What's next — CMIP7 and CORDEX-Africa

What this page covers

The next generation of coordinated climate-model experiments is CMIP7 — the Coupled Model Intercomparison Project Phase 7 — which will feed IPCC AR7. CMIP6 is being slowly superseded, but slowly is the operative word: the forcings are finalised, model runs are underway, and the first outputs are landing on ESGF-NG (the next-generation Earth System Grid Federation, launched May 2026), but no CMIP7-derived downscaled product for African adaptation work exists yet.

By the end of this page you’ll know:

What’s actually changing from CMIP6 — the design, the experiments, the scenarios
Where CMIP7 is in June 2026, and the timeline you should plan against
When CMIP7-derived African projections will be usable in practice
What proposal authors and CGIAR partners should do in the meantime

For the operational guidance (“which dataset, which baseline, which scenario today”), see Recommendations for CGIAR. For the existing CMIP6 methodology this transition will replace, see African CMIP6 Ensembling.

What’s actually changing from CMIP6

CMIP7 isn’t a renaming of CMIP6 — it’s a structural redesign. Three things are materially different (Dunne et al. 2025, GMD; WCRP CMIP7 page):

The Assessment Fast Track (AFT). CMIP6 grew to ~22 endorsed Model Intercomparison Projects, each with their own experiment requests. The result was a sprawling catalogue and unsustainable compute and human cost at modelling centres. CMIP7 explicitly does not endorse MIPs; instead the CMIP Panel curates a compact “AFT” — an expanded DECK plus a small selection from community MIPs — designed to feed IPCC AR7 directly.
Continuous CMIP, not phased. The CMIP organisation is repositioning CMIP as a continuous activity punctuated by periodic “Fast Tracks” rather than a single big-bang phase every 7–8 years. The AFT is the first such Fast Track.
Expanded DECK and tighter sensitivity diagnostics. New experiments include esm-hist (ESM-only historical), several piClim configurations, and an extended 300-year abrupt-4xCO2 protocol for sharper equilibrium-climate-sensitivity estimates.

The design responds directly to feedback after CMIP6 about the unsustainable scale and complexity. The headline implication for partners: fewer experiments, more standardisation, faster turnaround into IPCC assessment.

The new scenarios — H/M/L instead of SSPs

The most visible change is the scenarios. CMIP6 used the SSP-RCP framework: scenario names had the form SSPx-y.z where the y.z was the target radiative forcing at 2100 in watts per square metre (e.g. SSP2-4.5 = “middle of the road, ~4.5 W/m² by 2100”). CMIP7 drops the W/m² suffix entirely. The new design uses seven scenarios labelled by emissions character (van Vuuren et al. 2026, GMD; WCRP Explainer):

Code	Name	Character
H	High	Plausibly possible given a rollback of current climate policies. ~3.5 °C global warming by 2100. 0.9 °C cooler than CMIP6 SSP5-8.5 — the fossil-fuel-maximalist counterfactual is effectively retired.
HL	High-to-Low	Overshoot pathway — high near-term, sharp mitigation later
M	Medium	Anchored to currently-implemented policy (frozen at 2025) — not to a forcing target. A methodological shift.
ML	Medium-to-Low	Overshoot, less aggressive than HL
L	Low	Roughly fills the SSP1-2.6 niche
LN	Low-to-Negative	Significant deliberate carbon-dioxide removal (CDR-heavy)
VL	Very Low	Overshoots 1.5 °C by 0.2–0.3 °C before returning to 1.5 °C by 2100

Three things in this design matter for partners:

No clean SSP-to-CMIP7 crosswalk exists. Indicative correspondences from the literature: VL ≈ SSP1-1.9, L ≈ SSP1-2.6 niche, M ≈ SSP2-4.5 outcomes (but built differently), H is well below SSP5-8.5. Don’t relabel — when you migrate to CMIP7 you migrate the science, not just the name.
The high end is reframed. CMIP7-H is “plausibly possible given policy rollback,” not a fossil-fuel-maximalist counterfactual. This is an explicit response to the SSP5-8.5 implausibility critique that gathered momentum through the late CMIP6 era.
Overshoot and net-negative pathways are first-class citizens. HL, ML, LN — mitigation is framed as including deliberate carbon removal, not just emissions reductions.

