Regional climate simulations with the MAR V 3.9 model by University of Liège. Dynamical downscaling on the CORDEX EUR-11 domain and HRes domain over Belgium at convection-permitting scale. Model name: MAR V. 3.9 Important reference: Wyard et al. 2017: https://dx.doi.org/10.1002/joc.4879 Resolution: RCM: 50 km and 12.5 km; LAM: 5 km Nr. vertical levels: 30 Time step (s): Important scheme: Snow variables Focal time series / severity index: Snowfall events and snowmelt events inducing floods Host GCM: ERA-Interim / various Non-hydrostatic: no

Regional climate simulations with the ALARO-0 model by the Royal Meteorological Institute of Belgium and Ghent University (RMIB-UGent). Dynamical downscaling on the CORDEX EUR-11 domain and HRes domain over Belgium at convection-permitting scale. Model name: ALARO-0 Important reference: Giot et al. 2016: https://dx.doi.org/10.5194/gmd-9-1143-2016 Resolution: RCM: 50 km and 12.5 km; LAM: 4 km Nr. vertical levels: 46 Time step (s): 900 (50 km); 300 (12.5 km); 180 (4 km) Important scheme: 3MT Focal time series / severity index: (Sub-)hourly precipitation Host GCM: ERA-Interim / ARPEGE Non-hydrostatic: no

This is an atmospheric hourly hindcast for the German Bight using COSMO-CLM version 5.00_clm2 from 1948-August 2015 (http://www.cosmo-model.org/content/model/documentation/core/default.htm). The model uses a rotated grid with 250 x 180 grid points and a grid point distance of 0.025 degrees, the rotated North pole is located at 172.97 W, 34.925 N. The forcing is coastDat2 doi:10.1594/WDCC/coastDat-2_COSMO-CLM . In rotated coordinates the model area extends from 2.25 W to 2.25 E, 3.125 S to 3.125 N, in geographical coordinates this corresponds to about 1.3 E to 12.8 E, 52.7 N to 57.3 N.

Regional climate simulations with the COSMO-CLM V. 6.0-CLM6 model by UCLouvain. Dynamical downscaling on the CORDEX EUR-11 domain and HRes domain over Belgium at convection-permitting scale. Model name: COSMO-CLM V. 6.0-CLM6 Important reference: Wyard et al. 2017: https://dx.doi.org/10.1002/joc.4879 Resolution: RCM: 12.5km; LAM: 2.8 km Nr. vertical levels: 40 Time step (s): 80 (12.5 km); 20 (2.8 km) Important scheme: Two-moment microphysics scheme Focal time series / severity index: Hail mixing ratio and number concentration, detailed precipitation Host GCM: ERA-Interim / MPI-ESM Non-hydrostatic: yes

Regional climate simulations with the ALARO-0 model by the Royal Meteorological Institute of Belgium and Ghent University (RMIB-UGent). Dynamical downscaling on the CORDEX EUR-11 domain and HRes domain over Belgium at convection-permitting scale. Model name: ALARO-0 Important reference: Giot et al. 2016: https://dx.doi.org/10.5194/gmd-9-1143-2016 Resolution: RCM: 50 km and 12.5 km; LAM: 4 km Nr. vertical levels: 46 Time step (s): 900 (50 km); 300 (12.5 km); 180 (4 km) Important scheme: 3MT Focal time series / severity index: (Sub-)hourly precipitation Host GCM: ERA-Interim / ARPEGE Non-hydrostatic: no

Regional climate simulations with the MAR V 3.9 model by University of Liège. Dynamical downscaling on the CORDEX EUR-11 domain and HRes domain over Belgium at convection-permitting scale. Model name: MAR V. 3.9 Important reference: Wyard et al. 2017: https://dx.doi.org/10.1002/joc.4879 Resolution: RCM: 50 km and 12.5 km; LAM: 5 km Nr. vertical levels: 30 Time step (s): Important scheme: Snow variables Focal time series / severity index: Snowfall events and snowmelt events inducing floods Host GCM: ERA-Interim / various Non-hydrostatic: no

This is an atmospheric hourly hindcast for the German Bight using COSMO-CLM version 5.00_clm2 from 1948-August 2015 (http://www.cosmo-model.org/content/model/documentation/core/default.htm). The model uses a rotated grid with 250 x 180 grid points and a grid point distance of 0.025 degrees, the rotated North pole is located at 172.97 W, 34.925 N. The forcing is coastDat2 doi:10.1594/WDCC/coastDat-2_COSMO-CLM . In rotated coordinates the model area extends from 2.25 W to 2.25 E, 3.125 S to 3.125 N, in geographical coordinates this corresponds to about 1.3 E to 12.8 E, 52.7 N to 57.3 N.

