CMIP6 Climate Scenario data: Where do I get the data!

by Melanie Frazier, Ben Halpern, Alejandra Vargas, Mandy Lombard

Fri, Dec 16, 2022

If you are an ecologist or biologist and need predictions of future climate variables, then the CMIP6 Scenario data will probably be what you want!

In this blogpost, we provide an overview of the raw CMIP6 data because this is helpful for understanding these data. However, because we believe non-climate scientists should probably seek more derived CMIP6 products, we also provide information about some of these products that we have used in our research.

A previous post provides an overview of the CMIP6 Scenario data. If you are new to CMIP data, we recommend starting there. This article, “CMIP6: the next generation of climate models explained” also provides an excellent overview of CMIP6 that manages to be high-level and understandable. And, much of the content in this post is derived from this excellent source, which I highly recommend reading because it provides a super useful and down-to-earth description of CMIP6.

We are always on the hunt for data. Please let us know (frazier@nceas.ucsb.edu) if you discover CMIP6 ensemble datasets and/or downscaled data that are publically available. I will add them to this post. MANY THANKS!

Raw CMIP data

Earth System Grid Federation: A deep dive

The CMIP6 data archive is distributed through the Earth System Grid Federation (ESGF).


If the ESGF website seems daunting, don’t worry, we feel the same!

There are a series of dropdown menus on the left of the ESGF site that will guide you through the process of finding data. We recommend an iterative approach because if the “Search” button is clicked after selecting a menu option, the subsequent menus will be simplified as irrelevant categories are removed.


Source ID & Institution ID

The global climate models and institutions that produce them. I usually skip over this section because it is rare for me to select on the model, and depending on the climate variable of interest, many of these options will disappear.

Click the triangle to see all the climate models and institutions

CMIP6 data models and institutions (from this incredibly helpful resource):

goal 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
1 Index 69.4 69.9 70.3 70.6 70.7 70.4 70.4 70 68.8 69 69.2
2 Artisanal opportunities 75.8 76.3 76.7 76.7 75.9 75.1 75.1 74.9 75.4 74.8 77.1
3 Species condition (subgoal) 79.7 79.5 79.3 79.1 78.9 78.7 78.5 78.3 78.1 77.9 77.7
4 Biodiversity 79 78.5 78 78.2 77.9 77.4 77.1 76.7 76.4 76.3 76.3
5 Habitat (subgoal) 78.4 77.5 76.8 77.4 76.8 76.1 75.7 75.1 74.7 74.8 74.9
6 Coastal protection 82.6 82.5 82.6 82.6 82.7 82.2 81.9 81.7 81.8 81.9 82.3
7 Carbon storage 81 81 81 81 81 81 81 81.1 81.1 81 81
8 Clean water 67.7 67.6 68.5 68.8 69.2 69.3 69.4 69 69.2 70.2 70.2
9 Fisheries (subgoal) 53.7 54.4 54.9 54.8 54.1 53.8 53.4 53.5 53.5 53.4 53.4
10 Food provision 50.6 50.9 51.3 51.2 50.5 50.1 49.8 49.8 49.8 49.6 49.5
11 Mariculture (subgoal) 5.7 5.6 5.6 5.7 5.8 5.9 6 6.1 6.2 6.4 6.6
12 Iconic species (subgoal) 64 65 65 65.5 64.7 64.6 65.5 62.7 60.3 59.8 59.7
13 Sense of place 59.4 60 60.8 61.2 61.5 62.4 63.6 62.6 61.4 61.3 61
14 Lasting special places (subgoal) 54.8 55 56.6 57 58.2 60.1 61.6 62.6 62.5 62.7 62.3
15 Natural products 76.1 76 77.4 78.2 78.5 78.3 77 75.7 76 75.1 74.9
16 Tourism & recreation 44.5 45.6 46.3 47.4 49 48 48.4 48.1 35.3 38.5 38.8

Source Type

A description of the general circulation model. In most cases, the model is a coupled atmospheric and oceanic general circulation model (AOGCM), but other models are used.

