Scientific content of the C ESM EP (original) (raw)

The C-ESM-EP provides a set of pre-defined atlases (collection of diagnostics) to evaluate a climate model. Those atlases are the results of working groups on different scientific thematics.

The C-ESM-EP comprises a collection of atlases/diagnostics for:

• the atmosphere
• the oceans: blue, white and green
• the land surfaces
• ENSO
• the air-sea fluxes
• the monsoons
• evaluation and tuning metrics

The content of the atlases is detailed in this page.

Reminders on the directory structure (where the atlas directories actually are):

The comparison directory standard_comparison is located in ${my_cesmep_workdir}/standard_comparison (see here how to run it).

The atlas modules are located in this directory. The atlas modules are directories and they contain:

• a parameter file atlasmodule/params{atlas_module}/params_atlasmodule/params{atlas_module}.py; the user controls the content of the atlas from this parameter file (see the page on Atlas Explorer for a general review of the functionalities accessible from the parameter file; have a look, there are many things you can do!)
• error, output and log files (also the last.out and climaf.log files produced by CliMAF)

If you don't need an atlas in your comparison (as well as on the frontpage), simply remove the directory. Inversely, you can add an atlas to your comparison by copying the atlas directory in your comparison directory (see here how to add your own atlas directory).

Start with the frontpage

See here the front page of the standard comparison for an example of use of the C-ESM-EP.

Send an e-mail to the contact (bottom of the home page) if the page is not accessible.

We will now detail the content of the pages in the following.

Atmosphere: 6 atlases/pages (LMD)

The atlases are the same pages as the bias maps of the "LMDZ-patchwork" evaluation atlas. First column is the climatology of the reference, and the other columns are the differences between the simulations and the reference (bias maps if the reference is an observational reference). It is based on the mechanism of the Atlas Explorer.

Here is the description of the atmosphere atlases.

Atlas name on frontpage: Atmosphere Surface - seasonal

• Atlas directory: Atmosphere_Surface
• bias maps against a reference (first column: climatology of the reference):
  • activated by the instruction do_atmos_maps = True
  • for Global, annual mean, DJF and JJA
  • surface variables:
    * 2m air temperature: tas
    * Precipitation: pr
    * Latent and Sensible Heat Fluxes: hfls and hfss
    * Zonal and Meridional Wind Speed at 10m: uas and vas
    * Zonal and Meridional Wind stress: tauu and tauv
    * Relative Humidity at 2m: hurs
    * albedo at TOA and surface: albt and albs
    * Upward ShortWave (total and clear-sky): rsut, rsutcs
    * Upward LongWave (total and clear-sky): rlut, rlutcs,
    * Total Cloud Radiative Effect at TOA: crett
    * Cloud Radiative Effect on LongWave and ShortWave at TOA: crest, crelt
    * Cloud Radiative Effect on SW at Surafce: cress
• additional instructions in the parameter file:
  • my_seasons: python list with the seasons
  • atmos_variables_list: python list of variables
  • atmos_variables: final python list used by the atlas; the loop creates dictionaries with one variable, one season from my_season, and adds additionnal project-dependant specifications to access the datasets

Atlas name on frontpage: NH Polar St. Atmosphere Surface and SH Polar St. Atmosphere Surface

• Atlas directories: NH_Polar_Atmosphere_Surface and SH_Polar_Atmosphere_Surface
• Northern Hemisphere Polar Stereographic projection, annual mean, DJF and JJA
• Instructions and variables are the same as Atmosphere_Surface

Atlas name on frontpage: Atmosphere Standard press. lev. - seasonal

• Atlas directory: Atmosphere_StdPressLev
• Global Robinson projection, annual mean, DJF and JJA
• Instructions are the same as Atmosphere_Surface
• Variables:
  • zonal and meridional wind speed (ua, va) and temperature (ta) on the column, at 850 and 200mb
  • Geopotential Height (zg) at 500mb

Atlas name on frontpage: NH Polar St. Atmosphere Standard press. lev. and SH Polar St. Atmosphere Standard press. lev.

