🌊 Carbon Sinks — Oceans, Forests, Soil & Blue Carbon ~54% of CO₂ emissions re-absorbed annually Sinks are weakening relative to emissions

Natural carbon sinks — oceans, forests, and soils — absorb roughly half of human CO₂ emissions each year, slowing atmospheric accumulation. Their capacity is finite, vulnerable, and changing Sources: Global Carbon Project 2023; IPCC AR6 WG1 Ch. 5; Friedlingstein et al. 2023; Pan et al. 2011; DeVries et al. 2023; Regnier et al. 2022; IPCC Blue Carbon report
36.8 Gt CO₂
Fossil fuel + cement emissions (2023)
Annual total from burning coal, oil, gas, and cement production; a new record high
+4.2 Gt CO₂
Land use change emissions (2023)
Deforestation, peatland drainage, and agricultural conversion; includes some reforestation offset
−10.5 Gt CO₂
Ocean carbon sink (2023)
Surface ocean absorbs ~27% of total anthropogenic CO₂ through dissolution and biological pump; growing in absolute terms but shrinking as fraction of emissions
−11.1 Gt CO₂
Land (biosphere) sink (2023)
Net forest growth, vegetation uptake, and soil accumulation; absorbs ~29% of total emissions; highly variable year-to-year with ENSO
+19.4 Gt CO₂
Net atmospheric accumulation (2023)
What stays in the atmosphere = emissions − ocean sink − land sink. 2023 growth rate: ~3.4 ppm/yr — the highest ever recorded
~3,000 Gt C
Soil carbon reservoir (0–100 cm)
The largest terrestrial carbon pool — 2× the atmosphere. Vulnerable to warming-accelerated decomposition and tillage disturbance
38,000 Gt C
Ocean total carbon reservoir
The ocean holds ~50× more carbon than the atmosphere; deep-ocean carbon has a residence time of ~1,000 years
~55 Gt CO₂/yr
Total sink capacity needed by 2050 (net-zero)
At net-zero, remaining hard-to-abate emissions (~5–10 Gt/yr) must be balanced by engineered and natural sinks; requires massive sink enhancement or CDR

Global Carbon Budget — Emissions vs. Sinks (1960–2023)

Source: Global Carbon Project (Friedlingstein et al. 2023, Earth System Science Data); Le Quéré et al. 2018; Friedlingstein et al. 2022; IPCC AR6 WG1 Chapter 5 Box 5.1 (carbon budget); Ciais et al. 2013.

Where Does Our CO₂ Go?

Sources: Fossil fuels + land use change

In 2023, humanity emitted ~41 Gt CO₂/yr total (36.8 Gt fossil + 4.2 Gt land use change). This is a new record, and reflects a 70% increase from 1990 levels. Land use change emissions (primarily deforestation) have been relatively flat at 4–6 Gt/yr for two decades, while fossil emissions continue growing.

Ocean sink — growing, but not keeping pace

The ocean absorbed ~10.5 Gt CO₂ in 2023 (~27% of total emissions). Ocean uptake has grown in absolute terms as atmospheric CO₂ rises (Henry's Law: more CO₂ pressure → more dissolution), but has declined as a fraction of emissions from ~35% in the 1960s to ~27% today. Warmer oceans hold less CO₂, creating a negative feedback loop as warming accelerates.

Land sink — large but variable

Land ecosystems absorbed ~11.1 Gt CO₂ in 2023 (~29% of emissions), primarily through enhanced forest growth (CO₂ fertilisation, regrowth on previously disturbed land) and boreal/temperate forest expansion. This sink is highly variable: strong El Niño years (tropical droughts, wildfires) can reduce land uptake by 2–3 Gt; La Niña years see enhanced uptake. The 2023 sink estimate is provisional.

Total 2023 emissions~41.0 Gt CO₂
Ocean sink−10.5 Gt CO₂ (−27%)
Land sink−11.1 Gt CO₂ (−27%)
Atmospheric accumulation+19.4 Gt CO₂ (+46%)
Source: Friedlingstein et al. 2023; GCP 2023 annual budget; Dlugokencky & Tans NOAA/GML; IPCC AR6 Ch.5.

Carbon Sink Fraction of Emissions — 1960 to 2023

Source: Global Carbon Project 2023; Le Quéré et al. 2018; Friedlingstein et al. 2022; Knorr 2009 (airborne fraction); Raupach et al. 2008. The "airborne fraction" (~46%) = fraction of total emissions remaining in atmosphere = 1 − (ocean fraction + land fraction).

Ocean Carbon Uptake — Mechanisms & Depth Profile

Source: Sarmiento & Gruber 2006 (Ocean Biogeochemical Dynamics); Sabine et al. 2004 (Science — anthropogenic CO₂ in ocean interior); Gruber et al. 2019 (Nature — ocean carbon sink 2000s); DeVries et al. 2023; Landschützer et al. 2015 (surface pCO₂ decadal trends).