Atmospheric (AGCMs) and oceanic GCMs (OGCMs) can be coupled to form an atmosphere-ocean coupled general circulation model (CGCM or AOGCM). With the addition of submodels such as a sea ice model or a model for evapotranspiration over land, AOGCMs become the basis for a full climate model. – Wikipedia

Experiment ID (aka “scenarios”)

The IPCC Sixth Assessment Report (CMIP6) includes 8 future scenarios (2015-2100) and one historical scenario (1850-2014). These scenarios represent different pathways the world might follow, which will lead to different predictions of future climate.

Click the triangle to see the climate scenarios

CMIP6 data climate scenarios (from this incredibly helpful resource) and here.

IPCC Scenarios Description Estimated Warming 2041-2060, C
historical Simulation of climate variables from the recent past from 1850 to 2014. These predictions are from a coupled atmosphere-ocean general circulation model (AOGCM) using observed variables such as atmospheric composition, land use and solar forcing. The historical simulation can be used to evaluate model performance against present climate and observed climate change. NA
SSP1-1.9 Based on SSP1 with low climate change mitigation and adaptation challenges which leads to a future pathway with a radiative forcing of 1.9 W/m2 in the year 2100. The SSP1-1.9 scenario fills a gap at the very low end of the range of plausible future forcing pathways, due to interest in informing a possible goal of limiting global mean warming to 1.5°C above pre-industrial levels based on the Paris COP21 agreement. 1.6
SSP1-2.6 Based on SSP1 with low climate change mitigation and adaptation challenges which leads to a radiative forcing of 2.6 W/m2 in the year 2100. The SSP1-2.6 scenario represents the low end of plausible future forcing pathways. SSP1-2.6 depicts a “best case” future from a sustainability perspective. 1.7
SSP4-3.4 Based on SSP4 in which climate change adaptation challenges dominate which leads to a radiative forcing of 3.4 W/m2 in the year 2100. The SSP4-3.4 scenario fills a gap at the low end of the range of plausible future forcing pathways. SSP4-3.4 is of interest to mitigation policy since mitigation costs differ substantially between forcing levels of 4.5 W/m2 and 2.6 W/m2.
SSP5-3.4OS Based on SSP5 in which climate change mitigation challenges dominate with a peak and decline in forcing towards an eventual radiative forcing of 3.4 W/m2 in the year 2100. The SSP5-3.4OS scenario branches from SSP5-8.5 in the year 2040 whereupon it applies substantially negative net emissions. SSP5-3.4OS explores the climate science and policy implications of a peak and decline in forcing during the 21st century. SSP5-3.4OS fills a gap in existing climate simulations by investigating the implications of a substantial overshoot in radiative forcing relative to a longer-term target.
SSP2-4.5 Based on SSP2 with intermediate climate change mitigation and adaptation challenges which lead to a radiative forcing of 4.5 W/m2 in the year 2100. The SSP2-4.5 scenario represents the medium part of plausible future forcing pathways. SSP2-4.5 is comparable to the CMIP5 experiment RCP4.5. 2.0
SSP4-6.0 SSP4-6.0 is based on SSP4 in which climate change adaptation challenges dominate and RCP6.0 which lead to a radiative forcing of 6.0 W/m2 in the year 2100. The SSP4-6.0 scenario fills in the range of medium plausible future forcing pathways. SSP4-6.0 defines the low end of the forcing range for unmitigated SSP baseline scenarios.
SSP3-7.0 Based on SSP3 in which climate change mitigation and adaptation challenges are high which leads to a radiative forcing of 7.0 W/m2 in the year 2100. The SSP3-7.0 scenario represents the medium to high end of plausible future forcing pathways. SSP3-7.0 fills a gap in the CMIP5 forcing pathways that is particularly important because it represents a forcing level common to several (unmitigated) SSP baseline pathways. 2.1
SSP5-8.5 SSP5-8.5 is based on SSP5 in which climate change mitigation challenges dominate which leads to a radiative forcing of 8.5 W/m2 in the year 2100. The ssp585 scenario represents the high end of plausible future forcing pathways. SSP5-8.5 is comparable to the CMIP5 experiment RCP8.5. 2.4

Variant Label

Modeling centers often run the same climate model with slightly different settings and initial conditions. A model and its collection of runs is referred to as an ensemble (not to be confused with “ensembles” that combine the various climate models, typically from different institutions). For some models, there is only one variant, but some models (e.g., CanESM5) include large numbers of variants.

