Climate & Environmental Science

In 1958 a geochemist named Charles David Keeling began measuring carbon dioxide on a Hawaiian volcano. Within two years his instrument showed something nobody had recorded before: the whole planet was breathing. Every northern spring the CO₂ reading fell as forests leafed out, and every autumn it rose again as leaves decayed. Riding on that annual pulse was a steady climb. One instrument on one mountain had caught a global system in motion.

The big picture

Environmental science treats the Earth not as a list of separate topics but as one connected system: a set of parts that interact so the whole behaves differently from any part alone. The reason this matters is simple. If everything is linked, then a push in one place shows up somewhere else, often later and larger than you expect.

Every big idea in this course, from the greenhouse effect to sea-level rise, is really a story about stocks, flows, and feedbacks. Learn that vocabulary once here, and the rest of the course becomes a set of examples.

Systems thinking is not vague holism. It is a precise bookkeeping habit. You name the quantity you care about, you name the rates that add to it and drain it, and you look for the loops that connect the two. Do that faithfully and a surprising number of environmental arguments settle themselves, because most bad arguments about climate turn out to be bookkeeping errors dressed up as opinions.

Stocks, flows, and boundaries

Three plain ideas describe any system. A stock is the amount of something stored right now, like the water sitting in a bathtub, or the carbon dissolved in the ocean. A flow is the rate at which that something moves in or out, like the faucet filling the tub and the drain emptying it. A boundary is the imaginary line you draw around the parts you are studying, separating the system from everything outside it.

Think of a bank account. The balance is the stock. Deposits and withdrawals are the flows. If deposits beat withdrawals the balance climbs; if they match, it holds steady. The same picture works for carbon in the air, heat in the ocean, or water in a reservoir.

Notice the units, because they do a lot of the work. A stock is a plain amount: liters, gigatonnes of carbon. A flow is an amount per unit of time: liters per minute, gigatonnes of carbon per year. Whenever someone quotes an environmental number, ask which of the two it is.

Key idea: A stock is a quantity stored; a flow is the rate it changes; whether a stock grows or shrinks depends only on the balance of its flows.

Residence time: how long something stays put

Pair a stock with its outflow and you get a third quantity that predicts how sluggish a system will be. Residence time is the average length of time a unit of material spends in a reservoir, and for a system in rough balance it equals the stock divided by the outflow rate.

Work an example. A lake holds 600 million cubic meters of water and its river drains 200 million cubic meters a year. The residence time is 600 divided by 200, which is 3 years, so a pollutant dumped today is still flushing out three years later. Change the drain to 6 million a year and the residence time becomes 100 years, turning the same spill into a century-long problem.

That calculation explains much of the urgency in climate policy. Fine particles from a smokestack wash out of the air in days, so cleaning them up brings quick relief. Carbon dioxide is different: no single number describes it, because the ocean absorbs some within decades while a meaningful fraction stays airborne for centuries to millennia.

Key idea: Residence time equals stock divided by outflow, and it predicts how long a system stays disturbed after the disturbance stops.

Open, closed, and isolated systems

Systems differ by what is allowed to cross the boundary. An open system exchanges both energy and matter with its surroundings. A lake is open: it gains sunlight and rainwater and loses water vapor. A closed system exchanges energy but essentially no matter. An isolated system exchanges neither, an idealization that almost nothing in nature reaches.

On the scale of a human lifetime, planet Earth is very nearly a closed system. Sunlight enters and heat leaves, so energy flows through. But the total amount of matter, the water, carbon, and nitrogen, stays almost constant apart from a trickle of meteorite dust arriving and a little hydrogen escaping to space. That single fact, energy flows through while matter cycles within, organizes the entire course.

Key idea: Earth is effectively closed on human timescales: energy passes through it, but its matter is recycled rather than replaced.

Equilibrium, steady state, and why "balance" is a verb

People often picture a healthy environment as one sitting perfectly still. Systems scientists picture something busier. A steady state is a stock that stays roughly constant because large flows in and out happen to cancel, not because nothing is moving. A river-fed lake at a constant level is in steady state while millions of liters pass through it every day.

