Research method
Coupled Model Intercomparison (CMIP)
20 studies
CMIP is a coordinated archive of coupled atmosphere–ocean climate models (GCMs/ESMs) run under shared historical and scenario protocols — CMIP5 RCPs and CMIP6 SSPs, plus endorsed MIPs such as HighResMIP. A CMIP result is usually a multimodel mean and a spread, not a single GCM. Papers in this library also borrow CMIP warming deltas, NEX-GDDP downscaled members, or one-model large ensembles checked against a few CMIP runs; those are uses of the archive, not ‘the CMIP model’.
Climate papers reach for CMIP when they need a comparable set of 21st-century extremes, sea-level sensitivity, or regional rain under a chosen warming level. It answers ‘does this response survive other models and scenarios?’ Its main limitation is that the ensemble mean is not an observation: CESM2 under SSP3-7.0 can sit on the hot end, ice emulators inherit ISMIP6’s weak marine dynamics, and a lexicon hit can tag a methane inversion or a nested 2.2 km CPM as ‘CMIP’ because a GCM supplied the large-scale climate.
Studies
20
Findings
6
7 supporting · 0 challenging · 0 qualifying citations
Open tensions
2
Currently
What we know
- Multimodel-mean regional sensitivity of yearly hottest days, coldest nights, and heaviest 1-day rain (ETCCDI TXx, TNn, Rx1day) is very similar in CMIP5 and CMIP6 at +1.5, +2 and +4 °C. At +1.5 °C, uncertainty in regional transient sensitivity often exceeds uncertainty from global climate sensitivity, especially in CMIP6. Local differences include hotter SAM extremes and wetter West Africa/Sahel in CMIP6. The CMIP6 archive was still incomplete in late March 2020.
- Weighted CMIP6 plus ice emulators give steric transient sea-level sensitivity of 1.5±0.2 mm/yr/K (observations 1.4±0.5) and Greenland 0.8±0.2; East Antarctic ice-sheet TSLS is slightly negative (−0.1±0.2). Full GMSL TSLS is 5.3±1.0 mm/yr/K for 2016–2050 versus 3.0±0.4 after 2050. Historical all-but-glacier sensitivity (3.1±0.4) sits ~30% above the models; observed Antarctic dynamics are +0.4±0.2 mm/yr/K against near-zero modelled WAIS/EAIS dynamics.
- A 100-member CESM2 large ensemble under SSP3-7.0, checked with other CMIP6 ensembles, finds that record-shattering 5-year block-maximum daily rain (beating each member’s own record by ≥1 1850–1949 standard deviation) becomes more likely over most land. In some tropics the late-21st-century probability ratio reaches 15. Variability (σ) changes dominate CESM2-LE in most regions, while other CMIP6 models split mean and variability more evenly. A CESM2 analogue resembles the 2021 western European floods, which exceeded a 1000-year return level in observations.
- NEX-GDDP CMIP6 rain, quantile-mapped and fed to SWAT+, turns a historical 50- and 200-year Nile peak of 13,400 and 14,900 m³/s into 21,800 and 24,100 m³/s (median SSP2-4.5) or >24,700 and >27,400 m³/s (SSP5-8.5). Extreme streamflow intensity rises 49–63% and 73–85%; a historical 100-year flood (~14,200 m³/s) occurs about every 4 years or 2.75 years. Mean AMS rain rises ~29% and ~41%. CMIP6 still disagrees on the sign of mean Nile rain.
- CMIP is also a catalogue of process fields, not only of warming. Across 26 Indian monsoon extremes, the dynamic (vertical-velocity) moisture-budget term contributed more than 90% on average; six CMIP6 models with omega/humidity agreed that dynamics lead thermodynamics. Twenty-seven CMIP6 models underestimate North American oceanic cyclone depth and overestimate it over land versus ERA5; 12 km CRCM6 has the smallest domain errors even when driven by different CMIP6 models.
