier degradation under varying
emission scenarios and contributes valuable insights for
environmental policy, climate impact assessment, and
adaptation strategies.