Potential and limitations of convection‐permitting CNRM‐AROME climate modelling in the French Alps

• Published in: International journal of climatology 2022-11, Vol.42 (14), p.7162-7185
• Main Authors: Monteiro, Diego, Caillaud, Cécile, Samacoïts, Raphaëlle, Lafaysse, Matthieu, Morin, Samuel
• Format: Article
• Language: English
• Subjects: AROME; Barriers; Boundary conditions; Climate change; Climate change research; Climate models; Climatic conditions; Climatology; Convection; CP‐RCM; Earth Sciences; Elevation; Future climates; Global climate; Global climate models; Low temperature; Modelling; mountain; Mountain regions; Numerical simulations; Precipitation; Regional climate models; Regional climates; Resolution; Sciences of the Universe; Simulation; Snow; Snow accumulation; Snow depth; Spatial variability; Spatial variations; Temperature
• ISSN: 0899-8418; 1097-0088
• DOI: 10.1002/joc.7637

Convection‐permitting climate modelling is a promising avenue for climate change research and services especially in mountainous regions. Work is required to evaluate the results of high‐resolution simulations against relevant observations, and put them in a broader context against coarser resolution modelling frameworks. Here we evaluate numerical simulations with the convection‐permitting regional climate model CNRM‐AROME ran at 2.5 km horizontal resolution over a large pan‐Alpine domain in the European Alps, using either the ERA‐Interim or climate model output as boundary conditions. This study analyses annual and seasonal characteristics of 2 m temperature, total precipitation, solid fraction of precipitation and snow depth at the scale of the French Alps under past and future climate conditions. The results are compared with the local reanalysis S2M, and raw or adjusted, with the ADAMONT method, simulations of the regional climate model CNRM‐ALADIN driven either by the ERA‐Interim reanalysis or the CNRM‐CM5 global climate model. The study highlights generally similar differences in past and future climate between the datasets, as well as obstacles to the use of some CNRM‐AROME outputs as they stand. These consist of excessive accumulation of snow on the ground above 1,800 m a.s.l., as well as lower temperature values at same elevations than the S2M reanalysis and the ADAMONT‐adjusted outputs. Besides these obstacles, CNRM‐AROME simulations present several advantages compared to the raw CNRM‐ALADIN outputs. Among them, a significantly smaller cold bias, more realistic values of accumulated precipitations, as well as a better representation of the spatial variability of the different variables investigated, which always stand closer to the reference data than in the CNRM‐ALADIN outputs. As suggested by many studies, CNRM‐AROME could even produce more realistic accumulated precipitations at high elevation than the S2M reanalysis taken as our reference and consequently than the ADAMONT‐adjusted projections, but the lack of a reliable set of high‐resolution observations at high elevation remains an obstacle to their evaluation. 3‐days cumulated precipitation (in kg m−2) during the Eleanor storm 2 to 4 January 2018 simulated by AROME, ALADIN (both driven by ERA‐Interim) and the S2M reanalysis at 2100 m elevation