Evidence for Human-Caused Climate Change
In 1958, American chemist Charles David Keeling began recording CO₂ at 315 ppm from Hawaii's Mauna Loa Observatory, starting the world's most important climate dataset, now showing over 420 ppm.
● Know
- The Keeling Curve shows continuous CO₂ monitoring from Mauna Loa since 1958
- Ice-core records show CO₂ over 800,000 years, never above 300 ppm until recently
- Carbon isotope ratios (¹³C/¹²C) provide a chemical fingerprint for fossil fuel CO₂
● Understand
- Why decreasing δ¹³C in the atmosphere is a "smoking gun" for fossil fuel burning
- How attribution science uses climate models with and without human forcing to show human causation
- What multiple converging lines of evidence mean for scientific confidence
● Can do
- Describe each line of evidence and explain what it shows about human causation
- Evaluate which type of evidence is most direct for attributing CO₂ to fossil fuel burning
- Explain how scientists distinguish human-caused from natural climate variation
In 1958, American chemist Charles David Keeling began measuring atmospheric CO₂ concentrations at the Mauna Loa Observatory in Hawaii, chosen because it is far from local pollution sources and sits at high altitude. The resulting dataset, known as the Keeling Curve, is one of the most important scientific records ever compiled.
When measurements began, atmospheric CO₂ stood at approximately 315 ppm. By 2024 the concentration had exceeded 420 ppm a rise of more than 33% in just 66 years. The record shows an unmistakable upward trend with a characteristic seasonal "wobble": CO₂ dips slightly each northern-hemisphere spring as plants grow and absorb CO₂, then rises again in autumn and winter. This oscillation shows that the monitoring is precise enough to detect the breathing of the biosphere.
The rate of increase has accelerated over time. In the 1960s CO₂ rose by roughly 0.9 ppm per year; by the 2020s the annual increase exceeded 2.4 ppm per year. This acceleration is consistent with growing global fossil fuel consumption.
Pre-industrial CO₂ (before 1750): ~280 ppm. Start of Keeling record (1958): ~315 ppm. Year 2000: ~370 ppm. Year 2024: ~422 ppm. The 280-to-315 rise happened over 200 years of industrialisation; the 315-to-422 rise happened in just 66 years. The rate is accelerating, not slowing.
Cape Grim, Tasmania: Australia runs its own baseline atmospheric monitoring station at Cape Grim on the northwest tip of Tasmania, chosen because the prevailing "Roaring Forties" winds bring clean Southern Ocean air. The CSIRO Cape Grim station has been recording CO₂ continuously since 1976, and its data mirrors the Keeling Curve, confirming the global nature of the CO₂ rise. Cape Grim data is used in IPCC assessments and has made Australia a significant contributor to global climate science.
"The seasonal wobble in the Keeling Curve shows CO₂ is naturally fluctuating, so the long-term trend isn't real." This misreads the data. The seasonal oscillation (~6 ppm peak-to-trough) is real and expected, it reflects northern-hemisphere plant growth cycles. But it rides on top of a strong upward trend that has never reversed and is not seasonal. The trend and the wobble are distinct signals operating on different timescales.
To put the Keeling Curve in long-term context, scientists drill ice cores from Antarctica and extract tiny air bubbles trapped in the ice as it formed layer by layer over hundreds of thousands of years. Cores from Vostok Station and EPICA Dome C provide a continuous record of past atmospheric CO₂ going back 800,000 years.
The results are stark: throughout this entire 800,000-year record, covering eight complete glacial-interglacial cycles, atmospheric CO₂ never exceeded 300 ppm. During cold glacial periods it fell as low as ~180 ppm; during warm interglacials it reached ~280 ppm. The current value of 420+ ppm is completely outside the range of natural variation seen over eight ice ages. It is unprecedented in at least 800,000 years.
The ice-core record also shows that CO₂ and temperature have historically moved together: higher CO₂ correlates with warmer periods. This demonstrates the connection between CO₂ and climate over geological time, even before human influence.
Natural interglacial peaks reached ~280 ppm over thousands of years. Humans have raised CO₂ from 280 to 420+ ppm in roughly 200 years. That is not just a higher value, it is occurring roughly 100 times faster than any natural CO₂ increase seen in the ice-core record. The rate is as important as the level.