Largest unresolved question
Single-model large ensembles and the CMIP multimodel archive disagree about where uncertainty lives. CESM2-LE attributes most extra record-shattering rain to variability changes; other CMIP6 models are more balanced between mean and σ, and CESM2’s high climate sensitivity plus SSP3-7.0 make the probability ratio of 15 an upper-end figure. The CMIP5/6 extremes paper finds similar regional-mean sensitivity across generations but, at +1.5 °C, larger regional- than global-sensitivity uncertainty. Opening CESM2-LE is not the same as opening CMIP6.
Procedure
How it works
The main stages of the method. Specific protocols and analysis choices vary between studies.
Run coupled models under shared protocols
Atmosphere–ocean GCMs and Earth-system models are integrated with agreed historical forcings and future scenarios (CMIP5 RCPs, CMIP6 SSPs, plus endorsed MIPs such as HighResMIP). A single lab's large ensemble is still one model.
Archive comparable fields
Standardised output — temperature, rain, sea level, extremes — is deposited so groups can compute the same diagnostics. Using a CMIP warming delta, a downscaled member, or a nested convection-permitting run is a use of the archive, not 'the CMIP model.'
Summarise as a multimodel mean and a spread
Papers usually report an ensemble mean and a range across models or members. The mean is an average of simulations. It is not an observation, and one hot model under a high-forcing SSP can pull the central estimate.
Ask whether a response survives other models
The scientific use is robustness: does a regional extreme, a sea-level sensitivity, or a rain change appear across models and scenarios? Agreement among 100 members of one GCM samples weather, not structural diversity.
Do not read the ensemble as the climate
CMIP answers whether this modelled response survives other models and scenarios. Historical sensitivity can sit above the models, reanalysis is an imperfect referee, and upgrading the archive version does not automatically rewrite every regional extreme.
Common misconceptions
The CMIP ensemble mean is an observation of the climate.
It is an average of models. Cyclone ‘errors’ versus ERA5 are differences from an imperfect reanalysis; historical sea-level sensitivity sits ~30% above the models; CESM2’s record-shattering ratios may be upper-end because of high climate sensitivity and SSP3-7.0.
CMIP6 is one GCM, so upgrading from CMIP5 automatically changes regional extremes.
CMIP is a multi-model archive. Regional-mean TXx/TNn/Rx1day sensitivity is very similar in CMIP5 and CMIP6; what often grows at +1.5 °C is the spread in regional transient sensitivity, especially in CMIP6. Local SAM and Sahel differences are exceptions, not a wholesale rewrite.
If 100 CESM2 members agree, that is the CMIP6 result.
A large ensemble from one model samples initial-condition weather, not structural model diversity. Other CMIP6 ensembles do not all assign the same share of record-shattering rain to variability, and an eight-model CMIP6 mean still leaves Sahel internal variability alive.
Related
Claim ledger
What the evidence shows
Drawn from 20 studies in this library. Mix labels say which citation roles are present; they are not a strength score. Supports means evidence for a finding; Challenges means evidence against a stated position; Qualifies marks scope.
Multimodel-mean regional sensitivity of yearly hottest days, coldest nights, and heaviest 1-day rain (ETCCDI TXx, TNn, Rx1day) is very similar in CMIP5 and CMIP6 at +1.5, +2 and +4 °C. At +1.5 °C, uncertainty in regional transient sensitivity often exceeds uncertainty from global climate sensitivity, especially in CMIP6. Local differences include hotter SAM extremes and wetter West Africa/Sahel in CMIP6. The CMIP6 archive was still incomplete in late March 2020.
Weighted CMIP6 plus ice emulators give steric transient sea-level sensitivity of 1.5±0.2 mm/yr/K (observations 1.4±0.5) and Greenland 0.8±0.2; East Antarctic ice-sheet TSLS is slightly negative (−0.1±0.2). Full GMSL TSLS is 5.3±1.0 mm/yr/K for 2016–2050 versus 3.0±0.4 after 2050. Historical all-but-glacier sensitivity (3.1±0.4) sits ~30% above the models; observed Antarctic dynamics are +0.4±0.2 mm/yr/K against near-zero modelled WAIS/EAIS dynamics.
