Inter-comparison of medium-resolution satellite albedo retrieval techniques over snow and ice

Albedo plays a crucial role in regulating the amount of solar radiation absorbed by snow and ice surfaces, thereby influencing the production of meltwater and the mass balance of glaciers. Landsat and Sentinel-2 surface reflectance data products can be used to estimate albedo at medium resolution (30 m) on a near-daily to weekly basis. This research compared four medium-resolution methods for retrieving snow and ice albedo. It assessed their performance using ground-truth observations from 10 automatic weather stations on five glaciers around the world, along with linear mixed-effects models that included the retrieval method and surface type as fixed effects, and the site locations as random effects. The results indicated that the mean differences between satellite-derived and observed albedo depended on the surface type. The direct-estimation method using four spectral bands estimated snow albedo with the smallest difference with in situ observations of 0.03 ± 0.06. Over ice, the method utilizing two spectral bands and an empirical model of the bidirectional reflectance distribution function (BRDF) produced the smallest difference between retrieved and observed albedo (0.05 ± 0.08). The empirical BRDF method using five spectral bands, along with the six-band method incorporating a semi-empirical kernel-based BRDF model, also showed strong performance on both surfaces, with differences of 0.06 ± 0.06 on snow and 0.06 ± 0.08 on ice.