Field evaluation of a new method for estimation of liquid water content and snow water equivalent of wet snowpacks with GPR

Abstract

Estimates of snow water equivalent (SWE) with ground-penetrating radar can be used to calibrate and validate measurements of SWE over large areas conducted from satellites and aircrafts.However, such radar estimates typically suffer from low accuracy in wet snowpacks due to a built-in assumption of dry snow. To remedy the problem, we suggest determining liquid water content from path-dependent attenuation. We present the results of a field evaluation of this method which demonstrate that, in a wet snowpack between 0.9 and 3 m deep and with about 5 vol% of liquid water, liquid water content is underestimated by about 50% (on average). Nevertheless, the method decreases the mean error in SWE estimates to 16% compared to 34% when the presence of liquid water in snow is ignored and 31% when SWE is determined directly from two-way travel time and calibrated for manually measured snow density.

Measurement of the physical properties of the snowpack

“…The attenuation of electromagnetic radiation in snow was recently investigated by Sundström et al [] using Ground‐Penetrating Radar (GPR) and a path‐dependent attenuation function. Sundström et al [] found that liquid water content was underestimated by ∼50%, but the mean error in determining SWE was reduced from 34% for a dry snowpack to 16% when the snowpack was wet.…”

“…An upward looking FMCW radar system can be used to determine liquid water content by measuring the attenuation of radar waves in snow and using an empirical relationship to relate attenuation to water content [Mitterer et al, 2011]. The attenuation of electromagnetic radiation in snow was recently investigated by Sundström et al [2013] using Ground-Penetrating Radar (GPR) and a path-dependent attenuation function. Sundström et al [2013] found that liquid water content was underestimated by ∼50%, Reviews of Geophysics 10.1002/2015RG000481 but the mean error in determining SWE was reduced from 34% for a dry snowpack to 16% when the snowpack was wet.…”

“…The attenuation of electromagnetic radiation in snow was recently investigated by Sundström et al [2013] using Ground-Penetrating Radar (GPR) and a path-dependent attenuation function. Sundström et al [2013] found that liquid water content was underestimated by ∼50%, Reviews of Geophysics 10.1002/2015RG000481 but the mean error in determining SWE was reduced from 34% for a dry snowpack to 16% when the snowpack was wet. Williams and Knoll [2002] used tomographic imaging of radar waves sent between two antennas to determine the dielectric constant of snow and measure snow wetness.…”

Multiscale geophysical imaging of the critical zone

“…GPR is effective for traversing large areas to estimate snow depth [Marchand and Killingtveit, 2001] and SWE for dry snowpack [Jaedicke, 2003;Lundberg et al, 2006] using snow density relationships [Sand and Bruland, 1998]. More recently, techniques have been pioneered to extract SWE information about snowpacks with liquid phase water distributed among the solid phase ice/water matrix [Bradford et al, 2009;Sundström et al, 2013]. Such geophysical investigations have resulted in better understanding of the relationships between snow depth and slope aspect and snow depth and elevation [Marchand and Killingtveit, 2001].…”

Estimating snow water equivalent in a Sierra Nevada watershed via spaceborne radiance data assimilation

“…Remote sensing offers spatially continuous observation of seasonal snowpack properties from local to global scale [ Nolin , ]. Efforts to remotely sense SWE have utilized passive microwave (PM) radiometry [ Chang et al ., ; Stankov et al ., ; Hardy et al ., ; Vuyovich et al ., ], active microwave radar scatterometry [ Cline et al ., ; Yueh et al ., ; Sundström et al ., ], airborne gamma counters [ Peck et al ., ; Carroll and Carroll , ; Choquette et al ., ], satellite gravity recovery [ Frappart et al ., ; Niu et al ., ; Forman et al ., ], reconstruction or reanalysis using visible and near‐infrared measurements [ Cline et al ., ; Molotch and Margulis , ; Molotch , ; Slater et al ., ; Guan et al ., ; Girotto et al ., , ; Margulis et al ., ], airborne LiDAR altimetry [ Deems et al ., ; Harpold et al ., ; Mattmann et al ., ; Painter et al ., ], synthetic aperture radar interferometry [ Deeb et al ., ], and photogrammetric reconstruction [ Vander Jagt et al ., ; Nolan et al ., ]. However, as noted recently by Lettenmaier et al .…”