• Atlas directory: NH_Polar_Atmosphere_StdPressLev and SH_Polar_Atmosphere_StdPressLev
• Northern Hemisphere Polar Stereographic projection, annual mean, DJF and JJA
• Instructions and variables are the same as Atmosphere_StdPressLev

Atlas name on frontpage: Atmosphere Zonal mean - seasonal

• Atlas directory: Atmosphere_zonmean
• Zonal averages lat/pressure sections, annual mean, DJF and JJA
• Instructions are the same as Atmosphere_Surface
• Variables (atmos_variables_list):
  • zonal and meridional wind speed: ua, va
  • temperature: ta
  • specific and relative humidity: hus, hur

Ocean: 4 atlases with the NEMO group

The NEMO atlases are the result of a working group that specified the essential diagnostics to evaluate the ocean in a coupled model.

Atlas name on frontpage: NEMO General Diagnostics:

• Atlas directory: NEMO_main
• Bias maps of surface variables:
  • activated by the instruction do_ocean_2D_maps=True
  • Global, annual mean, DJF and JJA
  • for the variables in ocean_2D_variables:
    * Sea Surface Temperture (SST): tos
    * Sea Surface Salinity (SSS): sos
    * Sea Surface Height (SSH): zos
    * Precipitation - Evaporation: wfo
• Mixed Layer Depth (mlotst):
  • activated by the instruction do_MLD_maps=True
  • climatology of De Boyer Montegut (first column)
  • full field of the simulations (subsequent columns)
  • on annual mean, DJF, JJA and annual Max for global and polar stereographic projections
• Wind Stress Curl (socurl):
  • activated by the instruction do_curl_maps=True
  • method: cdfcurl (cdftools); only for ORCA grid
  • full field of the simulations
  • on annual mean, DJF, JJA for global, North Atlantic and Southern ocean (polar stereographic)
• MOC Diagnostics:
  • activated by the instruction do_ATLAS_MOC_DIAGS=True
  • Global, Atlantique, Pacific and Indian MOC (python list MOC_basins)
  • lat/depth sections
  • method: uses the zomsf[basin] variables if they exist, or compute the Meridional Overturning Circulation using msft
• Sea ice maps:
  • activated in parameter file by do_seaice_maps=True
    * Sea ice concentration (sic) map (colors) of the simulation vs isocontour 15 of the reference (NSIDC by default)
    * sea ice thickness (sit) and sea ice volume (sic*sit), full field
• Annual cycle of the total volume of sea ice:
  • activated with do_seaice_annual_cycle=True
  • Northern and Southern Hemisphere separately

Atlas name on frontpage: NEMO T - S @depth :

• Atlas directory: NEMO_depthlevels
• Bias maps:
  • switch on in parameter file with do_ocean_2D_maps=True
  • potential temperature (thetao) and salinity (so)
  • at 200m and 1000m
  • method: 'cdo intlevel' on thetao and so
  • runs in parallel with do_parallel=True

Atlas name on frontpage: NEMO zonal means

• Atlas directory: NEMO_zonmean
• Zonal averages of 3D potential temperature (thetao) and salinity (so) per basin
  • instruction do_ATLAS_ZONALMEAN_SLICES=True
  • lat/depth sections of zonal averages
  • method: uses the zo[var][basin] variables when available; if not, regrid the model both horizontally and vertically on WOA13 and uses the WOA13 land-sea mask and basins masks to compute the zonal averages ('cdo zonmean' on the model regridded on the regular grid of WOA13)

Atlas name on frontpage: NEMO PISCES:

• Atlas directory: NEMO_PISCES
• Bias maps against WOA09:
  • instruction do_biogeochemistry_2D_maps=True
  • Variables (python list ocebio_2D_variables):
    * Nitrates: NO3
    * phosphate: PO4
    * Silicate: Si
    * Oxygen: O2
    * at surface (first model level), 300, 1000 and 2500m at depth
    * method: 'cdo sellevidx,1' on the 3D variables to get the first level and 'cdo intlevel,${depth}' to get the 300, 1000 and 2500m levels

Land Surfaces: ORCHIDEE group

This atlas was born from consultations with the ORCHIDEE group.