How the Ocean Absorbs Carbon

The solubility pump

CO₂ dissolves in surface seawater according to Henry's Law — cold, high-latitude waters (Arctic, Southern Ocean) absorb more CO₂ per unit volume. Surface currents carry dissolved CO₂ downward into deep water via thermohaline circulation (the "ocean conveyor belt"), isolating it from the atmosphere for hundreds to thousands of years. The Southern Ocean is responsible for ~40% of total ocean CO₂ uptake despite covering ~20% of ocean area.

The biological pump

Phytoplankton absorb dissolved CO₂ through photosynthesis, fixing it into organic matter. When these organisms die or are eaten, a fraction of their carbon sinks as "marine snow" to the deep ocean — effectively removing it from contact with the atmosphere. The biological pump transports ~5–12 Gt C/yr to depth, but most is remineralised above 1,000 m; only ~0.1–0.2 Gt C/yr reaches the permanent deep sediment reservoir.

Ocean acidification — the cost of the sink

As CO₂ dissolves, it forms carbonic acid, which dissociates into bicarbonate and hydrogen ions — reducing seawater pH. Ocean surface pH has declined from ~8.18 in 1750 to ~8.07 today — a 26% increase in acidity (pH is logarithmic). Continued acidification threatens calcifying organisms (corals, molluscs, pteropods) that build calcium carbonate shells, and may eventually weaken the biological pump itself.

Source: Orr et al. 2005 (Nature — ocean acidification); Caldeira & Wickett 2003; Hoegh-Guldberg et al. 2007; IPCC AR6 WG1 Ch. 5.3; DeVries 2022.
The Southern Ocean — Earth's most important carbon sink: The Southern Ocean (south of 35°S) absorbs ~40% of global ocean CO₂ uptake and ~15% of all annual anthropogenic emissions. This enormous uptake is driven by cold surface waters, strong westerly winds driving upwelling, and phytoplankton blooms. However, Southern Ocean sink efficiency has been variable: it weakened significantly in 1990–2005 (possibly due to stronger westerly winds driven by ozone hole recovery), then recovered. The 2016 "blob" of warm water and other events show its vulnerability to both climate change and ozone recovery dynamics.

Ocean Carbon Reservoir vs. Atmosphere vs. Land

Source: Ciais et al. 2013 (IPCC AR5 Ch. 6); Sabine et al. 2004; IPCC AR6 WG1 Ch. 5 Table 5.1 (global carbon stocks); Friedlingstein et al. 2022 (flux estimates); Scharlemann et al. 2014 (soil carbon); Jobbágy & Jackson 2000.

Land Carbon Flux by Region (average 2010–2023)

Source: Pan et al. 2011 (Science — global forest carbon sink); Harris et al. 2021 (Nature Climate Change — global forests carbon sink); Spawn et al. 2020 (global biomass); Bastin et al. 2019; GCP 2023; IPCC AR6 WG1 Ch. 5 Box 5.1; Gatti et al. 2021 (Amazon source).

Forest and Land Carbon — Key Dynamics

The global forest sink

Intact and recovering forests are the dominant component of the land sink, absorbing ~7–9 Gt CO₂/yr gross through net biomass accumulation. Boreal forests (Russia, Canada) and temperate forests (Europe, China, eastern US) account for most of this sink. Tropical forests (Amazon, Congo, SE Asia) are globally important but their net status depends on whether deforestation exceeds regeneration — the Amazon may have flipped to a net carbon source in some regions.

CO₂ fertilisation — real but overstated

Higher atmospheric CO₂ directly stimulates plant photosynthesis ("CO₂ fertilisation" or the "greening" effect). Satellite data show global vegetation greenness increased ~14% from 1982–2015. However, this effect is nutrient-limited — extra plant growth requires nitrogen and phosphorus, which are often scarce. Water limitation and heat stress increasingly constrain the CO₂ fertilisation effect in warm, dry regions. CMIP6 models may overestimate this effect.

Deforestation — the source that erodes the sink

Land use change releases ~4–6 Gt CO₂/yr globally, primarily from tropical deforestation. The net land sink (gross uptake minus land use change emissions) has remained roughly stable at 10–12 Gt CO₂/yr even as gross uptake from CO₂ fertilisation has grown — because deforestation emissions have increased in parallel. Protecting existing forests is 2–3× more climate-effective than planting new trees of equivalent area, because established forests have decades of stored carbon that would be released if cleared.