The distinction matters because a steady state can be broken from either side. You can raise a lake by opening the faucet wider, or just as effectively by partly blocking the drain. Both routes appear in this course: burning fossil fuels opens the CO₂ faucet, while clearing forests and warming the ocean weaken the drains. When a system is pushed out of steady state, the push is called a forcing and the settling into a new balance is called the response. Module 2 uses exactly those words for the climate.

Key idea: A steady state means inflows and outflows cancel, not that nothing moves, so a stock can rise either because inflow grew or because outflow shrank.

Feedbacks: the engine of system behavior

A feedback is what happens when a change loops back to affect itself. There are two kinds, and the words describe direction only, not good or bad.

• A negative (stabilizing) feedback opposes the original change and holds the system steady. A thermostat is the classic example: when a room gets too warm, it shuts off the furnace, and the room cools back down.
• A positive (amplifying) feedback reinforces the change and pushes the system further in the same direction. Think of a microphone held too close to its own loudspeaker: a small sound gets amplified, comes out louder, is picked up again, and squeals out of control.

Much of climate science is really the bookkeeping of which feedbacks dominate. Here is one we will meet again: as bright ice melts it exposes darker ocean, the darker ocean absorbs more sunlight, the extra warmth melts more ice. That is an amplifying loop.

Key idea: Negative feedbacks stabilize a system like a thermostat; positive feedbacks amplify a change like a microphone squeal.

How much does a feedback amplify?

Amplifying feedbacks do not usually run away forever. Most of them settle on a finite multiplier, and a simple arithmetic model shows why. Suppose a push warms a system by 1 unit, and that warming triggers a loop returning half the change as extra warming. The total is 1 plus 0.5 plus 0.25 plus 0.125, and so on. That series converges on 2. The feedback doubled the original push and then stopped.

The general pattern is that a loop returning a fraction f of each round multiplies the original push by 1 divided by (1 minus f). At f of two thirds the multiplier is 3. As f approaches 1 the multiplier explodes, so runaway behavior needs a loop that returns nearly everything it receives, which is rare in nature.

This is roughly how climate scientists reason about why doubling carbon dioxide warms the planet by about 3 degrees Celsius rather than the roughly 1 degree the added gas produces on its own. The known amplifying loops, chiefly water vapor and shrinking ice, roughly triple the direct effect. Lesson 12 does this with real numbers.

Key idea: A loop that returns a fraction f of each round multiplies the original push by 1/(1 - f), so strong feedbacks amplify greatly without running away.

Lags, thresholds, and tipping points

Systems also surprise us in three ways. A lag time means a response can arrive years after its cause, so the system seems calm long after the push has been given. A threshold is a level past which behavior changes sharply. A tipping point is a threshold beyond which a system flips into a new state that is hard, or impossible, to reverse, the way a canoe leans and leans and then suddenly capsizes. Delays and thresholds together make systems hard to steer, because the damage from a push may not be visible until it is too late to take the push back.

Key idea: Delays hide consequences and thresholds can trigger abrupt, sometimes irreversible change, so acting only after harm appears is often acting too late.

A small worked example

Imagine a bathtub holding 100 liters (the stock). The faucet adds 5 liters per minute and the drain removes 3 liters per minute. The net flow is 5 minus 3, which is +2 liters per minute, so after 10 minutes the tub holds 100 plus 2 times 10, which is 120 liters. Now suppose the drain slows to 1 liter per minute. The net flow rises to 5 minus 1, which is +4 liters per minute, and the tub fills twice as fast even though the faucet never changed. This is exactly how slowing a carbon sink, not just adding emissions, can speed the buildup of CO₂.

Push the same tub one step further and you get the most important policy result in the course. To hold the level steady rather than merely slow its rise, the faucet must come down to 3, matching the drain. Cutting it from 5 to 4 halves the rate of rise, but the tub still fills. Swap the labels and this is the argument for net-zero emissions in Lesson 18: halving global emissions slows the growth of atmospheric CO₂ without stopping it. Nothing in that conclusion depends on climate physics; it follows from the arithmetic of a tub.