Atlas name on frontpage: ORCHIDEE

• Atlas directory: ORCHIDEE
• Three main thematics: Water, energy and carbon budget
• Section on Energy Budget:
  • variables (python list variables_energy_budget):
    * Latent Heat Flux: fluxlat
    * Sensible Heat Flux: fluxsens
    * Surface Albedo 'Visible': albis
    * Surface Albedo 'Near Infra Red': albnir
    * 2m air temperature: tair
    * SW down at surface: swdown
    * LW down at surface: lwdown
  • instruction do_ORCHIDEE_Energy_Budget_climobs_bias_modelmodeldiff_maps=True: section with climatologies, bias maps and model-model differences
  • instruction do_ORCHIDEE_Energy_Budget_climobs_bias_maps=True: section with climatology of the reference and bias maps
  • instruction do_ORCHIDEE_Energy_Budget_climrefmodel_modelmodeldiff_maps=True: section with climatology of first simulation and model-model differencies
• Section on Water Budget:
  • variables (python list variables_water_budget):
    * Evaporation Bare soil: evapnu
    * Sublimation: subli
    * Total evaporation: evap
    * Runoff: runoff
    * Drainage: drainage
    * Snow fall: snow
  • instruction do_ORCHIDEE_Water_Budget_climobs_bias_modelmodeldiff_maps=True: section with climatologies, bias maps and model-model differences
  • instruction do_ORCHIDEE_Water_Budget_climobs_bias_maps=True: section with climatology of the reference and bias maps
  • instruction do_ORCHIDEE_Water_Budget_climrefmodel_modelmodeldiff_maps=True: section with climatology of first simulation and model-model differencies
• Section on Carbon Budget:
  • variables (python list variables_carbon_budget):
    * Global Primary Productivity: gpptot
    * Leaf Area Index: lai
    * GPP treeFracPrimDec
    * GPP treeFracPrimEver
    * GPP c3PftFrac
    * GPP c4PftFrac
    * Total soil carbon: total_soil_carbon_PFT_tot
  • instruction do_ORCHIDEE_Carbon_Budget_climobs_bias_modelmodeldiff_maps=True: section with climatologies, bias maps and model-model differences
  • instruction do_ORCHIDEE_Carbon_Budget_climobs_bias_maps=True: section with climatology of the reference and bias maps
  • instruction do_ORCHIDEE_Carbon_Budget_climrefmodel_modelmodeldiff_maps=True: section with climatology of first simulation and model-model differencies

ENSO CLIVAR diagnostics

Atlas name on frontpage: ENSO CLIVAR diagnostics

This collection of diagnostics correspond to the recommandations of the CLIVAR group for the evaluation of ENSO.

• Atlas directory: ENSO
• one global instruction do_ENSO_CLIVAR=True to activate the whole section; additional instructions activate:
  • SSTA Nino3.4 index: monthly SST averaged over 210-270W / -5/5N, annual cycle removed => do_ENSO_CLIVAR_sstanino3_timeseries=True
  • map of the standard deviations of SSTA (monthly SST anomalies against annual cycle) => do_ENSO_CLIVAR_SSTA_std_maps=True
  • climatology of the precipitations (pr) => do_ENSO_CLIVAR_pr_climatology_maps=True
  • climatology of the zonal wind stress (tauu) => do_ENSO_CLIVAR_tauu_climatology_maps=True
  • map of regression coefficients between zonal wind stress (tauu, by grid point) and the Nino3.4 index: d(tauu)/d(SSTA Nino3.4) => do_ENSO_CLIVAR_linearRegression_dtauu_dsstanino3_maps=True
  • map of regression coefficients between Shortwave (rsut, by grid point) and the Nino3.4 index: d(rsut)/d(SSTA Nino3.4) => do_ENSO_CLIVAR_linearRegression_drsds_dsstanino3_maps=True
  • annual cycles of:
    * SST Nino3.4 (raw value)
    * SSTA Nino3.4 (anomalies)
    * Standard deviation of SSTA month by month
    * one plot with all simulations together and plots with one simulations vs reference
    * do_ENSO_CLIVAR_SSTA_annualcycles=True
  • profile of the annual mean climatology of zonal wind stress (-5/5N):
    * one plot with all simulations together and plots with one simulations vs reference
    * do_ENSO_CLIVAR_longitudinal_profile_tauu=True