These are interesting from a statistical perspective because they give some idea of the variation at this modeling scale.

Click the triangle to learn more about these ensembles

CMIP6 esembles (taken entirely from this incredibly helpful resource): Within these ensembles, four different categories of sensitivity studies are done, and the resulting individual model runs are labelled by four integers indexing the experiments in each category

e.g. ripf, where W, X, Y and Z are positive integers as defined below:

  • The first category, labelled realization_index (referred to with letter r), performs experiments which differ only in random perturbations of the initial conditions of the experiment. Comparing different realizations allow estimation of the internal variability of the model climate.
  • The second category, labelled initialization_index (referred to with letter i), refers to variation in initialisation parameters. Comparing differently initialised output provides an estimate of how sensitive the model is to initial conditions.
  • The third category, labelled physics_index (referred to with letter p), refers to variations in the way in which sub-grid scale processes are represented. Comparing different simulations in this category provides an estimate of the structural uncertainty associated with choices in the model design.
  • The fourth category labelled forcing_index (referred to with letter f) is used to distinguish runs of a single CMIP6 experiment, but with different forcings applied.


Many different climate variables (e.g., sea surface temperature, near surface air temperature, rainfall) are modeled for the CMIP6 project.

The climate variables often go by a short identifier. For example, to search for sea surface temperature data, you will often need to use it’s shortname which is “tos”. You can learn more about some of the common CMIP6 climate variables and their shortname (i.e., ESGF Variable ID) in the below table.

Click the triangle to see a long list of CMIP6 climate variables

Some common CMIP6 climate variables (from this incredibly helpful resource):

Parameter name ESGF variable id CDS parameter name for CMIP5 Units
Near-surface air temperature tas 2m temperature Kelvin
Daily maximum near-surface air temperature tasmax Maximum 2m temperature in the last 24 hours Kelvin
Daily minimum near-surface air temperature tasmin Maximum 2m temperature in the last 24 hours Kelvin
Surface temperature ts Skin temperature Kelvin
Sea level pressure psl Mean sea level pressure Pa
Surface air pressure ps Surface pressure Pa
Eastward near-surface wind uas 10m u component of wind m s-1
Northward near-surface wind vas 10m v component of wind m s-1
Near-surface wind speed sfcWind 10m wind speed m s-1
Near-surface relative humidity hurs 2m relative humidity 1
Near-surface specific humidity huss 2m specific humidity 1
Precipitation pr Mean precipitation flux kg m-2 s-1
Snowfall flux prsn Snowfall kg m-2 s-1
Evaporation Including sublimation and transpiration evspsbl Evaporation kg m-2 s-1
Surface downward eastward wind stress tauu Eastward turbulent surface stress Pa
Surface downward northward wind stress tauv Northward turbulent surface stress Pa
Surface upward latent heat flux hfls Surface latent heat flux W m-2
Surface upward sensible heat flux hfss Surface sensible heat flux W m-2
Surface downwelling longwave radiation rlds Surface thermal radiation downwards W m-2
Surface upwelling longwave radiation rlus Surface upwelling longwave radiation W m-2
Surface downwelling shortwave radiation rsds Surface solar radiation downwards W m-2
Surface upwelling shortwave radiation rsus Surface upwelling shortwave radiation W m-2
TOA incident shortwave radiation rsdt TOA incident solar radiation W m-2
TOA outgoing shortwave radiation rsut TOA outgoing shortwave radiation W m-2
TOA outgoing longwave radiation rlut TOA outgoing longwave radiation W m-2
Total cloud cover percentage clt Total cloud cover %
Air temperature ta Air temperature K
Eastward wind ua U-component of wind m s-1
Northward wind va V-component of wind m s-1
Relative humidity hur Relative humidity 1
Specific humidity hus Specific humidity 1
Geopotential height zg Geopotential height m
Surface snow amount snw Surface snow amount kg m-2
Snow depth snd Snow depth m
Total runoff mrro Runoff kg m-2 s-1
Moisture in upper portion of soil column mrsos Soil moisture content kg m-2
Sea-Ice area percentage (ocean grid) siconc Sea-ice area percentage 1
Sea Ice thickness sithick Sea ice thickness m
Sea-Ice mass per area simass Sea ice plus snow amount kg m-2
Surface temperature of sea Ice sitemptop Sea ice surface temperature K
Sea surface temperature tos Sea surface temperature K
Sea surface salinity sos Sea surface salinity PSU
Sea surface height above geoid zos Sea surface height above geoid m
Grid-cell area for ocean variables* areacello NOT AVAILABLE m2
Sea floor depth below geoid* deptho NOT AVAILABLE m
Sea area percentage* sftof NOT AVAILABLE %
Grid-cell area for atmospheric grid variables* areacella NOT AVAILABLE m2
Capacity of soil to store water (field capacity)* mrsofc NOT AVAILABLE kg m-2
Percentage of the grid cell occupied by land (including lakes)* sftlf NOT AVAILABLE %
Land ice area percentage* sftgif NOT AVAILABLE 1
Surface altitude* orog NOT AVAILABLE m