A 100-member CESM2 large ensemble under SSP3-7.0, checked with other CMIP6 ensembles, finds that record-shattering 5-year block-maximum daily rain (beating each member’s own record by ≥1 1850–1949 standard deviation) becomes more likely over most land. In some tropics the late-21st-century probability ratio reaches 15. Variability (σ) changes dominate CESM2-LE in most regions, while other CMIP6 models split mean and variability more evenly. A CESM2 analogue resembles the 2021 western European floods, which exceeded a 1000-year return level in observations.
NEX-GDDP CMIP6 rain, quantile-mapped and fed to SWAT+, turns a historical 50- and 200-year Nile peak of 13,400 and 14,900 m³/s into 21,800 and 24,100 m³/s (median SSP2-4.5) or >24,700 and >27,400 m³/s (SSP5-8.5). Extreme streamflow intensity rises 49–63% and 73–85%; a historical 100-year flood (~14,200 m³/s) occurs about every 4 years or 2.75 years. Mean AMS rain rises ~29% and ~41%. CMIP6 still disagrees on the sign of mean Nile rain.
CMIP is also a catalogue of process fields, not only of warming. Across 26 Indian monsoon extremes, the dynamic (vertical-velocity) moisture-budget term contributed more than 90% on average; six CMIP6 models with omega/humidity agreed that dynamics lead thermodynamics. Twenty-seven CMIP6 models underestimate North American oceanic cyclone depth and overestimate it over land versus ERA5; 12 km CRCM6 has the smallest domain errors even when driven by different CMIP6 models.
- Indian downpours are driven more by uplift than extra humidity
- How well CMIP6 captures North American storm pulses
Study Role Design N Population Outcome Indian downpours are driven more by uplift than extra humidity Supports Computational / modellingMoisture-budget decomposition of Indian monsoon extremes in reanalyses and CMIP6 N=26 · 26 IMD extremes (2005–2018) plus Kerala 2018 and Uttarakhand 2013 case studies Precipitation extremes over India during the monsoon Large-scale dynamics (upward motion) contribute more than Clausius–Clapeyron humidity increases to Indian precipitation extremes How well CMIP6 captures North American storm pulses Supports Computational / modellingEulerian intense-cyclone metrics comparing ERA5, 27 CMIP6 GCMs, and CRCM6 N=27 · 27 CMIP6 simulations plus 5 CRCM6-GEM5 runs evaluated against ERA5 (1980–2014) Intense low-pressure systems over North America in reanalysis and climate models Biases in cyclone depth/tendency and skill of 12 km CRCM6 vs GCMs Fixing aerosols in CESM1, then comparing CanESM2-LE and an eight-model CMIP6 mean with HadISST/NOAA SST and CRU rain, shows that the 1925–1955 wet, 1955–1985 dry, then wet Sahel/NASST swing vanishes without evolving aerosols. NASST correlates with net surface energy (r = 0.90) and sulfate burden (r = −0.82). The CMIP6 mean does not fully kill internal variability; aerosol–cloud physics remains the main spread source.
Debates
Tensions and limits
Some items are genuine disagreements on the same question. Others mark different assays, populations, or outcomes.
Single-model large ensembles and the CMIP multimodel archive disagree about where uncertainty lives. CESM2-LE attributes most extra record-shattering rain to variability changes; other CMIP6 models are more balanced between mean and σ, and CESM2’s high climate sensitivity plus SSP3-7.0 make the probability ratio of 15 an upper-end figure. The CMIP5/6 extremes paper finds similar regional-mean sensitivity across generations but, at +1.5 °C, larger regional- than global-sensitivity uncertainty. Opening CESM2-LE is not the same as opening CMIP6.