Source: Pan et al. 2011; Harris et al. 2021; Gatti et al. 2021; Le Quéré et al. 2018; Bastin et al. 2019; Walker et al. 2022 (CO₂ fertilisation limits); IPCC AR6 WG1 Ch. 5.
Tropical forests — from sink to source: A landmark study (Gatti et al. 2021, Nature) found that eastern Amazon regions have become net carbon sources even before considering deforestation — rising temperatures and drought are driving higher tree mortality and increased respiration. The western Amazon remains a sink. The Amazon as a whole may have crossed or be approaching the "tipping point" where it transitions from net sink to net source, driven by deforestation (now ~20% of original forest area lost) and climate-change-amplified drought stress. A degraded Amazon would release ~90 Gt CO₂ total — equivalent to ~9 years of current global fossil fuel emissions.

Soil Carbon Pools — Size and Vulnerability

Source: Scharlemann et al. 2014 (Global Ecology & Biogeography — world soil carbon); Jobbágy & Jackson 2000 (vertical distribution SOC); Jackson et al. 2017 (soil C vulnerability); Tarnocai et al. 2009 (permafrost carbon); Hugelius et al. 2014 (circumpolar permafrost carbon); Friedlingstein et al. 2022.

Soil Carbon — The Giant Underground Reservoir

Scale of soil carbon storage

Soils contain roughly 1,500–2,400 Gt C in the top 1 metre (3,000 Gt C to 3 m depth) — approximately 2× the atmospheric carbon pool (~880 Gt C) and 3× all living vegetation (~450 Gt C). Most of this carbon consists of partially decomposed plant matter, fungal biomass, and mineral-associated organic carbon (stabilised by binding to clay and mineral surfaces). The deeper the soil, the older and more stable the carbon — some deep soil carbon is thousands of years old.

The permafrost carbon bomb

Permafrost soils in Arctic regions store ~1,460–1,600 Gt C — the largest single terrestrial carbon pool. Currently frozen and inactive, this carbon is mobilised as permafrost thaws. Arctic warming is occurring at 2–4× the global average rate. Models project 37–174 Gt C released by 2100 under high-emission scenarios (RCP8.5), with a tail risk of abrupt "thermokarst" collapses releasing large pulses in decades rather than centuries. Permafrost carbon release is largely irreversible on human timescales.

Warming accelerates decomposition

Soil organic matter decomposition by microbial communities is temperature-sensitive — roughly doubling for every 10°C increase (Q₁₀ ≈ 2). A 2°C warming of deep soil layers could release an extra 55 Gt C (200 Gt CO₂) by 2100, according to the most comprehensive study of deep soil carbon stocks (Crowther et al. 2016, Nature). This creates a positive feedback loop: more CO₂ → more warming → more soil decomposition → more CO₂.

Source: Crowther et al. 2016 (Nature — soil warming and organic C); Scharlemann et al. 2014; Hugelius et al. 2014; Koven et al. 2015; Turetsky et al. 2019 (thermokarst); Jobbágy & Jackson 2000.
Managed soils — from source to sink: Agricultural soils have lost ~50–70% of their original organic carbon through tillage, chemical inputs, and compaction — a historical emission of ~130 Gt C since the start of agriculture. Regenerative practices (no-till, cover cropping, compost application, diverse rotations) can rebuild this carbon at rates of 0.1–0.9 t C/ha/yr. The "4 per mille" initiative (COP21, 2015) estimates that increasing global soil carbon by 0.4%/yr would offset all current anthropogenic CO₂ emissions — a theoretical maximum that would require transforming essentially all agricultural land management simultaneously.

Blue Carbon Ecosystem Carbon Density

Source: Mcleod et al. 2011 (Nature Geoscience — blue carbon); Duarte et al. 2013 (blue carbon review); Pendleton et al. 2012 (Nature Climate Change); Howard et al. 2017 (IUCN blue carbon); Nellemann et al. 2009 (UNEP blue carbon report); Murdiyarso et al. 2015 (mangrove peatlands).

What Is Blue Carbon?

Coastal ecosystems — disproportionately powerful sinks

Blue carbon refers to the carbon captured and stored by marine and coastal ecosystems — primarily mangroves, seagrasses, and tidal salt marshes. Despite covering less than 0.5% of the sea floor, these ecosystems store more carbon per unit area than most terrestrial forests — primarily because their waterlogged, anoxic soils prevent the decomposition of organic matter, allowing peat to accumulate for thousands of years.

Mangroves — the most carbon-dense forests on Earth

Mangrove forests store 3–4× more carbon per hectare than most tropical rainforests (1,000+ t CO₂-eq/ha total ecosystem carbon, including deep peat soils). They are also among the most threatened — losing ~0.3–0.5% of global area per year to coastal development, aquaculture, and sea level rise. Destroying a mangrove releases stored peat carbon accumulated over centuries — a deforestation carbon debt far larger than equivalent terrestrial forest loss.

