How the treatment of sea surface temperature affects the water cycle in EURO-CORDEX simulations

Regional climate models (RCMs) over Europe often exhibit wet precipitation biases, primarily attributed to excess oceanic evaporation across time scales. One likely source for such wet biases are therefore imperfections in the models’ lower boundary condition (LBC) over the ocean, as realized by time-evolving sea surface temperature (SST) fields. SST data from atmospheric reanalyses (e.g., ERA5) are commonly adopted in RCMs, but ambiguity exists about the exact SST variable in these products (e.g., foundation or skin temperature) and the manner with which they represent the diurnal cycle and spatial gradients. Here we explore these questions with a ~12-km setup of ICON-CLM (Icosahedral Nonhydrostatic Model in Limited-Area Mode) over the EURO-CORDEX domain, run repeatedly for 6 years with various SST datasets. We use ERA5-based daily SST and skin temperature and hourly upper-layer SST drawn from our own global ocean simulations with FESOM2 (Finite Element Sea-Ice Ocean Model) at ~10-km node spacing in the eastern North Atlantic. Specifically, prescribing the FESOM2 SST fields in ICON-CLM both with and without spatial smoothing allows us to examine the effects of oceanic eddies and fronts on precipitation characteristics onshore. Preliminary results from 7 months of integration with ICON-CLM suggest that the choice of the SST data appreciably impacts latent heat fluxes, moisture transport onto land, and cumulative continental precipitation, generally in areas of pronounced moisture recycling. “Mind your SST” is therefore the advice we can give to ongoing dynamical downscaling efforts aimed at modeling future precipitation changes over land.