Nonlinearity, and why systems fool people

Human intuition is trained on straight lines: twice the cause, twice the effect. Systems rarely oblige. A nonlinear response is one where output is not proportional to input, and environmental systems are full of them. A river channel absorbs rain with no trouble until the water reaches the top of the bank, at which point a small extra amount of rain produces an enormous change in outcome.

Three habits of mind help. Look for the accumulating stock rather than the eye-catching flow, because the stock usually causes the harm. Ask where the delays are, since delay hides consequences. Ask what the system does to itself, because loops generate most surprising behavior. A fishery, a groundwater basin, and the atmosphere differ wildly in detail, yet each is a stock filled and drained by flows and steered by feedbacks.

Key idea: Environmental systems respond nonlinearly, so the useful questions are what is accumulating, where the delays are, and which loops the system runs on itself.

Common misconceptions

• "Positive feedback is good and negative feedback is bad." No. The labels describe whether a loop amplifies or opposes a change, not whether the outcome is desirable.
• "A stock and a flow are the same thing." No. A stock is an amount at one moment; a flow is a rate over time. Water in the tub versus water per minute.
• "If a cause stops, the effect stops immediately." Not with lags. Because of delays, a system can keep changing long after the original push ends.
• "Cutting a flow in half makes the stock go down." No. It only makes the stock rise more slowly. The stock falls only when outflow exceeds inflow.
• "An amplifying feedback must run away." No. A loop returning a fraction of each round converges on a finite multiplier; runaway needs a loop that returns nearly everything.

Recap

• A system is interacting parts whose whole behaves differently from its pieces.
• Stocks store, flows change stocks, and a boundary defines the system.
• Residence time equals stock divided by outflow and predicts how long a disturbance lingers.
• Earth is effectively closed: energy flows through, matter cycles within.
• A steady state means flows cancel, so a stock can rise from a bigger inflow or a weaker outflow.
• Negative feedbacks stabilize; positive feedbacks amplify by a factor of 1/(1 - f).
• Lags, thresholds, and nonlinearity make systems hard to manage and can produce abrupt change.

Sources

1. Meadows, D. H. (2008). Thinking in systems: A primer. Chelsea Green Publishing. find source ↗
2. UCAR Center for Science Education. (n.d.). Earth as a system. scied.ucar.edu
3. NASA Earth Observatory. (n.d.). Earth Observatory: Earth system features. NASA Science. earthobservatory.nasa.gov
4. Chen, D., Rojas, M., Samset, B. H., et al. (2021). Chapter 1: Framing, context, and methods. In Climate change 2021: The physical science basis (IPCC AR6 WGI). ipcc.ch
5. Canadell, J. G., Monteiro, P. M. S., Costa, M. H., et al. (2021). Chapter 5: Global carbon and other biogeochemical cycles and feedbacks. In IPCC AR6 WGI (on carbon reservoirs, fluxes, and the long CO₂ tail). ipcc.ch
6. NOAA Global Monitoring Laboratory. (2026). Trends in atmospheric carbon dioxide (Keeling curve, Mauna Loa and global means). gml.noaa.gov
7. Forster, P., Storelvmo, T., Armour, K., et al. (2021). Chapter 7: The Earth's energy budget, climate feedbacks, and climate sensitivity. In IPCC AR6 WGI. ipcc.ch

Look at the famous Apollo 17 photograph of Earth taken in 1972 and you can see the whole system in one frame: white swirls of cloud, blue ocean, a white cap of Antarctic ice, brown continent, and a green fringe of vegetation along the African coast. Nothing in that picture is separate from anything else. The clouds came from the ocean, the ice came from the clouds, the green depends on both, and the brown land is being slowly eaten by all three. Earth science begins by naming those parts, and immediately proceeds to studying where they meet.