Single-model large ensembles and the CMIP multimodel archive disagree about where uncertainty lives. CESM2-LE attributes most extra record-shattering rain to variability changes; other CMIP6 models are more balanced between mean and σ, and CESM2’s high climate sensitivity plus SSP3-7.0 make the probability ratio of 15 an upper-end figure. The CMIP5/6 extremes paper finds similar regional-mean sensitivity across generations but, at +1.5 °C, larger regional- than global-sensitivity uncertainty. Opening CESM2-LE is not the same as opening CMIP6.
- Wilder rainfall swings make record storms more likely
- Regional extreme sensitivity similar in CMIP5 and 6
Study Role Design N Population Outcome Wilder rainfall swings make record storms more likely Supports Computational / modelling100-member CESM2-LE analysis of record-shattering daily extremes under SSP3-7.0 N=100 · 100-member CESM2 large ensemble (other CMIP6 ensembles as checks) Land daily precipitation extremes in CESM2 large-ensemble climate projections Record-shattering extreme-rain probability driven by growing variability vs mean shifts Regional extreme sensitivity similar in CMIP5 and 6 Supports Computational / modellingCMIP5 vs CMIP6 ETCCDI extremes scaled to shared global-warming levels Multimodel CMIP5/CMIP6 ensembles at +1.5/+2/+4 °C; ensemble count not a single primary N Regional climate extremes (TXx, TNn, Rx1day) across AR6 regions Similarity of regional extreme sensitivity in CMIP6 vs CMIP5 despite ECS differences
PaperFren reads this as a limit on how far one study travels — different assays, populations, or outcomes — not a forced fight between papers.
CMIP can match one budget and miss another. Steric TSLS (1.5±0.2) sits on observations (1.4±0.5), yet historical all-but-glacier sea-level sensitivity is ~30% above the models and Antarctic dynamics are observed, not modelled. Downstream, QQM-corrected NEX-GDDP members intensify Nile flood peaks while the same CMIP6 generation still disagrees on the sign of mean Nile rain. Agreement of a multimodel mean on ETCCDI sensitivity does not license treating every CMIP-forced hydrology run as an observation.
CMIP can match one budget and miss another. Steric TSLS (1.5±0.2) sits on observations (1.4±0.5), yet historical all-but-glacier sea-level sensitivity is ~30% above the models and Antarctic dynamics are observed, not modelled. Downstream, QQM-corrected NEX-GDDP members intensify Nile flood peaks while the same CMIP6 generation still disagrees on the sign of mean Nile rain. Agreement of a multimodel mean on ETCCDI sensitivity does not license treating every CMIP-forced hydrology run as an observation.
- CMIP6 sea-level sensitivity lags observations
- Nile flood peaks may jump this century
- Regional extreme sensitivity similar in CMIP5 and 6
Study Role Design N Population Outcome CMIP6 sea-level sensitivity lags observations Supports Computational / modellingTransient sea-level sensitivity from weighted CMIP6/ISMIP6 components vs AR6 historical rates Multimodel CMIP6 and ice emulators; not a single sample N Global mean and component sea-level contributions under transient warming Model TSLS ~30% below historical all-but-glacier rate, mainly from weak Antarctic response Nile flood peaks may jump this century Supports Computational / modellingSWAT+ Nile hydrology forced by bias-corrected CMIP6 members under SSP2-4.5 and SSP5-8.5 N=30 · 30 CMIP6 climates (high/median/low members per SSP run through SWAT+) Nile basin streamflow regimes at Dongola under 21st-century climate scenarios Projected increases in 50- and 200-year flood peaks by 2100 Regional extreme sensitivity similar in CMIP5 and 6 Supports Computational / modellingCMIP5 vs CMIP6 ETCCDI extremes scaled to shared global-warming levels Multimodel CMIP5/CMIP6 ensembles at +1.5/+2/+4 °C; ensemble count not a single primary N Regional climate extremes (TXx, TNn, Rx1day) across AR6 regions Similarity of regional extreme sensitivity in CMIP6 vs CMIP5 despite ECS differences
PaperFren reads this as a limit on how far one study travels — different assays, populations, or outcomes — not a forced fight between papers.