Australian Antarctic Division: Australian scientists at Mawson, Davis and Casey stations participate in the global ice-core drilling program. The Australian Antarctic Division has been involved in projects extracting cores from Law Dome, East Antarctica, where annual snowfall layers allow year-by-year CO₂ resolution going back 2,000 years. This record clearly shows CO₂ was stable for 1,700 years, then began rising rapidly after 1750 with the Industrial Revolution.
"CO₂ has changed naturally in the past, therefore current changes are also natural." This is a logical fallacy (sometimes called "appeal to nature"). The ice-core record confirms past natural variation, but it also shows that current CO₂ levels are outside the entire range of that natural variation. Natural causes cannot explain a value that exceeds the natural maximum by 40%.
The most direct chemical evidence that the CO₂ increase comes from fossil fuels is the isotopic fingerprint. Carbon has three naturally occurring isotopes: ¹²C (the most abundant), ¹³C (~1.1% of all carbon), and ¹⁴C (radioactive, trace amounts). When plants carry out photosynthesis, they preferentially incorporate the lighter ¹²C over the slightly heavier ¹³C. Coal, oil and gas are formed from ancient plant (and marine organism) matter, so they are depleted in ¹³C relative to the atmosphere.
When fossil fuels are burned, this ¹³C-depleted carbon enters the atmosphere. If fossil fuel burning is adding CO₂, the atmospheric ¹³C/¹²C ratio (expressed as δ¹³C) should decrease over time. That is precisely what precise measurements show: atmospheric δ¹³C has been falling steadily since the Industrial Revolution, in parallel with rising CO₂. This is called the Suess effect, named after Hans Suess who first detected it in the 1950s.
Additionally, fossil fuels contain no ¹⁴C (they are too old, ¹⁴C decays with a half-life of 5,730 years). Measuring the decreasing proportion of ¹⁴C in atmospheric CO₂ provides further confirmation that the extra CO₂ is fossil-derived, not volcanic or oceanic.
Think of δ¹³C as a forensic fingerprint. If someone commits a crime while wearing a very specific type of mud on their boots, you can prove they were at a specific location even without a witness. Fossil-fuel CO₂ carries a specific isotopic "mud", low ¹³C, no ¹⁴C, that scientists can detect in the atmosphere. No volcano, ocean or land ecosystem has this exact signature. Only fossil fuels do.
CSIRO and Australian universities use isotopic analysis to track the carbon cycle. In Australia, isotopic measurements have been used not only to track fossil fuel emissions but also to study how Australia's land (including fire events like the 2019-20 Black Summer) affects the carbon budget. Isotopic techniques can even distinguish CO₂ from coal combustion vs natural gas combustion, helping to allocate emissions precisely for policy purposes.
"Volcanoes release huge amounts of CO₂, they could be responsible for the rise." In fact, global volcanic CO₂ emissions are estimated at 0.1–0.3 billion tonnes per year, while human fossil fuel emissions exceeded 37 billion tonnes in 2023, more than 100 times larger. And volcanic CO₂ does not have the isotopic signature (low δ¹³C, no ¹⁴C) seen in the growing atmospheric surplus. Both the quantity and the fingerprint rule out volcanoes.
Attribution science uses climate models to determine whether an observed change (like global warming) would have occurred without a particular forcing factor. The key method is to run two parallel simulations of the climate system: one with only natural forcings (solar variation, volcanic eruptions) and one with both natural forcings and human forcings (greenhouse gas emissions and aerosols).
The results are clear: models run with only natural forcings cannot reproduce the observed warming since 1950. The natural-only models predict roughly stable temperatures, or slight cooling after major volcanic eruptions. Only when human greenhouse gas emissions are added to the models does the simulated temperature record match the observed one closely. The IPCC (Intergovernmental Panel on Climate Change) Sixth Assessment Report (2021) concluded it is "unequivocal" that human influence has warmed the climate.
Additional independent lines of evidence support this conclusion: the NASA GISS surface temperature record and the Berkeley Earth project (which used a different statistical methodology) both show the same ~1.1°C warming trend above pre-industrial levels. Ocean heat content has increased in every decade since the 1970s. Sea level rise has been accelerating, with the rate roughly doubling from the 20th-century average to the 21st-century average.
