Assessing snow mixed pixels dynamics to better understand snow spatiotemporal variability in Mediterranean mountain catchments

The headwaters catchments of the Sierra Nevada mountain range in Southern Spain are a clear example of Mediterranean mountain catchments where climate variability enhances the spatiotemporal complexity of snow dynamics. The changeable patterns of snowfall combined with the usually mild and sunny winters result in shallow snowpacks that favor various accumulation and melting cycles and, consequently, the appearance of a characteristic snow patchy distribution. Remote sensing techniques has proven to be the most effective solution to monitor this characteristic snow distribution. Among the different satellite constellations, Landsat still provides the most extended time series with an adequate spatial resolution for capturing the long-term snow spatial variability over these areas. Applying a spectral mixture analysis to the long-term Landsat dataset over the area has allowed us to not only improve the spatial representation of snow that binary classification gave but also to define and idetenfigy the presence of pixels that are not fully covered by snow: mixed pixels. 

This work proposes using these mixed pixels as an indicator of snow cover occurrence and persistence and linking its frequency and evolution with snow dynamics, from snowfall to snow ablation patterns. Twenty years of Landsat imagery has been analyzed over an area composed of the five main headwaters in the Sierra Nevada mountain range. A spectral mixture analysis, considering the three main land cover over the region: snow, shallow vegetation, and rocks, was performed to define the land cover partitioning in each pixel in the area. The distributed snow-mixed pixels' spatiotemporal persistence and evolution over the region were statistically analyzed. 

The analysis of the occurrence of these pixels shows that their presence can reach up to 40% of the mountain range during some specific years, such as wet and cold years. The clustering of mixed pixels has also allowed us to identify common areas where patchy conditions prevail. A clear differential pattern has been observed between catchments in the southern face, which is highly influenced by the presence of the sea, and in the southern face, which has a more continental climate. Finally, analyzing the temporal evolution of these pixels has allowed for the spatial assessment of areas where snowfalls can be significant and/or frequent. Still, persistence is not enhanced by the local conditions. In general, this work highlights that accounting for subgrid variability is key in this area for understanding snow spatiotemporal patterns, determining the more vulnerable regions facing potential changes in the snow regime due to global warming and climate variability, and further assessing water resources planning through the improvement of hydrological models predictions.

Acknowledgment: This research was funded by the Spanish Ministry of Science and Innovation through the research project PID2021-12323SNB-I00, HYPOMED—“Incorporating hydrological uncertainty and risk analysis to the operation of hydropower facilities in Mediterranean mountain watersheds.”