The big picture

Scientists split the Earth system into a handful of interacting parts called spheres, one for the air, one for the water, one for the ice, one for the rock, and one for life. The point is not to file topics into tidy boxes but to see how the boxes leak into each other.

Almost every environmental event, a rainstorm, a wildfire, a warming ocean, is really two or more spheres trading energy and matter. Get comfortable with the spheres and their handoffs, and later chapters on carbon, weather, and climate change will feel like variations on one theme.

The five spheres

Here are the five, each defined plainly with an everyday touchstone.

• Atmosphere, the envelope of gases held to Earth by gravity, like a thin shell of air you would pop through in a fast elevator ride to space.
• Hydrosphere, all the liquid and vapor water: oceans, lakes, rivers, groundwater, and the moisture in the air.
• Cryosphere, the frozen water: ice sheets, glaciers, sea ice, and permafrost. Think of it as the planet's freezer.
• Lithosphere (or geosphere), the rocky crust and interior, including the soils that blanket it.
• Biosphere, the sum of all living things and the zones they occupy, a thin film of life smeared across the other four.

Key idea: The spheres are the air, water, ice, rock, and life of the planet, and they overlap rather than sit in separate boxes.

Putting numbers on the spheres

Sizes are worth knowing, because they explain which sphere pushes which. The atmosphere has a mass of roughly 5.1 x 10^18 kilograms, which sounds enormous until you compare it with the ocean at about 1.4 x 10^21 kilograms, some 270 times heavier. That single ratio is why the ocean, not the air, sets the pace of climate change.

The hydrosphere is also badly misnamed if you picture drinking water. According to the U.S. Geological Survey, about 96.5 percent of Earth's water sits in the oceans as salt water. Roughly 1.7 percent is locked in glaciers and ice caps, and another 1.7 percent is groundwater. Everything else, all the rivers, lakes, soil moisture, and atmospheric vapor combined, amounts to well under 1 percent of the total.

Work through what that means. If all of Earth's water were a 100-liter drum, the ocean would be 96.5 liters, ice would be about 1.7 liters, groundwater about 1.7 liters, and the liquid fresh surface water that supports nearly all terrestrial life would be less than a teaspoon and a half. The freshwater the biosphere depends on is a rounding error in the planet's water budget.

The biosphere is smaller still by mass. A 2018 census of global biomass estimated that all living things together hold roughly 550 gigatonnes of carbon, with plants making up about 450 of that and all animals about 2. Yet this thin film of life moves carbon in and out of the atmosphere at more than a hundred gigatonnes a year. Small stock, enormous flow, which is exactly the distinction Lesson 1 insisted on.

Key idea: The ocean outweighs the atmosphere roughly 270 to 1 and holds 96.5 percent of Earth's water, while the biosphere is tiny in mass but enormous in the flows it drives.

Coupling: where the action is

A coupling is an interaction through which two or more spheres exchange energy or matter. This is the interesting part, because nearly every process you care about is a coupling. A few concrete examples:

• A rainstorm is the atmosphere and hydrosphere swapping water.
• A forest fire is the biosphere releasing carbon into the atmosphere.
• Weathering of rock by rainwater is the hydrosphere slowly pulling carbon dioxide out of the atmosphere and into the lithosphere over geological time.
• The oceans (hydrosphere) absorbing heat and carbon from the atmosphere, with marine plankton (biosphere) influencing how much.

Because the spheres are coupled, a push on one ripples through the others. That is precisely why adding a greenhouse gas to the atmosphere ends up changing the ice, the oceans, and living things.

Couplings also run in both directions, which is what turns them into the feedbacks of Lesson 1. Sea ice is the cleanest illustration. Warming air (atmosphere) melts sea ice (cryosphere), exposing dark water (hydrosphere) that absorbs more sunlight and warms the air further. The arrow goes atmosphere to cryosphere to hydrosphere and straight back to atmosphere, closing a loop. Whenever you can trace a chain of couplings that returns to where it started, you have found a feedback, and it is worth asking whether it amplifies or damps.

