Machine Learning Maps 400 Bright Slope Streaks on Mercury, Hinting at Ongoing Volatile Loss

Overview

• A peer-reviewed study in Communications Earth & Environment presents the first global inventory of roughly 400 slope lineae identified with deep learning across about 100,000 MESSENGER images.
• The mapped streaks occur preferentially on sun-facing slopes inside relatively young impact craters that cut through volcanic deposits into volatile-rich bedrock.
• Many features are linked to hollows—small bright depressions long tied to volatile loss—supporting a common volatile-driven origin for both.
• Researchers propose that impact-fracture networks expose volatile-rich layers, with solar heating driving present-day escape that generates or refreshes the bright streaks.
• The inventory will serve as a reference for ESA/JAXA’s BepiColombo mission, expected to arrive in November 2026, to assess whether new streaks have formed or existing ones have changed and to refine estimates of Mercury’s volatile budget.