Choosing SSP scenarios and climate models for local projections

When you request downscaled CMIP6 projections, two choices shape your results most: which future scenarios (SSPs) to look at, and which climate models to use. This guide explains both in plain language and suggests sensible starting points for NASA's NEX-GDDP-CMIP6 data.

Updated 16 November 2026

Part 1: Scenarios

Climate scenarios are "what if" stories about possible futures. They combine socioeconomic pathways (how society might develop) with different levels of greenhouse gas emissions, so you can explore and prepare for a range of possible futures.

Scenarios aren't predictions of what will happen. They explore what could happen under different assumptions about emissions, technology and policy.

What does "SSP2-4.5" mean?

The name has two parts:

  • SSPx (Shared Socioeconomic Pathway) is the societal storyline, from SSP1 to SSP5.
  • y.z is the climate "forcing" level reached by 2100, in watts per square meter.

Higher numbers mean more warming. SSP1-2.6 represents a sustainable world with strong climate policy, while SSP5-8.5 depicts fossil-fueled growth with minimal climate action.

Scenarios in NEX-GDDP-CMIP6

ScenarioWorld in 2100Warming by 2100*Often described as
Historical
1950–2014
What already happened+1.1°C so farOur current warming level
SSP1-2.6
2015–2100
Sustainable development, rapid decarbonization, net-zero by 2070≈ 1.8°CParis Agreement target (nearly met)
SSP2-4.5
2015–2100
"Middle of the road": some climate action but not transformative≈ 2.7°CCurrent policies trajectory
SSP3-7.0
2015–2100
Regional rivalry, less cooperation, continued fossil fuel use≈ 3.6°CHigh emissions
SSP5-8.5
2015–2100
Fossil-fueled economic growth, no climate policy≈ 4.4°CWorst case / no climate action

*Global mean temperature increase compared with pre-industrial (1850–1900) levels, based on IPCC AR6 central estimates.

What these warming levels mean

  • 1.5–2°C: more frequent heatwaves, some crop impacts, coral reef decline, sea level rise of about 0.5 m by 2100.
  • 3°C: substantial agricultural challenges, more extreme weather, significant sea level rise, some regions becoming difficult to inhabit.
  • 4°C and above: severe and widespread impacts, large uninhabitable zones, major ecosystem transformation, more than 1 m of sea level rise.

How to choose scenarios

  • Always include Historical. It is your baseline and shows the changes that have already happened.
  • Use a range of futures. Select at least one lower-emission scenario (SSP1-2.6 or SSP2-4.5) and one higher-emission scenario (SSP3-7.0 or SSP5-8.5) to see the range of possibilities.
  • Mind the time horizon. Scenarios look similar until about 2040, then diverge significantly in later decades.

Recommended starter set: Historical + SSP2-4.5 + SSP5-8.5 gives you the past baseline, a moderate future and a high-impact scenario. Each job on Climate Data Download includes up to two scenarios, and Historical counts as one, so run Historical + SSP2-4.5 as one job and SSP5-8.5 as a second.

Part 2: Climate models

Think of climate models as digital twins of Earth. Each one simulates the planet's climate system, but different research teams make slightly different choices about how to represent clouds, oceans and other processes. That diversity is useful: when many models agree on an outcome (such as hotter summers) you can be more confident in it, and when they disagree, the disagreement shows the uncertainty you should plan for.

No single model is "perfect". Even the best models have strengths and weaknesses, which is why climate scientists typically use several models rather than picking one.

Key terms

  • GCM (global climate model): a supercomputer program that simulates Earth's climate by solving equations for the atmosphere, oceans, land and ice on a 3D grid.
  • Ensemble mean: the average across several models. It smooths out individual model quirks and often gives the most reliable projection. Your files include it as the MULTI-MODEL-MEAN column.
  • Model spread: the range between the coolest and warmest model results, which gives a sense of the uncertainty in the projections.
  • Climate sensitivity (ECS): equilibrium climate sensitivity is the global temperature increase expected from a doubling of atmospheric CO₂.