Resources for derived products

We’ve said this before, but we’ll say it again: If you are an ecologist or biologist you should, if possible, avoid the raw climate data and use more derived products that others have created.

Using derived products can provide many advantages:

Derived data products are generally easier to work with because the raw ESGF has many idiosyncrasy’s. One example: data are often reported using an irregular grid that must be interpolated to a regular grid.

You will be able to harness the power of the ensemble! Most derived CMIP6 data products are “ensembles” that combine a subset of the CMIP6 climate models into a single dataset. In most cases, ensembles are more accurate than any given single climate model.

Much of the raw CMIP data is very coarse resolution (e.g., 1 to 1.25 degrees). You may be able to find downscaled climate data which has translated coarse resolution spatial information into a finer spatial resolution. This is often better suited to the scale we want to work at.

Of course, it is possible to download raw data and generate ensembles and downscaled data, however, avoiding this will save lots of time and computational energy. Derived datasets are generally created by experts who have spent a good deal of time thinking about how to deal with the idiosyncrasies of CMIP6 data, selecting and combining the climate models, and correcting for bias.

But: if you must work with the raw data. Here are some resources that look useful:

Global CMIP6 derived models that we are using

Source Variables Scenarios Time Frame Ensemble Details Resolution Instructions
1 land and ocean: SST, sea level pressure, surface pressure, air temperature, zonal wind, meridional wind, relative humidity, geopotential height, soil moisture, soil temperature SSP245, SSP585 1979-2100, 6h intervals bias corrected, 18 cmip6 models 1.25 degrees
2 human population SSP1_2.6, SSP2_3.4, SSP3_7.0, SSP4_3.4, SSP5_8.5 2010-2100, yearly modelled to be consistent with both the CMIP6 RCP-specific urban fraction dataset (LUH2-v2f, luh.umd.edu) and the country level SSP population and urban fraction scenarios from the SSP database 30 arcseconds (~1km equator) ftp: sidekick.nateko.lu.se
3 land and ocean: Mean/min/max air temp, extreme air temp variables, SST, sea ice conc, SLR, pH, surface ozone, surface PM2.5, population density SSP1-2.6, SSP-4.5, SSP3-7.0, SSP5-8.5 near term (2021-2040), medium term (2041-2060), long term (2081-2100), historical ensemble model means (vary for climate variables) 1 degree
4 land only: monthly avg. min. temp, monthly avg. max temp, monthly total ppt, range of bioclimatic variables ssp126, ssp245, ssp370, ssp585 2021-2040, 2041-2060, 2061-2080, 2081-2100 downscaled data for N=23 climate models (not ensemble averages) 10 min, 5 min, 2.5 min, 30 sec access through R library sdmpredictors
5 ocean (mostly) and land: depends on scenario, temperature, ppt, wind, sst (and at different depths), salinity, oxygen, NPP ssp126, ssp245, ssp370, ssp585 30 year periods: 1955-1984, 1965-1994, 1975-2004, 1985-2014; 2020-2049, 2030-2059, 2040-2069, 2050-2079, 2060-2089, 2070-2099 data provided for many ensembles or averaged (not bias corrected) 1 degree
