Climate projections for Africa — a 10-minute primer

Africa is warming. Each stripe is one year of continental annual temperature anomaly, 1880–2025, relative to the 1991–2020 baseline. Style inspired by Ed Hawkins's #ShowYourStripes. Data: NASA GISTEMP v4 tropical band (24°S–24°N), 1991–2020 baseline. data.giss.nasa.gov/gistemp. See commission brief for methodology. Regenerate with scripts/regen-warming-stripes.py (switches to Berkeley Earth Africa land data automatically when their server is reachable).

If you remember one thing

A climate projection isn’t a forecast of what will happen. It’s a model’s answer to “if emissions follow trajectory X, here’s what the physics says could happen.” The honest reading of any projection chart is always: which scenario, which models, and how much do they agree — not a single number.

What this page covers

Your 10-minute orientation to climate projections for African adaptation work. By the end you’ll know:

• What a “future climate projection” actually is (and what it isn’t)
• The three sources of uncertainty in every projection chart you’ll see
• How to read a projection chart honestly — central estimate, spread, scenario dependence
• Where projections fit into a climate-rationale workflow
• What the rest of this wiki section can answer for you

What a projection is, and isn’t

A climate projection is a model-based answer to the question what would happen to the climate if greenhouse-gas emissions and land use follow a specific future trajectory? That trajectory is called a scenario. For CMIP6, scenarios are the Shared Socioeconomic Pathways (SSPs) — SSP1-2.6 (sustainability), SSP2-4.5 (middle of the road), SSP3-7.0 (regional rivalry), SSP5-8.5 (fossil-fuelled development).

Projections are not predictions. A prediction would assume we know which scenario will play out; we don’t. A projection says: given this scenario, the climate would behave roughly like this. Every projection chart you read is conditional on a choice the chart-maker made about emissions (IPCC AR6 WGI Ch 1; AR6 WGI Summary for Policymakers).

Where projections come from — the short version

Projections come from global climate models (GCMs) — large physics-based simulations run at major modelling centres around the world. CMIP6, the current coordinated experiment, gathered runs from ~50 models. NASA’s statistically-downscaled NEX-GDDP-CMIP6 product provides 18 of them at 0.25° resolution; this is the most widely used dataset for African projection work, and it’s the one the AAA Adaptation Atlas’s Build a Climate Rationale notebook is built around (see African CMIP6 Ensembling for which 18 models and why). The next page goes deep on what a GCM actually is: Climate models 101.

The three sources of uncertainty

Hawkins & Sutton (2009) gives the standard framing — every projection carries three layered uncertainties, and you should know which one dominates for your time horizon:

• Scenario uncertainty — we don’t know which emissions path will play out (which SSP humanity ends up on).
• Model uncertainty — even given a scenario, different models give different answers because they represent atmosphere, ocean, land, and clouds differently.
• Internal variability — the climate’s own year-to-year and decade-to-decade noise, driven by modes like ENSO and the monsoon. It doesn’t go away with averaging across models.

A rough rule: for near-term horizons (2030s), internal variability dominates the spread. For end-of-century horizons (2080s+), scenario uncertainty does. Model uncertainty is in the middle and matters most at the regional scale — which is exactly where you’re working.

How to read a projection chart

A typical chart shows two things layered on top of each other:

• An ensemble mean — a single thick line through the average of multiple models. This is your central estimate.
• A spread band — usually shaded — showing the inter-model range (typically ±1 standard deviation, or the 17–83% inter-model interval). This is the uncertainty.

If multiple emissions scenarios are shown, each one gets its own coloured mean-and-spread band. The bands almost always overlap in the near term and diverge later as scenario uncertainty grows.

If a chart shows only the mean and not the spread, half the story is missing — treat it with caution and look for the source’s underlying data. Honest charts always include the spread.

Figure 1. Anatomy of a projection chart (interim schematic). Two stylised CMIP6 emissions scenarios — SSP2-4.5 in blue, SSP5-8.5 in amber — showing how every honest projection chart layers three things: a central estimate (ensemble mean), inter-model spread (shaded envelope), and scenario uncertainty (the gap between scenarios that opens by end-of-century). Y-axis is °C above 1991–2020 (the wiki's default baseline; see Baselines and reference periods). Interim schematic — the final figure will be derived from IPCC AR6 WGI SPM Figure 8, panel (a) (CC-BY 4.0) per the commission brief.

Where this fits in a climate rationale

A defensible climate rationale for a Green Climate Fund concept note, an Adaptation Fund proposal, or a national NAP submission usually has four parts:

1. Observed climate context — the current baseline. Best continental source: WMO State of the Climate in Africa 2024 (African temperature anomaly +0.86 °C vs the 1991–2020 baseline; North Africa fastest-warming; 2024 saw Sahel floods and Kenya MAM floods).
2. Observed change — recent trends in temperature, rainfall, extremes.
3. Projected change — what this wiki section helps you cite defensibly.
4. Adaptation logic — how the projected changes drive the intervention you’re proposing.

Projections sit in part (3), but always read alongside (1) and (2) — a projection that contradicts the observed trend (as for East African long rains; see East African Paradox) is a known-difficult case, not a free pass to use the model number.

What this wiki section can answer

Question	Page
What’s actually inside a climate model?	Climate models 101
Why do models disagree?	Why models disagree
What does “downscaled” mean?	Downscaling
What does “bias-corrected” mean?	Bias correction
Which datasets exist and what should I use?	Dataset landscape
Which models work best over my region?	Regional evaluation
What does the AAA Adaptation Atlas specifically use?	African CMIP6 Ensembling
Why is East African rainfall a special case?	East African Paradox
What’s coming next?	CMIP7 + CORDEX-Africa

Further reading

• IPCC AR6 WGI Chapter 1 — Framing, context, methods — the canonical reference for what climate projections are and aren’t.
• IPCC AR6 Interactive Atlas — region-by-region projection viewer; explore SSP bands for any AR6 reference region.
• WMO State of the Climate in Africa 2024 — the most authoritative observational baseline for continental and regional change.
• Hawkins & Sutton 2009 — The potential to narrow uncertainty in regional climate projections — the original framing of the three uncertainty sources.
• Carbon Brief — How Shared Socioeconomic Pathways explore future climate change — non-specialist orientation to SSPs.
• Carbon Brief — CMIP6: the next generation of climate models explained — what changed from CMIP5 to CMIP6 and why it matters.