Supplementary material for "Rare event algorithm study of extreme warm summers and heatwaves over Europe"

DOI10.5281/zenodo.4763283Zenodo4763283MaRDI QIDQ6707766FDOQ6707766

Dataset published at Zenodo repository.

Publication date: 20 May 2021

Copyright license: Creative Commons Attribution 4.0 International

This dataset contains supplementary material for thepaperRare event algorithm study of extreme warm summers and heatwaves over Europe (Ragone and Bouchet 2021) submitted toGeophysical Research Letters (preprint available atRagone and Bouchet 2021, arXiv:2009.02519).We include here: 1) a description of the model setup andof the simulations, 2) sample scripts for the application of the rare event algorithmto CESM1.2.2, 3) data on extreme warm summers and heatwaves from simulations performed with CESM 1.2.2. Description of model setup and computational environment All simulations have been performed with the standard setup of the F2000 compset ofCESM 1.2.2. with CAM4 as choice of theatmospheric model (Hurrel et al. 2013). This is aprescribed sea surface temperature (SST) setup, where the SST varies monthly and is repeatedcyclically every year. The values of the SST andthe greenhouse gases concentrationsare calibrated in order to generate a stationary climate consistent with the observedclimate of the year ca. 2000 (see Hurrel et al. 2013 and the documentation of the model for more details, https://www.cesm.ucar.edu/models/cesm1.2/). All the simulations have been performed using 144 processors on the supercomputer Occigen of the computing center CINES (https://www.cines.fr/calcul/materiels/occigen/). Description of the simulations Three set of simulations have been performed: 1) a controlensemble of 10 simulations of 100 years each. They are referred in the following and in the dataset as batch n, with n=1,...,10. The 10 ensemble members of each batch start from different initial conditions and are statistically independent. The initial condition for batch n=1is the default one of the model. The initial conditions for batch n=2,...,10have been generated by taking the 1st of January every 10 years from batch 1 (thus the beginning of years 10, 20,...,100), and adding a random perturbation.The perturbation is added multiplying at each vertical level the amplitudes of the spherical harmonics of the potential temperature field by a factor \(\gamma = 1+\epsilon r\), where \(\epsilon = 10^{−4}\) and \(r\)is arandom numberextracted between -1 and 1 (different for each spherical harmonic, but the same at each vertical level for a given spherical harmonic). 2) a set of 10 experiments where we have applied arare event algorithm (Ragone et al. 2018) to ensemble simulations withCESM 1.2.2 to study heatwaves over France. Each experiment consists of an ensemble simulation with 100 ensemble members. Foreach experiment the ensemble members start from 100 different initial conditions. The initial conditions are taken as the 1st of June of the 100 years of one of the control batches. The 10 rare event algorithmexperiments are therefore alsoreferred in the dataset as batch n, with n=1,...,10, indicating from which control batch the initial conditions of the ensemble members have been taken. Each of the 100 ensemble members consists of a simulation 90 days long, and along the time evolution of the ensemble we have applied the rare event algorithm to select trajectories leading to heatwaves over France, as described in Ragone and Bouchet (2021). 3) the same as 2), but with with the rare event algorithm set toselect trajectories leading to heatwaves over Scandinavia. Scripts We include the scripts for two applications of the rare event algorithm to CESM1.2.2, to select trajectories over France or Scandinavia. Refer to Ragone and Bouchet (2021) for a description of the rare event algorithm.CAM4_F2000_p144_GK_France_k30_PT10m4_scale1_batch_0001.GK_run andCAM4_F2000_p144_GK_Scandinavia_k30_PT10m4_scale1_batch_0001.GK_run are CESMrun scripts modified to perform the trajectory selection. Thesecall the other scripts, described in the following. The scriptextract_observable_CAM is used to extract from the partial output the control observable used in the definition of the weights.Thecore scripts performing the resampling areresampling_CAM_France.py andresampling_CAM_Scandinavia.py. The scripts perturb_ic_spectral_fac is used to add the random perturbation after the cloning. The script shuffleic_CAM.py is used to rearrange the restart files according to the cloning.The scripts launch_* are utility scripts used in the main run scripts to interface