Research Features

GISS and NCCS Contribute to CMIP6 International Climate Model Intercomparison Project

The NASA Goddard Institute for Space Studies contributed results from hundreds of the institute's historical and future climate simulations to the Coupled Model Intercomparison Project Phase 6 ( CMIP6. This international effort involves more than 110 climate modeling groups (see world map above). CMIP6 results serve as key inputs to the Intergovernmental Panel on Climate Change's Sixth Assessment Report due for release in 2021-2022.

Global map of contributors to CMIP6

The GISS CMIP6 simulations ran on the NASA Center for Climate Simulation (NCCS) Discover supercomputer, where they consumed approximately 104.5 million core hours. That much computing is like running a four-core laptop for nearly 3,000 years. The NCCS DataPortal and Earth System Grid Federation (ESGF) currently host 147.2 terabytes of data from 84,445 GISS simulation datasets for access by climate researchers and policymakers. 

“This work would be impossible without NCCS resources,” said Gavin Schmidt, acting Senior Advisor on Climate to the NASA Administrator and Principal Investigator for the GISS ModelE Earth System Model.

Impact: The model simulations GISS contributed to CMIP6 play an essential role in attributing climate changes in the historical period, understanding why climate is changing now, and providing projections of future climate change. Evaluation of these simulations against observed climate change provides confidence for policymakers working on climate, air quality, water, and coastal issues.

For CMIP6, GISS used two major versions of their ModelE global atmospheric model coupled to two distinct global ocean models in various combinations. Each model version has particular strengths in representing Earth system processes. GISS CMIP6 models include:

Model Grid Resolution Strengths
ModelE 2.1 Atmosphere 250 kilometers (km), 40 layers
  • Improved variability and climatology
  • Composition modeling (aerosols and short-lived gases)
  • Carbon Cycle (CC version)
ModelE 2.2 Atmosphere 250 km, 102 layers
  • Stratospheric dynamics
  • Quasi-Biennial Oscillation (global directional switch of tropical stratospheric wind circulation)
  • Stratosphere-troposphere exchanges
GISS Ocean 100 km, 40 layers
  • Oceanic bio-geochemistry (CC version)
HYCOM Ocean 100 km, 32 layers
  • Oceanic bio-geochemistry (CC version)

Using these models, GISS researchers ran several hundred paleoclimate (distant past), historical (starting in 1850), carbon cycle, and future simulations. GISS CMIP6 team papers in the Journal of Advances in Modeling Earth Systems and the Journal of Geophysical Research cover CMIP6 simulation results, model development, and model versions.

The graph below displays surface temperature anomalies (changes compared to the 1880–1899 baseline) from both historical (1850–2014) and future (2015–2100) GISS CMIP6 simulations:

Line plot of global mean surface temperature amomaly, with several alternatives (pathways) for future years

Global mean surface temperature anomalies from GISS model simulations for 1850–2014 driven by observed emissions and greenhouse gas concentrations and for 2015–2100 driven by four Shared Socioeconomic Pathways (SSPs).

Climate modelers compare historical simulation results to past observational records to verify model accuracy and give confidence in their projections of future climate changes. In the GISS CMIP6 simulations, historical drivers of emissions and greenhouse gas concentrations produce responses (purple line) that match well to GISTEMP surface temperature observations (black line) going back to 1880.

All CMIP6 future simulations are driven by a set of Shared Socioeconomic Pathways (SSPs), which represent a wide range of future climate outcomes based on differing emission, mitigation, and adaption scenarios and other factors. A shorthand for understanding the SSP outcomes are the global concentrations of carbon dioxide — a leading greenhouse gas — that each pathway is envisioned to produce by the year 2100. Compared to the March 2021 monthly average of 417.64 parts per million, the four SSPs shown in the graph would yield approximately the following:

Shared Socioeconomic Pathway (SSP) Description Global Carbon Dioxide Concentrations in 2100 Parts per Million
SSP1-2.6 Sustainability — Taking the Green Road (low challenges to mitigation and adaptation) 440
SSP2-4.5 Middle of the Road (medium challenges to mitigation and adaptation) 600
SSP3-7.0 Regional Rivalry — A Rocky Road (high challenges to mitigation and adaptation) 890
SSP5-8.5 Fossil-fueled Development — Taking the Highway (high challenges to mitigation, low challenges to adaptation) 1,180

Across the SSP-driven GISS future simulations, global mean surface temperature anomalies range from roughly 2°C (3.6°F) for SSP1-2.6 to nearly 5°C (9°F) for SSP5-8.5 by 2100. For reference, the global temperature change between the peak of the most recent ice age 20,000 years ago and the pre-industrial climate was around 9°F.

CMIP6 adds GISS's results to those from more than 60 other climate models, each having their own strengths and weaknesses. “If results from across the model ensemble are similar, that suggests the results are not related to any individual model feature (or bug),” Schmidt said. “However, in all cases, the model projections need to be evaluated against relevant observational data.”

GISS is currently running more CMIP6 simulations using its newest model — ModelE 3 — with 111 km grid resolution and 106 levels. ModelE 3 incorporates a cubed-sphere atmospheric dynamical core and grid as well as new “moist physics,” which includes updated cloud and aerosol microphysics. GISS runs ModelE 3 in the same coupled atmosphere-ocean configurations as ModelE 2.1 and 2.2.


+ The NASA GISS Contributions to CMIP6. Multiple articles in J. Adv. Model. Earth Syst. and J. Goephys. Res. Atmos.

This article was originally prepared as an NCCS Highlight for the NASA Center for Climate Simulations.