Arctic Permafrost Carbon Feedback
Arctic and sub-Arctic permafrost holds ~1,700 Gt of frozen organic carbon — twice the carbon currently in the atmosphere. As the Arctic warms 2–3× faster than the global mean, this carbon is thawing and releasing CO₂ and CH₄. This feedback is absent from IPCC AR6 central projections and could add 0.2–0.4 °C of additional warming by 2100 on top of modelled trajectories.
Annual CO₂e Flux by Scenario (2024–2100)
Cumulative CO₂e Released (Gt) by 2100
Scenario Summary
| Scenario | 2050 cum. (Gt CO₂e) | 2100 cum. (Gt CO₂e) | Peak flux (Mt/yr) | Add. warming 2100 (°C) | Tipping threshold yr |
|---|---|---|---|---|---|
| Low Warming | 42.529 | 114.193 | 1705.1 (2040) | +0.188 | — |
| Moderate Warming | 61.272 | 202.941 | 3280.2 (2060) | +0.335 | 2099 |
| High Warming | 78.343 | 252.052 | 4535.4 (2058) | +0.416 | 2080 |
| Abrupt Thaw | 148.644 | 305.409 | 7297.7 (2047) | +0.504 | 2058 |
Annual CO₂ Flux by Scenario (Mt/yr)
Annual CH₄ as CO₂e by Scenario (Mt CO₂e/yr)
Total Annual CO₂e Flux (all scenarios, Mt/yr)
Cumulative Permafrost CO₂e Release (Gt CO₂e)
% of Total 1,700 Gt C Stock Released
Even the high-warming scenario releases less than 10% of the total stock by 2100 — but this is a slow-moving feedback that accelerates over centuries. The concern is not a one-time pulse but a persistent source term that outlasts any net-zero transition.
Context: Permafrost Release vs Human CO₂ Budget
| Quantity | Gt CO₂e | Note |
|---|---|---|
| Remaining 1.5 °C budget (66% prob) | ~400 | IPCC AR6, as of 2024 |
| Remaining 2.0 °C budget | ~1,150 | IPCC AR6, as of 2024 |
| Permafrost by 2100 — Low | ~114.193 | This model |
| Permafrost by 2100 — Moderate | ~202.941 | This model |
| Permafrost by 2100 — High | ~252.052 | This model |
| Permafrost by 2100 — Abrupt | ~305.409 | This model |
Additional Warming from Permafrost Feedback (°C)
Global Temperature Trajectory per Scenario (°C above pre-industrial)
Gradual Active-Layer Deepening
The active layer is the seasonal thaw zone above perennially frozen ground. It deepens ~0.3–0.5 cm per year currently, but accelerates with warming. As organic material above the permafrost table is exposed to microbial decomposition, CO₂ is released aerobically.
This is the "baseline" thaw mode — slow, geographically widespread, well-characterised. It underlies the Koven et al. sensitivity of ~37 Gt C per °C of sustained warming.
Thermokarst (Abrupt Thaw)
Ice wedges and massive ground ice, when melted, cause ground subsidence and collapse — forming thermokarst lakes, retrogressive thaw slumps, and boggy depressions. These expose deep, previously stable carbon to rapid anaerobic decomposition.
Turetsky et al. 2020 showed thermokarst affects ~20% of permafrost area but contributes ~40–50% of total carbon release. It is rapid (years to decades, not centuries) and largely unrepresented in global models.
Yedoma — The Deep Frozen Reserve
Yedoma is exceptionally ice-rich (40–80% ice by volume) Pleistocene-age loess deposit found across Siberia and Alaska. It contains ~130 Gt C in particularly labile form — 10× more reactive than typical permafrost carbon.
Yedoma thaw events (e.g. the 2020 Siberian heat wave, +38 °C) can release CH₄ bubbles from lake bottoms ("ebullition"). These point sources are intermittent but potentially very large.
Self-Reinforcing Feedback Loop
Permafrost methane warms the Arctic faster than the global mean (amplification factor ~2.5×). Faster Arctic warming → deeper active layer → more carbon release → more warming. This creates a self-reinforcing loop that, once triggered, cannot be halted by cutting human emissions alone.
The loop accelerates when: (1) warming exceeds ~2 °C globally; (2) sea ice loss exposes dark ocean, adding further Arctic warming; (3) boreal forest fire frequency increases, burning the insulating peat layer above permafrost.