The experiment aims to investigate how the representation of convection influences the West African Monsoon during the mid-Holocene. Atmospheric and SST input data originate from the MPI-ESM Holocene simulations reflecting Holocene condition. External Parameters (surface condition) reflect present-day conditions similar to the experimental setup of PMIP1: The Sahara remains a desert. We use the ICON (ICOsahedral Nonhydrostatic) model framework version 2.5.0 (see Zängl et al. (2014) for more details). The provided data covers one simulation from June to October (JJASO) for the year 7023 before present (BP) with the year 2000 as the reference year. The time axes of the NetCDF files reflect the model year which is based on the time axes of the MPI-ESM slo0021a Holocene simulations. The artificial model year 1001 in slo0021a refers to the year 8000 BP. Therefore, the model year 1977 refers to the year 7023 BP. The experiment compares a 5km horizontal resolution, cloud-resolving simulation with a 40km-horizontal resolution, parameterized convection simulation. The 40km-domain (DOM01) covers a range from 70.5°W - 99.5°E; 49°S - 59°N The 5km-domain (DOM04) covers a range from 37°W - 53°E; 0°N - 40°N The dataset provides daily mean values on the triangular ICON grid. The datasets provide atmospheric (3D), surface (2D) and precipitation (2D) data an the following variables: rain_con_rate, rain_gsp_rate, clct, geopot, temp, rh, qv, u, v, w, w_so, runoff_g, runoff_s, lhfl_s, shfl_s, soiltyp

The experiment aims to investigate how the representation of convection influences the West African Monsoon during the mid-Holocene. Atmospheric and SST input data originate from the MPI-ESM Holocene simulations reflecting Holocene condition. External Parameters (surface condition) were adjusted to reflect mid-Holocene vegetation conditions. We prescribe an idealized, denser vegetation cover based on the simulated desert fraction (1-vegetation fraction) of the transient mid-Holocene MPI-ESM simulations (Dallmeyer et al., 2020). We use the ICON (ICOsahedral Nonhydrostatic) model framework version 2.5.0 (see Zängl et al. (2014) for more details). The provided data covers one simulation from June to October (JJASO) for the year 7024 before present (BP) with the year 2000 as the reference year. The time axes of the NetCDF files reflect the model year which is based on the time axes of the MPI-ESM slo0021a Holocene simulations. The artificial model year 1001 in slo0021a refers to the year 8000 BP. Therefore, the model year 1978 refers to the year 7024 BP. The experiment compares a 5km horizontal resolution, deep convection resolving simulation with a 40km-horizontal resolution, parameterized convection simulation. The 40km-domain (DOM01) covers a range from 70.5°W - 99.5°E; 49°S - 59°N The 5km-domain (DOM04) covers a range from 37°W - 53°E; 0°N - 40°N The dataset provides daily mean values on the triangular ICON grid. The datasets provide atmospheric (_atm_ 3D), surface (_sfc_ 2D) and precipitation ( _prec_ 2D) and forcing (extpar_) data and the following variables: w_so, lhfl_s, shfl_s, runoff_s, runoff_g, rain_con_rate, rain_gsp_rate, geopot, temp, rh, qv, u, v, w, clc Precipitation and forcing data are combined into 2 data files for DOM01 and DOM04. Surface data are combined into 2 data files for DOM01. The dataset with the suffix "_constSM_sep_" contains the data for the constant soil moisture simulations. In these simulations we prescribe the same constant soil moisture field, representing 1st September-soil moisture conditions, both for the 40km-P (DOM01) and the 5km-E (DOM04) simulations. The DS ("Dry Sahara") simulations refer to the simulations where we prescribe present-day land surface cover - the Sahara remains a desert. The GS ("Green Sahara") simulations represent idealized mid-Holocene conditions where we prescribe a higher vegetation cover that also extend further north. The Sahara reflects more savannah-like vegetation.