Citizen science trains a sharper eye on rare clouds

A volunteer working with NASA-supported citizen science has developed a machine-learning tool to help identify rare cloud types in submitted photographs.

The work supports Space Cloud Watch, which asks people around the world to photograph cloud formations. Researchers are examining why some unusual clouds appear more frequently and at lower altitudes than before.

Space Cloud Watch focuses on noctilucent clouds, extremely high-altitude formations that shine after sunset when lower layers of the atmosphere are already dark. Their observed range and altitude have changed over time, making consistent reports from many locations useful to atmospheric researchers.

Citizen submissions also create a classification problem: ordinary lower clouds can resemble the target formations in photographs. A volunteer, Namai, worked with project scientists to build a machine-learning pipeline trained on examples of noctilucent clouds and look-alikes. The tool is intended to help sort observations, not eliminate scientific review.

The project illustrates a productive division of labour. People supply geographic reach and local observation; software helps organise a growing image set; researchers interpret patterns against physical measurements. Its quality ultimately depends on labelled training data, documented uncertainty and continued checks when the model encounters unfamiliar conditions.

About this article: This is an original Orbit Daily synthesis written from the cited material. It is not a reproduction of any publisher’s article. Claims that remain scheduled, provisional or uncertain are described that way. Report an error if a factual claim needs correction.