Timeline
How understanding moved
Study years are when the paper was published. Evidence edits are dated changes to this page's claims. Explanations are when PaperFren added a Discovery — not a claim that the science happened that day.
2026
- Clouds amplified the 2022 Antarctic heatwave
- How well CMIP6 captures North American storm pulses
Concept page published
Coupled Model Intercomparison (CMIP)
Change log
What changed
Dated edits to this page's evidence: studies added or removed from a claim, claims added or withdrawn, and new explanations tagged here. Coverage edits are distinct from reviewed research changes and corrections. Dates record PaperFren updates; study publication dates appear in the timeline.
- Coverage updated · Concept page published
Papers
20 studies in this library bear on Coupled Model Intercomparison (CMIP), ordered by citations.
- Regional extreme sensitivity similar in CMIP5 and 6
CMIP6 and CMIP5 agree on how hot days, cold nights, and heavy rain scale with global warming, even though global climate sensitivity differs.
- Compound coastal storms become more common by 2100
Heavy rain and storm-driven high seas already coincide far more than by chance, and that pairing is projected to increase along most coasts.
- Europe’s first continent-wide 2.2 km rain projections
A 2.2 km convection-permitting model projects wetter northern winters and drier summers, with summer hourly extremes unlike coarser GCMs.
- Wilder rainfall swings make record storms more likely
Future record-shattering daily downpours become much more probable because extreme-rain variability grows, not only because the mean inches up.
- South Pacific SSTs, not India, built Brazil’s 2014 drought
Isca experiments show 2013/14 South Pacific and South Atlantic SST anomalies favored the blocking high that dried SE Brazil; Indian Ocean SSTs did not.
- Future Floyd–Florence rains could jump sharply
A design-rainfall method applied to three historic tropical cyclones projects large late-century increases in eastern North Carolina extreme rain.
- How Bølling warming and ice saddles fed MWP1a
North American ice-sheet ensembles show saddle collapse and Bølling warming can supply several meters of Meltwater Pulse 1a in 340 years.
- How North Atlantic aerosols steered Sahel rain
Fixing aerosols at 1920 levels removes the observed multidecadal Sahel drought–recovery cycle that tracks North Atlantic SST.
Show 12 more studiesShow fewer studies
- Indian downpours are driven more by uplift than extra humidity
For monsoon extremes over India, stronger upward motion supplies most of the extra moisture, not the Clausius–Clapeyron humidity increase alone.
- Why LGM methane needs a bigger wetland collapse
Ice-core CH4 and isotopes imply about a 50% source drop at the LGM; process models underpredict both concentration and isotopic shifts.
- US 1-in-1000-year rainstorms become much more common
Today’s thousand-year three-day downpours over the United States could strike several times more often at 2–4 °C of global warming.
- CMIP6 sea-level sensitivity lags observations
CMIP6/ISMIP6 transient sea-level sensitivity is ~30% below the historical “all but glaciers” rate, mainly because Antarctic ice-sheet models barely respond to warming.
- AMOC collapse adds only a few ppm of CO2
Forcing CESM2’s AMOC off with North Atlantic freshwater raises 2100 atmospheric CO2 by only 2.6–4.2 ppm under low and high emissions.
- Nile flood peaks may jump this century
A basin-wide hydrology model driven by 30 CMIP6 climates projects much larger 50- and 200-year Nile floods at Dongola by 2100.
- Neural nets map CMIP6 weather into Greenland melt
ANNs trained on CESM2 translate CMIP6 atmospheres into Greenland surface-melt increases of about 79–264% by 2100 depending on SSP.
- Less Arctic sea ice both snows and melts Greenland
Losing Arctic sea ice in a +2 °C world adds winter snow to Greenland but more than offsets it with extra summer melt.
- Why stronger stratification is changing ocean tides
Model and satellite M2 trends since 1993 point to stronger ocean stratification increasing tidal conversion, not just sea-level rise.