Turbulent Air-Sea Fluxes: Gainusa-Bogdan et al (2016) and Hotelling Test optional atlas:

Atlas name on frontpage: Turbulent Air-Sea Fluxes (GB2015)

This atlas shows the maps from the paper Gainusa-Bogdan et al 2016 (see the paper for more details). We use an ensemble or reference observational datasets as the reference. The maps show where the model is out of the range of the observational ensemble (dots on the maps). The differences are calculated against the ensemble mean of the observational ensemble. Note that we do not consider the ensemble mean of the observational ensemble as the best estimate of the fluxes climatologies, but it gives us an indication on the sign and magnitude of the difference.

• Atlas directory: TurbulentAirSeaFluxes
• Variables (python list TurbFluxes_variables):
  • Latent and Sensible Heat Fluxes (hfls and hfss)
  • Zonal and Meridional Wind stress (tauu and tauv)
• instruction do_GLB_SFlux_maps=True: Global / annual mean (fewer observational products in the ensemble)
• instruction do_Tropics_SFlux_maps=True: Tropical band (-30/30N), annual mean and seasonal

Large scale Metrics: parallel coordinates with PCMDI metrics package

Atlas name on frontpage: Parallel Coordinates - PMP PCMDI

• Atlas directory: ParallelCoordinates_Atmosphere

We use the PMP (PCMDI Metrics Package) to compute a set of evaluation/performance metrics. The page shows the results using the plotting method called parallel coordinates: the vertical axes display the results of the metrics of the models in raw values and the range of values covered by the axes corresponds to the range of values covered by the available results. The models are identified by lines connecting the scores obtained by each model on each metric. The CMIP5 ensemble (grey lines) gives us a background for a range of values. The columns are sorted so that the reference simulation appears as an ascending line from the left to the right to ease the visualization of the results. It helps highlighting whether the simulations are 'better' or 'worse' on the metrics against the 'model version to beat'. Syntax:

• rms: Root Mean Square Error
• rms: Centered (de-biased) Root Mean Square Error
• bias: explicit
• cor: spatial (pattern) correlation
• xyt: spatio temporal (the 12 maps of the climatological months)
• xy: spatial only (the metric is computed on a map)
• ann: annual mean if associated with xy, the annual cycle if associated with xyt
• djf, mam, jja, son: the season of the map (there is no xyt for the seasons)
• GLB, NHEX, TROPICS, SHEX: the geographical domain: global, Extratropical Northern Hemisphere (20/90N), Tropical band (-20/20N), Extratropical Southern Hemisphere (20/90S)
• variables:
  • atmosphere only -> tas, pr, prw, psl, rltcre, rstcre, [ua,va,ta][850,200], zg500
  • depending on the availability of the data (missing variables listed under the legend)

The first section shows metrics on the annual cycle (rms_xyt) and the annual mean map (rmsc_xy, bias_xy and cor_xy) over the globe. These are the results that are mostly used as first check for model evaluation of the mean state. The second section consists in seasonal/regional metrics to go in further details. The user can control the content of those sections using the metrics_sections python list. Each element of this list is a dictionary describing for the section:

• the statistics
• the region
• season
• and the title of the section

Tuning Metrics: metrics on SST 50°S/50°N

Metrics used for the tuning of the IPSL-CM6 model:

• raw value of the SST averaged over the 50°S/50°N area (mean state / bias)
• Root Mean Square (RMS) and de-biased RMS (RMSC) on the annual mean map
• results shown additionnally for all available CMIP5 models

Atlas Explorer

Atlas name on frontpage: Atlas Explorer

• atlas directory: AtlasExplorer

The scientific content of Atlas Explorer consists of the climatology of a reference and difference maps with this reference. The variables, seasons, projections... are up to you!

Optional atlases:

• Hotelling Test
• Atlantic Atmosphere Surface
• Focus Atlantic AMOC Surface
• Precip Annual Cycle
• Tuning Metrics
• Check data request ping