Dynamic identification of snow phenology in the Northern Hemisphere

Abstract

Snow phenology characterizes the cyclical changes in snow and has become an important indicator of climate change in recent decades. Changes in snow phenology can significantly impact climate and hydrological conditions. Previous studies commonly employed fixed-threshold methods to extract snow phenology, which cannot represent the differences in the beginning and/or end of the snow period under different snow conditions in the Northern Hemisphere, leading to potential uncertainties in terms of snow phenology. In this study, we observe that snow phenology extracted from different snow data and methods shows significant differences but consistently underestimates snow duration at low and middle latitudes. After normalizing, the percentage snow depth curve turns significantly at the 10 % position, marking the transition between the snow and non-snow seasons. Therefore, we propose a dynamic snow phenology method with a 10 % threshold. Using the dynamic-threshold method, there is an earlier snow cover onset day (SCOD), a later snow cover end day (SCED), and a longer snow cover duration (SCD) at low and middle latitudes, especially on the Tibetan Plateau, where the SCD differences can reach 28 d. The differences in terms of snow phenology at higher latitudes are reversed. The dynamic snow phenology accounts for the spatial heterogeneity of Northern Hemisphere snow cover and excludes the influence of inter-annual variability of snow cover on snow phenology extraction, providing a novel perspective for identifying and understanding snow cover variations in the Northern Hemisphere.