Why It's Excluded from IPCC Models
Earth System Models (ESMs) used for IPCC AR6 represent permafrost carbon with varying fidelity. Most models simulate gradual active-layer deepening but omit thermokarst, yedoma, and deep carbon pools below 1 m depth.
IPCC AR6 WG1 Box 5.1 explicitly notes that the "permafrost carbon feedback is not included in the IPCC assessed likely ranges." The best estimate (+0.2 °C under SSP2-4.5) is given as "medium confidence" in a separate box — not baked into headline temperature projections.
This means every 1.5 °C and 2.0 °C budget number in AR6 is implicitly optimistic relative to a world that properly accounts for permafrost feedbacks.
Key Milestones in Permafrost Carbon Science and Policy
| Year | Event | Detail |
|---|---|---|
| 1970s | Permafrost Monitoring Begins | International Permafrost Association establishes borehole networks across Siberia, Alaska, and Canada. Data reveal consistent active-layer deepening of 0.3–0.5 cm/yr. |
| 2008 | Permafrost Carbon Feedback Recognised in AR4 | IPCC AR4 acknowledges permafrost as a potential carbon feedback but does not quantify it in model scenarios — a gap that persists through AR6. |
| 2015 | Koven et al. — Sensitivity Quantified | First systematic estimate of permafrost carbon–climate feedback: 30–100 Gt C released per °C of sustained global warming. Implies 0.1–0.5 °C additional warming by 2100 under RCP 4.5. |
| 2018 | Schuur et al. — Stock Revised Upward to 1,700 Gt C | Revised soil carbon inventories (to 3 m depth) raise total permafrost carbon estimate from ~1,000 Gt C to ~1,700 Gt C — double previous IPCC assessment values. |
| 2020 | Turetsky et al. — Abrupt Thaw Doubles Estimates | Nature paper shows thermokarst (ground collapse from ice wedge melt) and retrogressive thaw slumps release carbon ~2× faster than gradual models predict. Affects ~20% of permafrost area but contributes disproportionately to flux. |
| 2021 | IPCC AR6 WG1 — Feedback Excluded from Central Projections | Box 5.1 explicitly states permafrost feedbacks are not included in the IPCC 'medium confidence' projections due to model limitations. Best estimate: +0.2–0.4 °C additional warming by 2100 under SSP2-4.5. |
| 2024 | IPA State of the Permafrost — Record Active Layer Depths | International Permafrost Association 2024 report documents record active-layer thickness at 70% of monitoring sites across the Arctic. Siberian permafrost showing first signs of deep (>3 m) carbon mobilisation. |
| 2027+ | Tipping Point Risk Window | Multiple models suggest that warming above 2 °C global mean triggers self-reinforcing permafrost feedback: released methane warms Arctic faster than global mean, accelerating further thaw. No international policy mechanism currently addresses this feedback loop. |
Sources & References
| Source | Description | Key Contribution |
|---|---|---|
| Schuur et al. 2022 | Permafrost Carbon Network synthesis — "Permafrost and Climate Change: Carbon Cycle Feedbacks" | 1,700 Gt C stock estimate (to 3 m depth); revised upward from prior ~1,000 Gt C |
| IPCC AR6 WG1 (2021) | Ch.5 Box 5.1 — Permafrost carbon feedbacks | +0.2 °C best estimate under SSP2-4.5; explicitly excluded from headline ranges |
| Turetsky et al. 2020 | Nature — "Carbon release through abrupt permafrost thaw" | Thermokarst doubles previous gradual-thaw estimates; ~20% area → ~50% flux contribution |
| Koven et al. 2015 | Nature Clim. Change — "A simplified, data-constrained approach to estimate the permafrost carbon feedback" | Sensitivity 30–100 Gt C / °C warming; central ~37 Gt C / °C used here |
| Gasser et al. 2018 | Nature Geoscience — "Path-dependent reductions in CO₂ emission budgets caused by permafrost carbon release" | Permafrost adds +0.2–0.4 °C by 2100 under RCP 4.5–8.5 |
| Nitzbon et al. 2020 | Nature Comm. — "Pathways of ice-wedge degradation in polygonal tundra under different hydrological conditions" | Thermokarst dynamics under RCP 8.5; quantifies ice wedge melt accelerations |
| IPA State of Permafrost 2024 | International Permafrost Association annual monitoring report | Record active-layer depths at 70% of monitoring sites; 2023 Siberian anomalies |
| Millar et al. 2017 | Nature Geoscience — TCRE (Transient Climate Response to Cumulative Emissions) | ~1.65 °C per 1,000 Gt CO₂; used for additional-warming conversion in this model |