The region of Earth’s atmosphere between 80 and 100 kilometers altitude—called the mesosphere and lower thermosphere (MLT)—is a key transition layer where energy and momentum from lower atmospheric weather and upper atmospheric space weather interact. One important feature of this region is the presence of ripple-like gravity waves: small, wave-shaped disturbances that can tell scientists how energy is moving through the atmosphere. These waves are often seen in nighttime images taken by special cameras that look at glowing layers of air high above Earth’s surface.

Detecting these ripples is difficult because they are small, short-lived, and often hidden in noisy or cloudy data. In this project, we use artificial intelligence—specifically a type of machine learning model called a convolutional neural network (CNN)—to automatically find and classify ripple patterns in airglow imager data. This system can quickly and accurately identify ripple events across large datasets, saving time and improving consistency compared to manual inspection.

By tracking where and when these ripples occur, we can better understand how gravity waves behave in the upper atmosphere and how they affect larger-scale weather and space weather patterns. This helps improve our understanding of atmospheric dynamics and supports future forecasting models.