The expected effect of climate change on snowfall amounts and snow depth in the major urban areas of Romania

Snowfall and snow depth are important elements of the climate system which can have a significant impact on the transport sector, local economy, water resources and local thermal regime.

This study aims to identify the future trends of snowfall and snow depth in the most important urban areas in Romania, based on high-resolution regional climate models (RCM) data, made available through the EURO-CORDEX, and bias-corrected RCM simulations available în the RoCliB dataset. Ten different regional climate models with a target resolution of 10 km and two emission scenarios were considered, namely the moderate (RCP4.5) and business-as-usual (RCP8.5) scenarios. The study covers the interval from 2021 to 2100.

In order to predict the future snow depth, snow cover duration and snowfall amounts based on the available parameters from the RCM simulations, it was necessary to identify the complex relationship between snowfall, snow melting, temperature and precipitation. In order to do this, we first extracted the ERA5 reanalysis data from 1981 to 2020 for each urban area, and then employed a Bayesian Regularized Neural Network (BRNN). The resulting model was used to predict the future variables for all major urban areas in Romania.

A general trend of decreasing snowfall amounts, mean snow depth and snow cover duration was observed for all analyzed areas and for both emission scenarios. Important regional variations were also observed, with some areas no significant change for the 2021 – 2050 interval compared to the observation period, which could be explained mostly by the increasing winter precipitation predicted by the RCM simulation. The results also showed an increased possibility of some years virtually lacking any snow cover and snowfall precipitation, especially after 2050 in the business-as-usual scenario (RCP8.5). However, an increased variability was also observed, with extreme snowfall events remaining possible even in the latter half of the study interval.