Changes in snow cover climatology in Romania (1961-2020)

Abstract The response of the cryosphere to a warmer climate is spatially diversified and requires accurate monitoring and understanding. The study analyses the changes in snow cover phenology (the first and last snow cover days - FSC and LSC), duration (SCD, SCDmax) and snow-free days (SFD) in Romania, which includes wide parts of the Carpathian Mountain range (the Eastern, the Southern and Southwestern Carpathians), using daily snow depth observations from 114 weather stations (WS), with long-term and gap-free time series, over the 1961–2020 period. The results are discussed over five elevation bands (< 500, 501-1,000, 1,001–1,500, 1,501-2,000, and > 2,000 m) and over the major Köppen-Geiger climate regions of the country. A delay in FSC and a retreat in LSC was systematically observed throughout the country, with average rates from 1.6 to 2.2 days decade− 1. The observed trends in snow cover phenology are prominent at mid-elevations (500-1,500 m) and in the lowlands (especially below 500 m), feedbacking the intense warming process. Consequently, declines in SCD and SCDmax have been also observed country-wide, especially at mid-elevations (500-1,000 and 1,000–1,500 m) and lowlands below 500 m, with an average rate exceeding 2.0 days decade− 1. Furthermore, the number of snow-free days have a significant growth in most areas, with an important contribution of significant trends (44% of WS). In mountain areas the SFD positive trends are weak and not statistically significant. The elevation dependency of the detected snow cover trends was not systematically observed throughout the major climate regions of the country. Accelerated snow cover changes with elevation were found specific only to the Dfc and Dfb climate regions, within narrow elevation bands (1,000–1,500 m). The high-elevation areas show weak changes in snow cover phenology and duration, mostly not statistically significant. The long-term variability in snow cover parameters shows breakpoints grouped during the 1980s, and mostly in the 1990s period. The analysis of the possible links with the large-scale atmospheric circulation (North Atlantic Oscillation - NAO) revealed that NAO has a significant negative correlation with LSC, SCD, and SCDmax for 30% of stations, mostly located in the north-eastern lowlands of Romania. NAO showed no statistically significant influence on FSC trends.