Glaciers are vital indicators of climate change and play a crucial role in regulating global freshwater resources and sea-level rise. Traditional approaches to studying glacier melt rely on remote sensing and physical measurements which often lack integration with local climate variables and predictive modeling these methods may also fall short in identifying short term climate anomalies or classifying individual glacier risk accurately. This paper investigates glacier melt dynamics by integrating temperature trend analysis with machine learning based risk prediction. The study analyzes long term monthly temperature data from the National Snow and Ice Data Center (NSIDC) focusing on the impacts of climate change particularly during the COVID 19 period. Notable shifts in seasonal temperature patterns were observed, with cooler summers and warmer winters during COVID years indicating reduced anthropogenic activity. Additionally, a glacier dataset “World Glacier Inventory” containing variables such as latitude, longitude, elevation, and morphological features is used. Multiple machine learning models viz. Random Forest, Logistic Regression, and Support Vector Machine (SVM) are evaluated to classify glaciers into high and low risk categories. Random Forest outperformed the Logistics Regression and SVM with an accuracy of 99.99%; key risk factors included glacier area, elevation, and depth. The results underscore the influence of both climatic variability and glacier-specific attributes on melt behavior and provide a data-driven approach to identifying glaciers at higher risk of outburst or rapid melting.