Novel Approach to Spatiotemporal Analysis of Snow Cover Pattern using ERA 5 Dataset Over Finland

Variations in Arctic snow cover, including Finland, can impact the ecosystem, hydrological cycle, biodiversity, and many other physical processes. Getting a consistent picture of long-term changes in relevant snow cover pattern (SCP), including phenology, duration of snow cover and snow-free days, is crucial for understanding the regional dynamics of the water resources. Prevalent SCP assessments excluded critical features such as the first and last days with maximum snow cover, which are essential for a thorough spatiotemporal analysis. To address these gaps, this study utilized a novel convolution-based method coupled with K-means clustering to analyze SCP features using ERA5-Land data spanning from 2000 to 2020 across Finland. This approach was employed to cluster the country into four distinct regions based on SCP, enhancing our understanding of spatiotemporal variability and dynamics.  The largest cluster spanned 114,738 km² with maximum snow cover duration (D_max) lasted 189 days of 220 snow-covered duration (D_total). Conversely, the smallest cluster in southern and coastal areas covered 41,630 km², experiencing D_max of 85 out of 123 days of D_total. Using K-nearest neighbours method and based on the mentioned four clusters, the 20 annual SCP features images of Finland were classified. The effect of air temperature and precipitation in the classification’s results were also investigated. To assess the accuracy of annual classification, and to analyze snow cover dynamics in relation to air temperature and precipitation, three indices were obtained to measure anomalies occurred during snow accumulation period, the period with maximum snow cover, and snowmelt period.