Why climate sensitivity matters

CMIP6 models span a wide range of climate sensitivity. A few "hot" models, such as CanESM5 and UKESM1-0-LL, have an ECS above 5°C, while INM-CM4-8 and INM-CM5-0 are below 2°C. The IPCC's Sixth Assessment Report puts the likely range of ECS at 2.5–4°C (best estimate 3°C) and the very likely range at 2–5°C. That's why many studies either screen models by sensitivity or deliberately choose a set that spans the range.

How many models should I choose?

  • Quick exploration (3–4 models): pick a diverse mini-ensemble of one lower-sensitivity, one medium and one higher-sensitivity model. This gives a reasonable spread of possible futures without excessive processing time.
  • Comprehensive analysis (8 or more models): choose more models or a predefined group. This gives more statistical confidence and better represents scientific uncertainty, but takes longer to process and produces larger files.

A balanced starter set

These three models give a good cross-section of possible climate futures:

  • INM-CM5-0: a cooler-running model that projects less warming (Institute of Numerical Mathematics, Russia).
  • MPI-ESM1-2-LR: a middle-of-the-road model with moderate warming (Max Planck Institute, Germany).
  • CanESM5: a warmer-running model that projects more significant warming (Canadian Earth System Model).

If you're unsure, the request form offers a "Recommended model group" of nine models whose climate sensitivity sits mid-range: ACCESS-ESM1-5, BCC-CSM2-MR, FGOALS-g3, MIROC6, MPI-ESM1-2-HR, MPI-ESM1-2-LR, MRI-ESM2-0, NorESM2-LM and NorESM2-MM. The group follows the model screening used in the U.S. Geological Survey's National Climate Change Viewer, and excludes the hottest- and coolest-running models.

Which combinations are unavailable

A small number of model, variable and scenario combinations are not available in NASA's NEX-GDDP-CMIP6 dataset. If you select one, Climate Data Download skips it and that column won't appear in your files. Based on the tool's current configuration:

ModelNot available
BCC-CSM2-MRRelative humidity (hurs)
CESM2Max/min temperature (tasmax, tasmin)
CESM2-WACCMMax/min temperature (tasmax, tasmin); SSP1-2.6 and SSP3-7.0
CMCC-CM2-SR5Max/min temperature (tasmax, tasmin) for SSP1-2.6 and SSP3-7.0
GFDL-CM4, GFDL-CM4_gr2SSP1-2.6 and SSP3-7.0
HadGEM3-GC31-LLSSP3-7.0
HadGEM3-GC31-MMSSP2-4.5 and SSP3-7.0
IITM-ESMMax/min temperature (tasmax, tasmin)
IPSL-CM6A-LRSpecific humidity (huss)
KIOST-ESMSSP3-7.0
MIROC6Specific humidity (huss) for Historical, SSP2-4.5 and SSP5-8.5
NESM3Relative and specific humidity (hurs, huss); SSP3-7.0

Two of these affect the Recommended group: if you request humidity, BCC-CSM2-MR has no relative humidity and MIROC6 has no specific humidity for most scenarios. For background, see NASA's technical note on the dataset.

Putting it together

Quick look at a new location

  • Variables: temperature (tas) and precipitation (pr)
  • Models: INM-CM5-0, MPI-ESM1-2-LR and CanESM5
  • Scenarios: Historical + SSP2-4.5

Planning or impact study

  • Variables: up to four that match your question (see the variable guide)
  • Models: the Recommended group, or up to nine models that span low to high sensitivity
  • Scenarios: job 1 with Historical + SSP2-4.5, and job 2 with SSP5-8.5 (or SSP3-7.0)

Once you have the files, the Excel guide shows how to calculate change from the historical baseline and the range across models.

More guides