Seagrass meadows — the ocean's grasslands

Seagrass meadows cover ~300,000 km² of shallow coastal ocean and sequester ~0.15 Gt C/yr — modest in absolute terms but concentrated in shallow, accessible areas. Seagrass sediments can store thousands of years of accumulated carbon. Seagrasses are declining at ~7% per year globally due to coastal eutrophication, turbidity, and dredging — and restored seagrass meadows can re-establish their carbon stocks within decades.

Source: Nellemann et al. 2009; Mcleod et al. 2011; Fourqurean et al. 2012 (Nature Geoscience — seagrass carbon stocks); Pendleton et al. 2012; Howard et al. 2017; Hamilton & Friess 2018.
Wetland peatlands — the largest terrestrial organic carbon store: Tropical peatlands (Indonesia, Congo) and boreal/subarctic peatlands (Canada, Russia, Scandinavia) store an estimated 400–600 Gt C in peat deposits up to 20 metres deep — carbon accumulated over 5,000–10,000 years since the last ice age. Draining peatlands for agriculture (palm oil in SE Asia, soy in Brazil) releases this ancient carbon at rates of 2–9 t CO₂/ha/yr. Indonesia's peatland fires during 2015 El Niño released ~1.75 Gt CO₂-eq in a single season — equivalent to ~5% of global annual fossil fuel emissions from an area smaller than California.

Sink Feedback Risks Under Different Warming Scenarios

Source: Cox et al. 2000 (Nature — carbon cycle climate feedback); Friedlingstein et al. 2006 (C4MIP); Gregory et al. 2009; Jones et al. 2013; IPCC AR6 WG1 Ch. 5.4 (carbon cycle feedbacks); Lowe et al. 2018 (constrained sink projections); MacDougall et al. 2020.

Carbon Cycle Feedbacks & Tipping Points

Why sinks may weaken — the positive feedback loop

Current carbon sinks exist partly because the atmosphere is rich in CO₂ (fertilising plant growth) and partly because cold, dense ocean water readily dissolves gas. As temperature rises, both mechanisms weaken: plants face heat and drought stress, oceans warm and hold less CO₂ (Henry's Law — solubility decreases with temperature), and soils decompose faster. IPCC AR6 models show a declining sink efficiency under all warming scenarios — meaning each tonne of CO₂ emitted has a progressively larger atmospheric impact.

Potential tipping point: Amazon dieback

Modelling studies suggest the Amazon may have a tipping point at approximately 20–25% deforestation (currently at ~20%), beyond which the forest loses enough moisture recycling to sustain itself — leading to a self-reinforcing dieback that releases ~90 Gt C over decades. Combined with climate-change-driven drying, this tipping point may be closer than deforestation percentage alone suggests.

Potential tipping point: Permafrost thaw cascade

Gradual permafrost thaw releases mostly CO₂; "abrupt thaw" (thermokarst lakes, retrogressive thaw slumps) releases CO₂ and CH₄, with faster timescales (years to decades). Current models include gradual thaw but poorly represent abrupt thaw — potentially underestimating permafrost carbon feedback by 50–100%. Some studies suggest permafrost carbon feedback alone could add 0.3–0.5°C to 2100 warming under high-emission scenarios.

Source: Lenton et al. 2019 (Nature — tipping elements); Armstrong McKay et al. 2022 (Science — tipping points); Turetsky et al. 2020 (thermokarst); Gatti et al. 2021; Nobre et al. 2016 (Amazon tipping); IPCC AR6 WG1 Ch. 5.4.
The carbon-climate feedback: sinks lose efficiency as warming increases: IPCC AR6 assesses the carbon-climate feedback (β-γ framework) as follows: for every 1°C of warming, land and ocean sinks together absorb ~7–12% less of each additional tonne of CO₂ emitted. At 3°C warming (without action), land sinks may absorb 25–40% less of our emissions than they do today — meaning the same emission rate produces faster atmospheric CO₂ accumulation. This is a self-reinforcing spiral: warming weakens sinks → more CO₂ stays in atmosphere → faster warming → weaker sinks. It represents one of the fundamental reasons early and deep emissions reductions are far more effective per tonne avoided than late reductions.
Carbon Dioxide Removal (CDR) — sinks we build: Natural sinks alone cannot achieve net-zero; hard-to-abate sectors (aviation, cement, agriculture) will still emit 5–10 Gt CO₂/yr by 2050 even under aggressive mitigation. All IPCC 1.5°C pathways require some CDR. Natural CDR (reforestation, soil carbon, wetland restoration) offers 5–10 Gt CO₂/yr by 2050, cost ~$20–100/t CO₂. Engineered CDR (BECCS, DAC, enhanced weathering, ocean alkalinity) offers 2–10 Gt CO₂/yr, cost ~$100–600+/t CO₂. No CDR pathway at scale currently exists; scaling any of them faces land, water, energy, or cost constraints.