Where CMIP7 is right now (June 2026)

Status	Component
✅ Done	CMIP7 forcings finalised (12 DECK categories + ScenarioMIP), distributed via input4MIPs under CC-BY-4.0 / CC0 licensing
✅ Done	ESGF-NG (next-generation Earth System Grid Federation) live, launched May 2026
✅ Done	REF (Rapid Evaluation Framework) CMIP7-ready since CMIP 2026 Community Workshop in Kyoto, March 2026
🟡 Partial	First AFT model output trickling onto ESGF-NG (ACCESS-ESM1.6 confirmed submitting); bulk publication targeted for second half of 2026
⏳ Pending	AR7 WGI First Order Draft expert review, August–October 2026
⏳ Pending	First CMIP7-derived downscaled / bias-corrected products for adaptation work
⏳ Pending	African regional-evaluation literature on CMIP7 models — the basis for AR-style regional sub-ensembles
❓ Undecided	AR7 WGI Summary for Policymakers approval timeline (IPCC-63 did not reach consensus)

The forcings are stable and citable; the raw model output is in the early-publication phase where centres may retract and republish. Don’t ingest raw CMIP7 model output into operational pipelines yet — wait for the centres’ v1.0 publications and the ESGF errata service to stabilise.

The timeline you should plan against

When	What	What this means for you
Now → end-2026	CMIP7 raw model output trickling onto ESGF-NG; AR7 WGI FOD review	Stay on CMIP6 in production. No migration.
2027 Q1–Q2	First CMIP7-derived downscaled / bias-corrected products begin to appear; ISIMIP4 forcings publish first	Pilot ingest of ISIMIP4 if your work is impact-sector (crops, water, health)
2027 H2 / 2028	Broader downscaled product availability; first peer-reviewed regional-evaluation papers for CMIP7 over African sub-regions	Start adopting CMIP7-derived products for new analyses; CMIP6 still acceptable for proposals where partner cited it
2028+	CMIP7 viable for CGIAR partner-facing production work	Migrate the AAA Adaptation Atlas’s Future Projections layer to CMIP7-derived inputs once the regional-evaluation basis exists
TBD (post-2027)	AR7 WGI SPM approval, AR7 publication	Cite AR7 directly once published; the wiki will be revised

The takeaway: CMIP6 stays the operational backbone for 2026 and 2027, with a transition window through 2028. Proposal authors writing in 2026 should cite CMIP6 confidently; those writing in late 2027 onwards will increasingly cite CMIP7.

What CMIP7-derived African data looks like — the downscaling pathway

Raw CMIP7 model output is at ~100 km resolution — too coarse to use directly for sub-national African adaptation work. The pathway from raw CMIP7 to data a CGIAR proposal can cite runs through several downscaling and bias-correction products, each on its own timeline:

NASA NEX-GDDP-CMIP7 — the global statistical-downscaling product that has anchored the AAA Adaptation Atlas’s CMIP6 ensemble. No public release timeline for the CMIP7 version as of mid-2026; expect 2027 by analogy with the CMIP6 cycle.
CIL-GDPCIR-CMIP7 — Climate Impact Lab’s downscaled product. Implicit follow-up; no public timeline.
CHELSA-CMIP7 — Swiss WSL’s high-resolution (~1 km) statistical downscaling. Named explicitly in the CMIP7 Impacts & Adaptation Data Request; expected 2027.
ISIMIP4 — the impact-sector default. Forcings published May–June 2026; sector-model runs through 2026–2027. Firmest timeline; most likely first to ship. Especially relevant for agriculture, water, health work.
CORDEX-CMIP7 — dynamical downscaling at ~22 km via regional climate models. Africa is a flagship CORDEX domain. Formally chartered task force; protocol finalisation late 2025 / 2026; simulations through 2026–2028. Slower but irreplaceable for extremes and monsoon dynamics.
WorldClim Future CMIP7 — bioclimatic variables at 30 arc-sec. No public roadmap; expect 2027–2028.

