Multi-decadal past winter temperature, precipitation and snow cover information over the European Alps using multiple datasets

Abstract

Abstract. Assessing past distributions, variability and trends of the mountain snow cover and its first order drivers, temperature and precipitation, is key for a wide range of studies and applications. In this study, we compare the results of various modelling systems (global and regional reanalyses ERA5, ERA5-Land, ERA5-Crocus, CERRA-Land, UERRA MESCAN-SURFEX, MTMSI, and regional climate model simulations CNRM-ALADIN and CNRM-AROME driven by the global reanalysis ERA-Interim) against observational references (in-situ, kriged datasets and satellite observations) across the European Alps, from 1950 to 2020. The comparisons are performed in terms of monthly and seasonal snow cover variables (snow depth and snow cover duration) and their main atmospherical drivers (near-surface temperature and precipitation). We assess multi-annual averages of regional and sub-regional mean values, their inter-annual variations, and trends over various time scales. CNRM-AROME and CNRM-ALADIN simulations, and ERA5-Land exhibit an overestimation of the snow accumulation during winter, increasing with elevations. ERA5, ERA5-Crocus, MESCAN-SURFEX, CERRA-Land and MTMSI offer a satisfying description of the monthly snow evolution albeit a spatial comparison against satellite observation indicates that all datasets overestimate the snow cover duration of the snow cover, especially the melt-out date. The analysis of the inter-annual variability and trends indicate that modelling snow cover dynamics remain complex across multiple scales, that none of the models evaluated here fully succeed to reproduce, compared to observational reference datasets. Indeed, while most of the evaluated model outputs perform well at representing the inter-annual to multi-decadal winter temperature and precipitation variability, they often fail to address the variability of the snow depth and snow cover duration. We discuss several artifacts potentially responsible for incorrect long-term climate trends in several reanalysis products (ERA5 and MESCAN-SURFEX), which we attribute primarily to the heterogeneities of the observation datasets assimilated. Reference datasets and some of the evaluated datasets provides past trends in line with current available literature. Over the last 50 years (1968–2017) at a regional scale, the European Alps have experienced a winter warming of 0.3 to 0.4 °C per decade, a weak reduction of winter precipitation, and a substantial decrease of the snowpack characteristics, with a decline of the winter snow depth and the snow cover duration reaching −10 % per decade and −10 days per decade, respectively, especially at low and intermediate elevations. Overall, we show that no modelling strategy outperforms all others within our sample, and that upstream choices (horizontal resolution, heterogeneity of the observations used for data assimilation in reanalyses, coupling between surface and atmosphere, level of complexity and configuration of the snow scheme etc.) have great consequences on the quality of the datasets and their potential use. Despite their limitations, in many cases these modeling outputs can be used to characterize the main features of the mountain snow cover for a range of applications.

Northern Hemisphere Snow Drought in Earth System Model Simulations and ERA5‐Land Data in 1980–2014

“…Although no data assimilation was used in ERA5-Land, the atmospheric forcings were directly constrained by assimilated observations (Räisänen, 2021). However, ERA5-Land has been found to show especially large overestimation of SWE in the European Alps (Monteiro & Morin, 2023). On the other hand, Lei et al (2022) found snow depth of ERA5-Land represents snow absence well and compares reasonably well with in situ observations of snow depth over the Tibetan Plateau.…”

“…Although no data assimilation was used in ERA5‐Land, the atmospheric forcings were directly constrained by assimilated observations (Räisänen, 2021). However, ERA5‐Land has been found to show especially large overestimation of SWE in the European Alps (Monteiro & Morin, 2023). On the other hand, Lei et al.…”

Above-treeline ecosystems facing drought: lessons from the European 2022 summer heatwave

“…First, land surface meteorological variables are notoriously difficult to model at high elevation because of the scarcity of available observations and the importance of processes that depend on topography and that are not accounted for by climate re-analyses (Vionnet et al, 2019). The ERA5-land products have proven useful for capturing the main features of surface variable trends in the European Alps (Monteiro and Morin, 2023), and have outcompeted other climate gridded products, such as E-OBS, in areas where the density of weather stations is low (Bandhauer et al, 2022), as is the case in the southwestern Alps. However, its coarse spatial resolution limits its usefulness for representing local heterogeneity in bioclimatic conditions on the ground.…”

Snow depth in high-resolution regional climate model simulations over southern Germany – suitable for extremes and impact-related research?

“…Furthermore, there are studies evaluating regional climate models based on multi-decadal simulations, but these are at coarser spatial resolution (Daloz et al, 2022: 0.11°;Matiu and Hanzer, 2022: 0.11°;Steger et al, 2013: 0.22°). Recently, Monteiro and Morin (2023) compared the performance of multiple model systems with spatial resolutions ranging from 2.5 to 30 km over the European Alps. They found that main features of the snow cover, snow depth, and driving climatic conditions could be reproduced by the models but with increasing deviations at higher altitude.…”

“…Climate change already affects and will further alter snow dynamics and conditions (Dong and Menzel, 2020;Monteiro and Morin, 2023), which is why observation-based analyses are limited. Climate impact research and data users benefit from information at the local scale (Orr et al, 2021).…”