with the other scripts. Data The dataset is organized as a collection of tar files as follows. Time series of spatially averaged quantities and related informations land_sea_masks.tar: contains an nc file with theland-sea mask of the model, andan npz file with the masks for the areas of France and Scandinavia that have been used to compute the surface temperature spatial averages TS_JJA_France_CAM4_F2000_p144_ctrl.tar: contains 10 npz files with the time series of thesurface temperature spatially averaged over France for the 10 control batches. Data are organized as 100x720 arrays, each row being 3 hourly valuesfrom June 1st to August 28 for each year of the control run. TS_JJA_Scandinavia_CAM4_F2000_p144_ctrl.tar: same as above, but for Scandinavia CAM4_F2000_p144_GK_France_k30_resampling.tar: contains 10 npz files withthe time series of thesurface temperature spatially averaged over France for the 10 experiments with the rare event algorithm analysing heatwaves over France, and all the informations on the selection of the trajectories perfomed by the algorithm. These correspond to a collection of theoutput generated by resampling_France.py during the simulation (see the script for the meaning of each variable). Quantities indicated by *_ens correspond to the straight output, while quantities indicated by *_eff correspond to the same quantity associated to the trajectories in the effective ensemble (see Ragone and Bouchet 2021). CAM4_F2000_p144_GK_Scandinavia_k30_resampling.tar: same as above, but for Scandinavia Composite averages of extremes in the control runs composite_heatwaves_CAM_{region}_{observable}_ctrl_JJA90_period90_perc{percentile}.tar: these files contain the composite averages of the time mean over summer of the fields of the observable {observable}over the entire planet in the control runs, conditional on years for which themean summer surface temperature averaged over the region {region} ishigher than its{percentile} percentile. The regions are France or Scandinavia, the observables surface temperature (TS) or the 500 hPa geopotential height (Z3.500hPa), the percentiles 99th and 99.9th. Each file contains 10 tar files, one for each control batch. The percentiles have been computed over the total 1000 years, so some files can be empty, as the corresponding batch did not contain any relevant event (most feature 1or 2 events). The tar files for each batch contain the composite average of the time averaged fields (*_m1.nc) and the composite average of the squares of the time averaged fields (*_m2.nc). composite_heatwaves_CAM_{region}_{observable}_k30_JJA90_period90_perc{perc}.tar: these files contain the composite averages of the time mean over summer of the fields of the observable {observable}over the entire planet in the experiments with the rare event algorithm, conditional on trajectories for which themean summer surface temperature averaged over the region {region} ishigher than its{percentile} percentile computed on the control runs. The regions are France or Scandinavia, the observables surface temperature (TS) or the 500 hPa geopotential height (Z3.500hPa), the percentiles 99th and 99.9th. Each file contains for each batch 4 nc files composite_heatwaves_CAM_{region}_{observable}_k30_batch{batch_id}_JJA90_period90_perc{perc}_data_m1.nc: composite averageof the time averaged fields; composite_heatwaves_CAM_{region}_{observable}_k30_batch{batch_id}_JJA90_period90_perc{perc}_data_weighted_m1.nc: as above, but the average has been taken weighing the contribution of each trajectory according to the weights of the resampling; composite_heatwaves_CAM_{region}_{observable}_k30_batch{batch_id}_JJA90_period90_perc{perc}_data_m2.nc: composite averageof the squares of the time averaged fields; composite_heatwaves_CAM_{region}_{observable}_k30_batch{batch_id}_JJA90_period90_perc{perc}_data_weighted_m2.nc: as above, but the average has been taken weighing the contribution of each trajectory according to the weights of the resampling; The files *_clim_m1.nc and *_clim_m2.nc are not used for the analysis presented in Ragone and Bouchet(2021). References 1) Ragone and Bouchet (2021), arXiv:2009.02519 2)Hurrell, J. W., Holland, M. M., Gent, P. R., Ghan, S., Kay, J. E., Kushner, P. J., ... Marshall, S. (2013). The community earth system model: A framework for collaborative research. Bulletin of the American Meteorological Society, 94(9), 1339-1360. doi: 10.1175/BAMS-D-12-00121.1

This page was built for dataset: Supplementary material for "Rare event algorithm study of extreme warm summers and heatwaves over Europe"