Key idea: A change forced on one sphere does not stay put; through couplings it spreads to the others, and a chain of couplings that closes on itself is a feedback.

Sphere	What it is	Example interaction
Atmosphere	Gases surrounding Earth	Delivers rain to the land
Hydrosphere	All liquid and vapor water	Oceans absorb atmospheric heat and CO2
Cryosphere	All frozen water	Reflects sunlight; stores fresh water
Lithosphere	Rock, crust, and soil	Weathering removes CO2 over eons
Biosphere	All living organisms	Photosynthesis and respiration cycle carbon

Tracing a signal through the spheres

Here is the master example of the whole course, told as a chain. Buried carbon in the lithosphere (coal, oil, gas) is burned, adding CO₂ to the atmosphere. The thicker blanket of greenhouse gas warms the hydrosphere (warmer, higher, more acidic oceans) and shrinks the cryosphere (melting ice). Those changes shift where and how the biosphere can live. One push at the start, felt in all five spheres. Following such chains is what environmental science does.

Key idea: Burning fossil fuels sends a single signal from rock through air into water, ice, and life, which is why its effects are so wide-ranging.

Each sphere runs on its own clock

Couplings would be easy to trace if every sphere responded at the same speed. They do not, and the mismatch is the source of most of the delay built into climate change. The atmosphere mixes globally in about a year, so a gas released in Ohio shows up in New Zealand within roughly twelve months. The surface ocean exchanges with the air over years to decades. The deep ocean turns over on a timescale near a thousand years. Ice sheets respond over centuries to millennia, and the rock cycle takes millions of years.

Stack those clocks and you get the pattern this course keeps returning to. A change made in the fast sphere, the atmosphere, is committed into the slow ones long before it finishes playing out. Emissions stopped today would leave the air responding within years, the surface ocean within decades, and the ice sheets and deep ocean still adjusting for centuries. That is not a prediction about policy; it is a statement about mixing times.

Key idea: The spheres respond on wildly different clocks, from a year for the atmosphere to millennia for deep ocean and ice, which is why climate change carries so much built-in delay.

Boundaries are a choice, not a fact

One practical warning. The five spheres are a convention, not a law of nature, and where you draw a boundary depends on the question you are asking. Soil is a good test case: it contains mineral grains from the lithosphere, water from the hydrosphere, air in its pore spaces, and a dense population of roots, fungi, and bacteria from the biosphere. Filing soil under any one sphere loses the point.

Researchers therefore sometimes speak of the critical zone, the thin layer from the tops of the trees down to the base of groundwater where all the spheres interact and where essentially all terrestrial life happens. Other groupings are equally legitimate. Some texts fold the cryosphere into the hydrosphere, since ice is simply frozen water, and some add an anthroposphere for human infrastructure. A boundary is a tool. Choose the one that makes the coupling you care about visible.

Key idea: Sphere boundaries are analytical choices, and features like soil sit in several at once, which is why researchers also talk about a critical zone where all of them overlap.

Common misconceptions

• "Each sphere can be studied on its own." Only up to a point. Because the spheres constantly exchange energy and matter, ignoring the couplings misses most of the important behavior.
• "The cryosphere is a separate thing from water." The cryosphere is simply the frozen part of the hydrosphere, water in solid form.
• "The biosphere is small and passive." Life is thin but powerful: it helped create the oxygen atmosphere and moves enormous amounts of carbon every year.
• "Most of Earth's water is fresh and available." No. About 96.5 percent is salty ocean, and the liquid fresh surface water on which land life depends is a small fraction of one percent.
• "The five spheres are fixed scientific categories." They are a useful convention. Soil belongs to four of them at once, which is why researchers also use the idea of a critical zone.

