However, this early in the year large uncertainties remain – particularly with a strong El Nino developing. It could end up anywhere between the warmest and fourth warmest, with about a 19% chance of 2026 surpassing 2024 as the warmest year on record.
Global temperatures in 2026 are on track to be the second warmest on record, at around 1.47C above preindustrial levels across the five different records assessed by Carbon Brief. Read more in our latest State of the Climate update for Q1 2026 here: www.carbonbrief.org/...
Average detrended SST anomalies in the Nino3.4 region of the tropical Pacific
Solar and EVs will matter a lot more than curtailing population growth, given that most population growth is happening in the parts of the world with the lowest per-capita emissions.
Sure, though both have their uses. But RONI is up to 2.5C (and even higher in the most recent 8 runs):
With the latest set of El Niño forecasts, NOAA's CFSv2 model needs to get a larger y-axis scale 😯
Yep, 0.25C per decade. Typo on my part!
2C by 2050 would be a warming rate of ~0.25C per year which is more or less in-line with climate model estimate: www.carbonbrief.org/analysis-wha...
As a percentage of GDP, yes, not not higher in real (inflation-adjusted) dollars. Both are useful metrics.
Note that clean energy here includes all energy transition investments: power sector, transportation, and building decarbonization.
The AI datacenter buildout is huge compared to past projects like the interstate highway system, the Apollo program, and the US railroad boom. But its still modest compared to global clean energy investment or oil and gas investment:
🌐 NEW | Analysis: Global fossil power generation fell in March after Hormuz closure due to solar & wind growth
⚠️ Fall in gas-fired generation offset by large increases in solar & wind power, not coal
w/CREA's @laurimyllyvirta.bsky.social
energyandcleanair.org/fossil-power...
May be on track to be closer to 2015/2016, but we will see!
2.2C is the Nino3.4 region anomaly, not the global average temperature anomaly.
Its a compilation of different sources, but the main ones are:
C3S: climate.copernicus.eu/seasonal-for...
NMME: www.cpc.ncep.noaa.gov/products/NMME/
CFS: www.cpc.ncep.noaa.gov/products/CFS...
CanSIPS: climate-scenarios.canada.ca?page=cansips...
Perhaps "is planned to be" would be more accurate, but yeh...
El Nino is just trying to make the 2027 values (which will be the latest full year reported in the AR7 WG1) more exciting and policy-relevant 😉
Its the Oceanic Nino Index, essentially detrended anomalies in the Nino3.4 region of the tropical Pacific: www.cpc.ncep.noaa.gov/products/ana...
If NCAR-CESM1 is right (it isn't) we would have a Mechagodzilla El Niño 😛
Also, to avoid potential confusion, the 2.2C value for September is the ONI value for the Nino3.4 region of the tropical Pacific, not global mean surface temperatures (those will be lower!).
You can find regularly updated data on these forecasts over at the Climate Dashboard: dashboard.theclimate...
Second, the models used here that are obtained solely from the NMME forecast (NCAR-CCSM4, NCAR-CESM1, NASA-GOES) do not include a separate amplitude correction as 1) NMME doesn't make the amplitude correction data available and 2) models from other sources do not include this step.
A few important caveats here: this uses the old ONI definition of El Niño rather than the new relative ONI (RONI) one, which would tend to reduce the magnitude of the event. ONI is used here as not all models provide the data necessary for me to easily calculate RONI.
There are some notable low end projections (CanESM5 and DWD) that include a moderate or even no El Niño formation, and a few high end outliers (NASA-GEOS and NCAR-CESM1) that are probably unrealistically high. But most models are clustering between +1.9C and +2.4C.
We now have 13 different models with 637 different ensemble members with El Niño forecasts out through at least September 2026.
They suggest a best estimate of 2.2C for September; interestingly ECMWF (which was seen as particularly hot when it came out) is right around the middle of the pack:
Nothing generalizable, as it would depend on the methane leakage rate for the plant in question (which is very region-specific). But you count probably roughly calculate it by assuming a ~2% average leakage rate and the molar mass ratio of CO2 to CH4 in combustion vs leakage.
What I've seen emerging is "CMIP7 High", "CMIP7 Medium", etc. Which from a communications perspective is probably an improvement over SSP2-4.5 as a name 😉
But would be interesting to get @benmsanderson.bsky.social's perspective here.
Neat! We also mentioned this idea in passing in Brunner et al 2024 (www.nature.com/articles/s43...) but never really elaborated on it.
That said, if you are indexing to the social cost of carbon and a ~3% discount rate, GWP100 gives you a much better match than GWP20: www.science.org/stoken/autho...
As you note the system design does matter a bit here in terms of predictability. A strict cap is different than a species-specific GHG price which in turn is different from a cap with price ceilings...