Item: MAPPING SURFACE HOAR FROM NEAR-INFRARED REFLECTANCE TEXTURE IN A COLD LABORATORY ENVIRONMENT
-
-
Title: MAPPING SURFACE HOAR FROM NEAR-INFRARED REFLECTANCE TEXTURE IN A COLD LABORATORY ENVIRONMENT
Proceedings: International Snow Science Workshop Proceedings 2023, Bend, Oregon
Authors:
- James W. Dillon [ Department of Civil Engineering, Montana State University, Bozeman, MT, USA ]
- Evan N. Schehrer [ Department of Civil Engineering, Montana State University, Bozeman, MT, USA ]
- Chris P. Donahue [ Department of Geography, Earth, and Environmental Sciences, University of Northern British Columbia, Prince George, British Columbia, Canada ]
- Karl W. Birkeland [ USDA Forest Service, National Avalanche Center, Bozeman, MT, USA ]
- Kevin D. Hammonds [ Department of Civil Engineering, Montana State University, Bozeman, MT, USA ]
Date: 2023-10-08
Abstract: Surface hoar crystals are snow grains that form when water vapor deposits on the snow surface. Once buried, surface hoar is a prominent concern in the avalanche forecasting community. The formation and persistence of surface hoar are highly variable in space and time. Therefore, surface hoar detection is an ideal candidate for remote sensing. Though the near-infrared (NIR) reflectance of snow is sensitive to microstructure, previous studies have fallen short in their efforts to utilize NIR reflectance for delineation of surface hoar. We hypothesize that NIR texture, as opposed to reflected magnitude, may produce an optical signature unique to surface hoar. We tested this by performing a reflectance experiment in a controlled cold laboratory environment to evaluate the accuracy of surface hoar mapping from NIR texture using a near-infrared hyperspectral imager (NIR-HSI). We analyzed thirty snow samples with widely varying microstructure and found that surface hoar exhibited greater median values of a texture metric than other samples. Leveraging this finding, we created a simple binary classification algorithm to map the extent of surface hoar on a pixelwise basis. The algorithm proved sufficient, particularly at the central spatial resolution of 5 mm, with a median accuracy of 96.5%. Last, we underwent a repeatability test with resulting accuracy upwards of 99%. As NIR-HSI detectors become increasingly available, our findings may play a key role in remotely assessing the spatiotemporal variability of surface hoar, which could provide a critically important tool for avalanche forecasters.
Object ID: ISSW2023_P3.34.pdf
Language of Article: English
Presenter(s): James W. Dillon
Keywords: surface hoar, hyperspectral imagery, near-infrared, mapping, texture, remote sensing
Page Number(s): 1437 - 1444
-