Determining snow material properties from near-infrared photography

It is well understood that snow is a complex, porous material, whose microstructural changes directly affect its physical properties. Therefore, – to gauge the snow's role within the climate system – it is of interest to accurately measure and characterize the spatio-temporal variability of snow surfaces and snowpacks.

On a local scale, for example inside a snowpit during a field campaign, snow measurements are often taken in a manual, point-like fashion resulting in single, one-dimensional profiles with a sampling resolution of a few centimeters. At this resolution thin layers are difficult to observe and spatial inhomogeneities of the snowpack are missed. State-of-the-art X-ray microtomography (μ‑CT) scans of snow provide excellent spatial resolution,¹ however, the added experimental constraints prevent sampling extended spatio-temporal domains.

To address some of these limitations, we propose to use near-infrared (NIR) photography² with 940 nm illumination to determine the snow's specific surface area (SSA) and density. Our device – called SnowImager – achieves millimeter resolution and covers a spatial extent of a few square meters, such as the surface area of a snowpit wall. While the SSA is determined directly from the measured NIR image using the well-established asymptotic radiation transfer theory,^3–6 the density dependence is introduced by physically truncating the illuminating and back-scattered light. It results non-trivially from the lateral component of the sub-surface scattering process and enables us to recover density profiles that compare well to reference data from density cutter and μ‑CT measurements. As a demonstration, we present the spatial variability of an Antarctic snowpack at an unprecedented level of detail, revealing an extremely high spatial variability of the snow microstructure.

Using near-infrared photography enables accurate and fast determination of snow material properties, whenever millimeter spatial resolution and a spatial extent of several square meters are required. It is thus ideally suited to simultaneously capture thin layers within the snowpack and spatial inhomogeneities over a centimeter to meter scale, which is relevant as ground truth measurement for climate research, remote sensing and avalanche forecasting among others.

 

1. Kerbrat, M. et al., Atmos. Chem. Phys. 8, 1261–1275 (2008).

2. Matzl, M. & Schneebeli, M., J. Glaciol. 52, 558–564 (2006).

3. Bohren, C. F. & Barkstrom, B. R., J. Geophys. Res. 79, 4527–4535 (1974).

4. Warren, S. G., Rev. Geophys. 20, 67–89 (1982).

5. Kokhanovsky, A. A. & Zege, E. P., Appl. Opt. 43, 1589–1602 (2004).

6. Libois, Q. et al., The Cryosphere 7, 1803–1818 (2013).