For Africa specifically, regional centres (ICPAC, AGRHYMET, ACMAD, CSAG) are positioned as consumers of these external products — none has announced a CMIP7-tuned African downscaled product of its own as of mid-2026.

The East African Paradox in CMIP7

The persistent CMIP-era disagreement between observed long-rains drying and modelled wetting over the Greater Horn of Africa (East African Paradox) is on the CMIP7 community’s radar. The leading hypothesis is that the paradox is rooted in Pacific sea-surface-temperature biases that propagate through the Walker circulation to East African rainfall, and CMIP7 development has prioritised improving those biases (Park et al. 2023; Schwarzwald et al. 2024).

Whether CMIP7 actually resolves it remains an open question — the first round of CMIP7-vs-observations evaluation papers for the Greater Horn won’t appear until 2027+. Until then, the existing paradox caveat continues to apply, and proposal authors working on East African long rains should frame adaptation logic around observed drying, not modelled wetting.

CORDEX-Africa CMIP7

CORDEX is the WMO-WCRP initiative producing dynamically-downscaled regional climate model simulations. Africa is one of its flagship domains (Region 5), with the CORDEX International Project Office hosted jointly by CSAG at the University of Cape Town and the European Space Agency.

CORDEX-CMIP7 is formally chartered — a Task Force on Preparation of CORDEX-CMIP7 (coordinator Jesús Fernández) plus a separate Task Team on Protocol and Infrastructure. Protocols are being finalised through late 2025 / 2026; simulations will run through 2026–2028.

For African adaptation work, CORDEX-CMIP7 is the slower path but the right path for use cases where dynamical fidelity matters — convection-permitting precipitation extremes, monsoon dynamics, mountain orographic effects. CORDEX-CMIP6 outputs over Africa are still being archived (specs v3 released May 2026); the CMIP7 generation will follow.

What to do in the meantime

For proposal authors and CGIAR partners writing in 2026 and 2027:

Frame on CMIP6. It remains the citable, defensible, downscaled-and-evaluated corpus. NEX-GDDP-CMIP6 v2.0, CHELSA-CMIP6, ISIMIP3b all stay in service.
Add a “CMIP7 readiness clause” to multi-year proposals. Note that the team will adopt CMIP7-derived ISIMIP / NEX-GDDP / CHELSA products as they publish, with re-baselining planned for the AR7 cycle (2028+).
For Recent Changes views: continue the 1991–2020 WMO baseline. See Baselines and reference periods. CMIP7 doesn’t change observational baselines.
For Future Projections views: continue the 1995–2014 CMIP6 reference period. CMIP7 will likely use a similar but not identical baseline once products land; the baselines page will be updated when that’s confirmed.
Watch ISIMIP4 first. It has the firmest timeline and the strongest sector specificity (crops, water, health) most relevant to CGIAR partner work.
Don’t commit specifically to NEX-GDDP-CMIP7 in roadmaps. No public release timeline exists; assume NEX-GDDP-CMIP6 v2 is the terminal NASA downscaling vintage for at least 18 months. For continuity-critical work, evaluate CIL-GDPCIR and CHELSA-CMIP7 as backup paths.

If your work requires partner-government-facing certainty over a multi-year horizon, consult the CGIAR Climate Action data team for the current internal recommendation on the CMIP7 transition pathway. The internal guidance is updated quarterly as the field moves.

Where this page sits in the wiki

This page is the forward-look part of the wiki. The companion pages cover:

The current methodology being applied: African CMIP6 Ensembling
The defaults CGIAR centres should harmonise around today: Recommendations for CGIAR
The dataset landscape your team will need to navigate during the transition: Dataset landscape
The regional evaluation literature that will be re-done for CMIP7: Regional evaluation
The paradox case that CMIP7 may help resolve: East African Paradox

When CMIP7-derived African projections become viable, the CGIAR Climate Data Hub will update the underlying methodology pages and add a CMIP7 sub-ensemble alongside the existing CMIP6 view — not as a flag-day cutover. Old versions of this page are retained for citation stability.