- How well CMIP6 captures North American storm pulses
An Eulerian three-parameter storm metric shows CMIP6 GCMs share cyclone biases versus ERA5, while 12 km CRCM6 is usually closest.
- Finer climate models capture South Asian monsoon rain better
High-resolution HighResMIP runs reduce dry bias and better match monsoon timing and intensity in the Ganga–Brahmaputra–Meghna basin.
- Clouds amplified the 2022 Antarctic heatwave
Water-vapor and supercooled-cloud feedbacks added several degrees to the March 2022 East Antarctic heatwave compared with a preindustrial storyline.
Compare studies
Select 2–10 studies. Design and N are labels, not a ranking.
Nothing selected yet.
Questions
What is still open
Single-model large ensembles and the CMIP multimodel archive disagree about where uncertainty lives. CESM2-LE attributes most extra record-shattering rain to variability changes; other CMIP6 models are more balanced between mean and σ, and CESM2’s high climate sensitivity plus SSP3-7.0 make the probability ratio of 15 an upper-end figure. The CMIP5/6 extremes paper finds similar regional-mean sensitivity across generations but, at +1.5 °C, larger regional- than global-sensitivity uncertainty. Opening CESM2-LE is not the same as opening CMIP6.
CMIP can match one budget and miss another. Steric TSLS (1.5±0.2) sits on observations (1.4±0.5), yet historical all-but-glacier sea-level sensitivity is ~30% above the models and Antarctic dynamics are observed, not modelled. Downstream, QQM-corrected NEX-GDDP members intensify Nile flood peaks while the same CMIP6 generation still disagrees on the sign of mean Nile rain. Agreement of a multimodel mean on ETCCDI sensitivity does not license treating every CMIP-forced hydrology run as an observation.
Ask PaperFren about Coupled Model Intercomparison (CMIP)
Study this conceptflashcards and short-answer questions
Why is ‘CMIP6 looks like CMIP5 for regional extremes’ not the same claim as ‘uncertainty is small at +1.5 °C’?
The multimodel-mean regional sensitivity of TXx, TNn and Rx1day is similar across generations, but at +1.5 °C the spread in regional transient sensitivity often exceeds the spread from global climate sensitivity, especially in CMIP6. Agreement of means is compatible with large member-to-member regional spread, and the March 2020 CMIP6 archive was still incomplete.
Steric TSLS is 1.5±0.2 mm/yr/K against observations of 1.4±0.5, yet historical all-but-glacier sensitivity is ~30% above the models. What did CMIP6 plus ice emulators miss?
Antarctic ice-sheet dynamics: observations are +0.4±0.2 mm/yr/K while modelled WAIS/EAIS TSLS is near zero (EAIS −0.1±0.2). Emulators inherit ISMIP6’s weak marine-ice dynamics and present-day zero-trend design. Matching the steric term does not match the full GMSL budget (5.3±1.0 mm/yr/K for 2016–2050 versus 3.0±0.4 after 2050).
A student treats a tropical late-century probability ratio of 15 as ‘what CMIP6 says about record storms.’ What two design choices make that an upper-end CESM2-LE figure rather than the archive mean?
CESM2 has high climate sensitivity, and SSP3-7.0 is a high-warming path. Variability (σ) changes dominate CESM2-LE in most regions, while other CMIP6 models split mean and variability more evenly. The ratio is 5yRx1d record-shattering probability versus a stationary climate in that ensemble, not an observed frequency.
How can NEX-GDDP CMIP6 members raise a historical 100-year Nile flood (~14,200 m³/s) to a ~4-year event under SSP2-4.5 while CMIP6 still disagrees on the sign of mean Nile rain?
The hydrology paper bias-corrects precipitation with QQM, then runs high/median/low AMS members through SWAT+ and fits log-Pearson III peaks. That pipeline can intensify extremes even when raw GCMs disagree on the mean. The result is a bias-corrected, hydrology-filtered subset, not the native CMIP6 rainfall field, and GERD/Lake Nasser operations are not fully resolved.