Recap

• The Earth system is divided into five spheres: atmosphere, hydrosphere, cryosphere, lithosphere, and biosphere.
• The ocean outweighs the atmosphere about 270 to 1 and holds 96.5 percent of Earth's water.
• The biosphere holds roughly 550 gigatonnes of carbon yet drives flows of over a hundred gigatonnes a year.
• A coupling is an exchange of energy or matter between spheres, and most environmental processes are couplings.
• Each sphere responds on its own clock, from about a year for the air to millennia for deep ocean and ice.
• Sphere boundaries are analytical choices; soil and the critical zone straddle several at once.
• Fossil-fuel burning is a signal that travels from rock through air into water, ice, and life.

Sources

1. NASA Earth Observatory. (n.d.). Earth system features. NASA Science. earthobservatory.nasa.gov
2. UCAR Center for Science Education. (n.d.). Earth as a system. scied.ucar.edu
3. U.S. Geological Survey. (n.d.). Where is Earth's water? Water Science School. usgs.gov
4. Bar-On, Y. M., Phillips, R., & Milo, R. (2018). The biomass distribution on Earth. PNAS, 115(25), 6506-6511. pmc.ncbi.nlm.nih.gov
5. National Snow and Ice Data Center. (n.d.). Ice sheets. nsidc.org
6. NOAA National Ocean Service. (n.d.). How much water is in the ocean? oceanservice.noaa.gov
7. Canadell, J. G., Monteiro, P. M. S., Costa, M. H., et al. (2021). Chapter 5: Global carbon and other biogeochemical cycles and feedbacks. In IPCC AR6 WGI. ipcc.ch

A satellite called CERES has been staring at Earth since 1997 with one job: to weigh the sunlight coming in and the heat going out, and report the difference. The difference is small, less than half a percent of the traffic in either direction, and it is the most consequential small number in climate science. It is the reason the planet is warming, and everything in this lesson is an attempt to explain where that number comes from.

The big picture

Earth's temperature is set by a simple balance: over the long run, the energy the planet takes in from the Sun must equal the energy it sends back to space as heat. That balance is called the energy budget, and it is just the accountant's rule that income must match spending, applied to a planet.

Almost all of that income arrives as sunlight. When something tips the balance, the planet warms or cools until the books balance again. Understanding this one idea is the foundation for understanding the greenhouse effect and global warming.

The Sun is the engine

Essentially all the energy driving Earth's climate arrives as solar radiation, the light and near-light streaming from the Sun. A tiny amount of heat also leaks out of Earth's hot interior, but it is thousands of times smaller than the solar input, so we treat the Sun as the engine and the interior as a rounding error.

The planet loses energy the only way an object in the vacuum of space can: by radiating it away. Because Earth is far cooler than the Sun, it radiates in the invisible infrared band, the same kind of warmth you feel from a radiator without seeing any glow.

Key idea: Sunlight is essentially the sole energy input, and the only way out is radiating heat to space as infrared.

The three numbers that run the planet

The budget is usually quoted in watts per square meter, a rate of energy delivered per unit area. Three numbers carry most of the story, and they are worth memorising.

Facing the Sun at Earth's distance, a square meter receives about 1,361 watts, a quantity called the solar constant. But the planet does not present a flat square meter to the Sun; it presents a sphere. Sunlight is intercepted by a disc of area pi times r squared, while the heat is radiated from a whole sphere of area 4 pi times r squared. Dividing the first by the second gives one quarter, so the globally averaged incoming sunlight is 1,361 divided by 4, which is about 340 watts per square meter. That factor of four is simply the geometry of a round planet lit from one side and radiating in every direction.

Of those 340 watts, roughly 100 are reflected straight back to space by clouds, bright surfaces, and airborne particles. That leaves about 240 watts per square meter absorbed by the Earth system, and in a balanced budget the planet must radiate the same 240 watts back to space as infrared. Incoming 340, reflected 100, absorbed and re-emitted 240. Everything else in climate science is a refinement of that accounting.

Key idea: Earth intercepts about 340 W/m2 of sunlight on average, reflects about 100, and must radiate the remaining 240 back to space as infrared heat.

