Comparison of Long-Term Albedo Products against Spatially Representative Stations over Snow

Multiple satellite products are available to monitor the spatiotemporal dynamics of surface albedo. They are extensively assessed over snow-free surfaces but less over snow. However, snow albedo is critical for climate monitoring applications, so a better understating of the accuracy of these products over snow is needed. This work analyzes long-term (+20 years) products (MCD43C3 v6/v6.1, GLASS-AVHRR, C3S v1/v2) by comparing them against the 11 most spatially representative stations from FLUXNET and BSRN during the snow-free and snow-covered season. Our goal is to understand how the performance of these products is affected by different snow cover conditions to use this information in an upcoming product inter-comparison that extends the analysis spatially and temporally. MCD43C3 has the smallest bias during the snow season (−0.017), and more importantly, the most stable bias with different snow cover conditions. Both v6 and v6.1 have similar performance, with v6.1 just increasing slightly the coverage at high latitudes. On the contrary, the quality of both GLASS-AVHRR and C3S-v1/v2 albedo decreases over snow. Particularly, the bias of both products varies strongly with the snow cover conditions, underestimating albedo over snow and overestimating snow-free albedo. GLASS bias strongly increases during the melting season, which is most likely due to an artificially extended snow season. C3S-v2 has the largest negative bias overall over snow during both the AVHRR (−0.141) and SPOT/VGT (−0.134) period. In addition, despite the improvements from v1 to v2, C3S-v2 still is not consistent enough during the transition from AVHRR to SPOT/VGT.