Attribution science doesn't just look at global average temperature. Greenhouse gas warming produces specific "fingerprint" patterns: the stratosphere should cool while the troposphere warms (the opposite of what solar changes would produce); warming should be greater at the poles than at the equator; nights should warm faster than days. All of these predicted fingerprints have been observed in the data, ruling out natural explanations and confirming greenhouse gas forcing.
ACCESS Climate Model, Australia's contribution: Australia's Bureau of Meteorology and CSIRO jointly develop the Australian Community Climate and Earth System Simulator (ACCESS). Australian climate scientists use ACCESS to run attribution analyses of specific Australian weather events, for example, showing that the 2019-20 Black Summer bushfire season was made more likely by climate change, and that record heat events in Australia are now occurring far more frequently than they would without greenhouse gas emissions. These studies directly inform Australian emergency management and policy.
"It's just models, models can be wrong." Climate models are imperfect, but their skill has been validated. Models from the 1990s accurately predicted the warming that has since been observed. Attribution science uses the models not to predict exact temperatures, but to test the counterfactual: "what would temperatures look like without human emissions?" The models' ability to replicate the observed warming only with human forcing is a powerful test, not just a circular argument.
Wrong: "Scientists only have one piece of evidence for human-caused climate change." There are multiple independent lines: direct CO₂ measurement, ice cores, isotopic fingerprints, ocean heat content, sea-level rise, and attribution modelling, all pointing in the same direction.
Right: Human-caused climate change is supported by at least five independent lines of evidence that all converge on the same conclusion. This "consilience" greatly increases scientific confidence, it is very unlikely that five unrelated methods would all give the same wrong answer.
Wrong: "The Sun is causing global warming." Solar output has been slightly declining since the 1980s, even as global temperatures have risen. Attribution models confirm that solar variation explains less than 0.1°C of the observed ~1.1°C warming.
Right: Solar activity has been monitored by satellites since 1979 and has shown no net upward trend. The observed warming pattern, with cooling in the stratosphere and warming in the troposphere, is the opposite of what increased solar output would produce, and is precisely what greenhouse gas theory predicts.
Wrong: "CO₂ only makes up 0.04% of the atmosphere, how can such a tiny amount matter?" A small quantity can have a large effect. A few milligrams of paracetamol can treat pain; a few micrograms of botulinum toxin can be lethal. Effect depends on mechanism, not just concentration. CO₂ absorbs infrared radiation efficiently at its specific wavelengths.
Right: Even at trace concentrations, CO₂ is an effective greenhouse gas because it absorbs and re-emits infrared radiation at specific wavelengths. The physics has been understood since Eunice Newton Foote (1856) and John Tyndall (1859). A 50% increase in CO₂ concentration (280 to 420 ppm) represents a very significant change in radiative forcing.
Australia's Role in Climate Science
Cape Grim Baseline Air Pollution Station in northwest Tasmania is one of the world's most important clean-air monitoring sites. Scientists from CSIRO and the Bureau of Meteorology have tracked greenhouse gases, aerosols and ozone-depleting substances there since 1976. Its data forms part of the global network that underpins IPCC assessments.
Australian Antarctic Division drilling projects at Law Dome and other sites have extended the ice-core CO₂ record with exceptionally fine time resolution, helping scientists pin down exactly when CO₂ began rising relative to the Industrial Revolution.
Indigenous knowledge and climate change: Aboriginal and Torres Strait Islander peoples have observed changing seasons, shifted plant and animal behaviour, and altered fire regimes over generations. This long-term observational knowledge increasingly complements scientific records, providing localised context for climate change impacts that instrumental records cannot fully capture.
Evidence Strength Ranking
Responding to Sceptical Claims
If the CO₂ rise in the atmosphere was caused by natural ocean outgassing rather than fossil fuels, what would you expect to happen to the δ¹³C ratio of atmospheric CO₂? Would it decrease, increase, or stay the same?
How close was your prediction?
The hook for this lesson revealed something clever: every CO₂ molecule has an isotopic "fingerprint" baked in at the moment it was made. Fossil fuels are ancient, and their carbon isotope signature is distinct, meaning scientists can actually detect fossil-fuel CO₂ specifically in today's atmosphere, as if each molecule carries a barcode proving its origin.
Now that you've worked through the full picture, look back at your earlier thinking. Did you anticipate that isotopic fingerprinting could be used this way? Compare your original list of evidence with what you've now discovered.
