The challenges of accurately modeling and measuring snow albedo: two new contributions to improve our understanding

An empirical time-based albedo decay equation, which was developed over 60 years ago, is still often used in snowmelt models. There have been a number of snow albedo models developed since, but they show remarkable spread in results, which is substantial considering that a change of a few percent in the predicted snow albedo can lead to large differences in the amount of snow melted over a season. Albedo sensitivity is especially high in snow water equivalent (SWE) reconstruction models, where the snowpack is built up in reverse from melt out to peak SWE. Validation remains a formidable problem, as accurate snow albedo measurements, especially over long time series are difficult and time intensive to prepare. Factors such as imperfect cosine response of radiometers, sloped snow surface, and non-snow objects in the downlooking radiometer's field of view must be accounted for.

Here we present two new contributions to improve our understanding of snow albedo: 1) a new snow albedo model using a two-stream radiative transfer solution where grain size and impurity content are solved simultaneously using nonlinear optimization; 2) seven years of hourly snow albedo measurements from a sub-alpine site in Mammoth Lakes, CA. Using smoothed grain size and impurity estimates from MODIS, we then examine the effects of different snow albedo models on our reconstructed SWE estimates in the upper Tuolumne Basin, CA using Airborne Snow Observatory measurements for validation.

Our snow albedo model shows improvement in model error during the snowmelt season, which in turns causes reduced error in basin-wide SWE estimates. Our modeled broadband direct snow albedos are consistently 5% lower than another leading model, however both models appear to be low biased during the accumulation season and high biased at the end of the melt season. There are numerous other errors, including those caused by the satellite retrievals themselves and how they are smoothed.

For further model validation, as they become available, we plan to compare independent surface measurements of specific surface area and reflectance taken during the SnowEx campaign, and to compare those with our snow albedo model and previous models.