Evaluating multiple canopy-snow unloading parameterizations in SUMMA with time-lapse photography characterized by citizen scientists

Abstract

Snow in the canopy can sublimate back into the atmosphere, or unload to the surface and contribute to streamflow. Snow unloading results in a drastic, and sometimes sudden, decrease in the land surface albedo. Snow unloading is a complex physical process that is difficult to parameterize due to limited observations. Time-lapse photos of snow in the canopy were characterized by citizen scientists to create a dataset of snow interception observations at multiple locations across the western United States. This novel interception dataset was used to evaluate three snow unloading parameterizations in the Structure for Unifying Multiple Modeling Alternatives (SUMMA) modular modeling framework. SUMMA was modified to include a third snow unloading parameterization (Roesch et al., 2001), which includes temperature and wind dependent unloading functions. This parameterization is compared to a meltwater drip unloading parameterization (Andreadis et al., 2009) and a time-dependent exponential unloading parameterization (Hedstrom & Pomeroy, 1998). The parameterizations were calibrated at Niwot Ridge, CO, using hourly interception observations, and simulations with these calibrated parameters were compared to those using default (i.e., literature recommended) parameterizations at sites located at Niwot Ridge, CO, Grand Mesa, CO, and the maritime Olympic Mountains, WA. The Roesch et al., (2001) parameterization with temperature and wind unloading performed best at Niwot Ridge, where it captured 87% of the observed snow interception events. When parameterizations with wind dependent and exponential unloading were transferred to Grand Mesa and the Olympic Mountains, they unloaded snow from the canopy too rapidly. This suggested that parameterizations would need to be recalibrated for the unique interception physics of those domains. As a result, Andreadis et al., (2009), which retained snow in the canopy longest, captured the duration of snow in the canopy best when transferred to these additional sites. Unloading schemes also impacted the total sublimation modeled at each site. At Niwot Ridge, cumulative sublimation from the canopy was 87.6 mm using Andreadis et al., (2009) parameterization, which is 6% of the total winter precipitation, compared to Roesch et al., (2001), which sublimated 1% of the total precipitation. While canopy-snow unloading parameterizations are often overlooked in land surface models, results show that the unloading scheme can have an effect on the duration of snow in the canopy, impacting the land-surface albedo, and whether canopy-snow contributes to streamflow or sublimates.