Earlier you were asked: What evidence would you need to prove that humans are responsible for rising CO₂ levels?
Now that you've worked through the lesson, write a fuller answer. How many independent lines of evidence did you discover, and which do you find most convincing?
Q1. Explain how the isotopic composition of atmospheric CO₂ (specifically the δ¹³C ratio) provides evidence that the increase in atmospheric CO₂ is caused by burning fossil fuels rather than by volcanic activity or ocean outgassing.
Q2. Describe what Antarctic ice-core records reveal about CO₂ levels over the past 800,000 years. Explain why this information strengthens the conclusion that current CO₂ levels are caused by human activity.
Q3. Explain what attribution science is and how it is used to show that observed global warming since 1950 is caused by human greenhouse gas emissions. In your answer, describe both what models with only natural forcings predict and what models with human forcings predict.
Revisit Your Thinking
Go back to your Think First answer. Has your understanding changed?
- Can you now name at least four independent lines of evidence for human-caused climate change?
- Which line of evidence do you find most scientifically compelling, and why?
Model answers (click to reveal)
Answers
▾MCQ 1
C The Keeling Curve shows continuous unambiguous rise from 315 ppm to 420+ ppm, and combined with ice-core data, current levels are unprecedented in at least 800,000 years.
MCQ 2
B Plants use ¹²C preferentially in photosynthesis (mass discrimination). Ancient plant/marine organic matter compressed into coal/oil/gas retains this ¹³C-depleted signature.
MCQ 3
D Attribution science prediction: natural-only models do not reproduce observed warming; only models with human greenhouse gas emissions match the observed temperature trend since ~1950.
MCQ 4
A Across eight glacial cycles the maximum was ~280 ppm; current levels of 420+ ppm are without precedent in 800,000 years of natural variation.
MCQ 5
B Both the scale (human emissions ~37 Gt CO₂/yr vs volcanic ~0.3 Gt/yr) and the isotopic fingerprint (low δ¹³C, no ¹⁴C) identify the source as fossil fuels.
Short Answer 1
Model answer: Plants preferentially absorb the lighter ¹²C isotope during photosynthesis, making their carbon depleted in ¹³C relative to the atmosphere. Since fossil fuels (coal, oil, gas) are formed from ancient plant and marine organic matter, they inherit this ¹³C-depleted signature. When fossil fuels are burned, this ¹³C-depleted CO₂ enters the atmosphere, causing the atmospheric ¹³C/¹²C ratio (δ¹³C) to decrease over time, a phenomenon called the Suess effect. Volcanic CO₂ and ocean outgassing do not have this same depleted signature, so the steadily declining δ¹³C points specifically to fossil fuel combustion as the source of rising atmospheric CO₂.
Short Answer 2
Model answer: Antarctic ice cores (e.g., from Vostok and EPICA Dome C) contain trapped air bubbles that allow scientists to reconstruct atmospheric CO₂ going back 800,000 years, covering eight glacial-interglacial cycles. Throughout this record, CO₂ varied between ~180 ppm (cold glacials) and ~280 ppm (warm interglacials), never exceeding 300 ppm. Current atmospheric CO₂ of 420+ ppm is therefore completely outside the range of natural variability over 800,000 years. Furthermore, the current rate of increase (~2.4 ppm/year) is approximately 100 times faster than any natural transition in the ice-core record. This makes it extremely unlikely that current elevated CO₂ is due to natural processes, strongly implicating human activity.
Short Answer 3
Model answer: Attribution science uses climate models to test whether an observed change can be explained by a specific cause. Scientists run two sets of model simulations: one including only natural forcings (solar variation and volcanic eruptions) and one including both natural and human forcings (primarily greenhouse gas emissions). Models with only natural forcings predict that global average temperature since 1950 would have remained roughly stable or declined slightly after major volcanic eruptions, they cannot reproduce the observed ~1.1°C warming trend. In contrast, models that include human greenhouse gas emissions closely match the observed temperature record. This "model detective" approach, combined with consistent results from multiple independent research groups (NASA GISS, Berkeley Earth) and the specific patterns of warming (tropospheric warming paired with stratospheric cooling, greater polar warming), provides very strong evidence that human emissions are the primary driver of observed global warming since the mid-20th century.