6 mostly coastal: sea level change variables ssp5-8.5 historical and 2020-2050 ensemble coastal = 0.1 degrees see R script;https://cds.climate.copernicus.eu/cdsapp#!/dataset/sis-water-level-change-indicators-cmip6?tab=overview
7 land/river: nitrogen and phosphorous loading and export SSP1 - SSP5 1970-2070 (5 year periods) modeled based on several assumptions about landuse, agriculture, wastewater, etc. 0.5 degree https://dataportaal.pbl.nl/downloads/IMAGE/GNM/
8 land: near-surface relative humidity (hurs), near-surface specific humidity (huss), daily mean precipitation rate (pr), surface downwelling longwave radiation (rlds), surface downwelling shortwave radiation (rsds), daily-mean surface wind speed (sfcWind), daily maximum near-surface air temperature (tasmax), daily minimum near-surface air temperature (tasmin), and daily mean near-surface air temperature (tas) SSP245, SSP585 monthly, 1950-2100 downscaled, bias corrected ensemble with 35 GCM models 0.25 degree https://nex-gddp-cmip6-cog.s3.us-west-2.amazonaws.com/index.html#monthly/

**1. Xu et al. 2021 2. Olen and Lehsten 2022 3. IPCC WGI Interactive Atlas 4. WorldClim 5. NOAA’s Climate Change Web Portal: CMIP6 6. Muis et al. 2022 7. Beusen et al. 2022 8. Thrasher et al. 2022

United States CMIP6 derived models that we have explored

Source Variables Scenarios Time Frame Ensemble Details Resolution Instructions
1 land: Two sets of variables are available for download. One consists of 33 biologically relevant variables, including seasonal and annual means, extremes, growing and chilling degree days, snow fall, potential evapotranspiration, and a number of drought indices. The second dataset consists of 48 monthly temperature and precipitation variables. See the metadata for complete description of variables. CMIP6: SSP1_2.6, SSP2_4.5, SSP3_7.0, SSP5_8.5 yearly and monthly variables averaged for 20 or 30 year periods The ensemble mean projections are calculated from the 8-model ensemble and a larger 13-model ensemble (see dendrogram below), the latter being representative of the full unconstrained CMIP6 ensemble (Mahony et al. 2022). Consistent with the guidance provided by Hausfather et al. (2022), users wishing to use an ensemble mean projection are encouraged to use the 8-model rather than the 13-model ensemble mean. Lambert Azimuthal Equal-Area projection, at 1km resolution These are derived from Mahoney et al. 2022
2 land: humidity, precipitation, min/max air temp, solar radiation, vapor pressure deficit, wind speed CMIP5: RCP4.5, RCP8.5 monthly and daily 20 climate models provided (no ensemble) ~4 or 6km can obtain and combine climate models using R packages climate futures toolbox (CFT) (https://github.com/earthlab/cft), or climateR (https://mikejohnson51.github.io/climateR-intro/#123)
3 land: Annual: MAT, MWMT, MCMT, TD, AHM, SHM, EMT, EXT and MAR; Seasonal: Tmax, Tmin, Tave and Rad; Monthly: Tmax, Tmin, Tave and Rad. SSP126, SSP245, SSP370, SSP585, SSP245, SSP370, SSP585 annual, 20-year, 30-year periods ensemble of 13 CMIP6 models ~4km for NA; ~800 m for Western North America https://climatena.ca/spatialData

**1. AdaptWest Project 2022 2. MACA Data Portal 3. Mahoney et al. 2022


This post was created as part of the MARISCO project which is funded by the Belmont Forum.

Belmont Forum members and partner organizations work together to direct and fund research on environmental change that affect us all.