Characterizing False Positives in CRISM Frost Detections

The martian frost cycle is a key driver of both climate and geomorphological features on the surface of Mars. To better understand the martian frost cycle, we are interested in building a global, monthly map of frost using automated detections from visible, thermal, and multispectral observations. The CRISM (Compact Reconnaissance Imaging Spectrometer for Mars) instrument, which collects data in infrared (IR) and visible/near infrared (VNIR) wavelengths remotely from orbit, can be used for automated detection and differentiation of different frost types. However, VNIR wavelengths, analyzed with the ICV parameter map, typically produce less robust detections of frost compared with parameter maps based on VNIR and IRwavelengths, with the potential for false positive detections. Our analysis aims to characterize terrains that can lead to false positive frost detections. These results inform potential improvements to the ICV parameter map and its usage for automated frost detection for global mapping.