Disturbance detection in snow using polarimetric imagery of the visible spectrum

Abstract

Optical polarimetric data were analyzed to determine their utility for detecting disturbances in snow. Research for this thesis was conducted in March of 2010 at Lake Tahoe in various settings and snow depths. Images of footprints, snowmobile tracks, and other disturbances were captured by Bossa Nova's linear stokes polarization camera named SALSA. This device implemented a fast switching liquid crystal polarizing filter to separate polarized light onto a 782 x 582 pixel detector operating in the 400 to 700 nanometer range. The data were then analyzed for polarimetric signatures by isolating the disturbances from the background and then comparing standard deviations of intensity and polarization occurrences. Additionally, texture filters were applied to determine if the disturbances could be enhanced and thus highlighted from the background. The results of the study showed that intensity was a stronger discriminant for disturbances in snow than polarization in the visible spectrum. This result was most likely due to the Umov Effect where bright objects typically have low polarization signatures. This conclusion discounts the significant polarization observed in shadowed regions due to polarized skyshine.