Where will the 2026 El Niño hit the carbon cycle?
Seasonal weather forecasts drive two land models and a fire predictor so we can see, months ahead, how much the land sink may weaken — and which tropical forests and peatlands face the highest risk.
Why it matters
A very strong El Niño is already unfolding in 2026, and land ecosystems usually soak up less carbon — or burn and release more — when the tropics turn hot and dry. This early-warning setup turns seasonal weather forecasts into maps of where forests and peatlands are most at risk, so fire crews and land managers can watch the right places before the peak months hit.
The paper
Forecasting carbon-risk hotspots ahead of the extreme 2026 El Niño
Xiaomeng Du, Runda Jiang, Philippe Ciais, Pierre Friedlingstein, Stephen Sitch, Mike O'Sullivan, Jefferson Goncalves de Souza, Tim Stripp, Piyu Ke, Yi Xi, Minxuan Sun, Xiaoran Wu, Mengjie Han, Shanshan Song, Qiang Zhang & Wei Li. arXiv:2610.16714 [physics.ao-ph], submitted 12 Oct 2026. Read on arXiv · PDF
In plain language: the team fed a 51-member ECMWF seasonal forecast (started 1 June 2026) into two vegetation models (ORCHIDEE-MICT and JULES) plus machine-learning burned-area predictions for July–December 2026. The ensemble-mean land sink weakens by about 0.23 PgC over that half-year — milder than 2015 or 2023 on average, but the worst member reaches about −1.0 PgC, like 2015. Tropical South America drives roughly 44% of the global decline; Equatorial Asia peatlands face burned area about 5.6× the recent non–El Niño mean.
Try it
Slide ensemble severity from an optimistic near-zero outcome toward the paper’s worst member (~−1.0 PgC). Month focus and flux bars use published Nov/Dec means and the Jul–Dec flux split; other months are shaped to match the −0.23 PgC cumulative mean and labelled illustrative.
The maths in this page
Net biome productivity anomalies follow the paper’s flux budget:
For the Jul–Dec 2026 ensemble mean the paper reports NBP = −0.23 ± 0.21 PgC, driven by increases of Ra 1.48 ± 1.61, Rh 0.86 ± 1.30, and fire 0.46 ± 0.02 PgC, partly offset by a GPP increase of 2.57 ± 2.72 PgC. Historical same-period NBP: 2015 −0.95 ± 0.03, 2023 −1.39 ± 0.02 PgC. Worst ensemble member: −1.0 ± 0.3 PgC.
- Severity slider: maps 0 → ~0 PgC (optimistic), ~23 → ensemble mean −0.23, 100 → worst-case −1.0. Tropical South America (~44% of the global decline at the mean), EQAS fire (5.6×), and Australia fire (+61.8%) scale with the same factor for the toy.
- Monthly curve: Nov −0.12 ± 0.08 and Dec −0.13 ± 0.03 PgC mon⁻¹ are from the paper; Jul–Oct are an illustrative slide that sums with them to the chosen Jul–Dec total.
- Regional context: Tropical SA NBP reduction 0.24 ± 0.1 PgC (44/51 members agree on sign); ~11.4% of global land (16.4% in the tropics) more negative than both 2015 and 2023; tropical precip 11.7 ± 2.3% below the 2016–2025 non–El Niño reference.
This is an illustrative toy from published ensemble means and spreads, not the authors’ DGVM or ML code. Absolute fluxes at intermediate severities are linearly scaled and should not be treated as new forecasts.
How to use it
Leave severity near 23 to sit on the ensemble mean, then drag toward 100 to see a 2015-like sink collapse and watch Tropical South America and Equatorial Asia fire risk climb with it. Slide the month focus into November and December — the paper’s worst months — and compare your bar to the 2015 and 2023 markers. Toggle flux decomposition on to see how rising respiration and fire outweigh the GPP bump. The takeaway for managers: watch northeastern and central Brazil for sink loss, and Indonesian, PNG, and Malaysian peatlands for fire, before the Nov–Dec peak.