Albedo: how much sunlight bounces off

Not all sunlight is absorbed. The fraction reflected straight back to space is the albedo, a number from 0 (perfectly dark, absorbs everything) to 1 (perfectly bright, reflects everything). It works just like clothing on a sunny day: a white shirt (high albedo) stays cool because it reflects sunlight, while a black shirt (low albedo) gets hot because it absorbs it.

Bright surfaces reflect a lot: fresh snow bounces back roughly 80 to 90 percent of the light hitting it. Dark surfaces reflect little: the open ocean reflects only about 6 percent and absorbs the rest. Clouds, deserts, and forests fall in between. Averaged over the whole planet, Earth's albedo is about 0.30, meaning roughly 30 percent of incoming sunlight is reflected and 70 percent is absorbed and later re-emitted as heat.

Key idea: Albedo is the reflected fraction of sunlight, like a white shirt versus a black shirt; Earth's average is about 0.30.

Clouds and aerosols: the reflective wildcards

Roughly two thirds of Earth's reflection is contributed not by the surface but by the atmosphere itself, principally by clouds and by suspended microscopic particles called aerosols. Clouds are simultaneously the planet's brightest reflectors and effective absorbers of infrared radiation, so their net influence depends on their altitude, thickness, and droplet size. Low, thick marine stratocumulus predominantly cools by reflecting sunlight, whereas high, thin cirrus predominantly warms by intercepting outgoing heat.

Aerosols complicate the accounting further. Sulfate particles emitted alongside fossil-fuel combustion scatter incoming sunlight and also seed additional cloud droplets, producing a cooling influence that has partially masked greenhouse warming. Soot, or black carbon, does the opposite by absorbing sunlight directly. The IPCC assesses aerosol effects as the largest single source of uncertainty in present-day radiative forcing, which is precisely why Lesson 12 identifies clouds as the dominant unknown in climate sensitivity.

Key idea: Clouds and aerosols supply most of Earth's reflectivity and can warm or cool depending on their properties, making them the least certain term in the energy budget.

Surface	Typical albedo	Fraction absorbed
Fresh snow	0.80 to 0.90	0.10 to 0.20
Sea ice	about 0.50 to 0.70	about 0.30 to 0.50
Desert sand	about 0.40	about 0.60
Grassland	about 0.25	about 0.75
Conifer forest	about 0.10	about 0.90
Open ocean	about 0.06	about 0.94

A small worked example

Suppose a patch of ground reflects 30 percent of the sunlight it receives, so its albedo is 0.30. Then it absorbs 1 minus 0.30, which is 0.70, or 70 percent. Now cover it in fresh snow with an albedo of 0.85. It now absorbs only 1 minus 0.85, which is 0.15, or 15 percent. Snow cutting absorption from 70 percent to 15 percent is why a snowy field feels so much cooler in sunlight, and why losing bright ice makes the planet absorb more energy.

Calculating Earth's bare temperature

The budget lets you predict a planet's temperature from first principles. Any object radiates in proportion to the fourth power of its absolute temperature, a relation called the Stefan-Boltzmann law, written as energy per square meter equals a constant (5.67 x 10^-8) times temperature in kelvin raised to the fourth power.

Set the outgoing radiation equal to the 240 watts per square meter Earth absorbs and solve for temperature. Dividing 240 by 5.67 x 10^-8 gives about 4.23 x 10^9, and taking the fourth root of that gives about 255 kelvin, which is roughly minus 18 degrees Celsius. That is what Earth's temperature would be with the same albedo but no heat-trapping gases.

The measured global average surface temperature is about 15 degrees Celsius. The gap of 33 degrees Celsius between the calculation and reality is the natural greenhouse effect, and Lesson 5 explains the mechanism. Notice how much this simple sum accomplishes: it shows that greenhouse warming is not an optional add-on but a quantity forced on us by the difference between two independent measurements.

Key idea: Balancing 240 W/m2 against the Stefan-Boltzmann law predicts a surface at minus 18 degrees C, and the 33 degrees C gap to the observed 15 degrees C is the natural greenhouse effect.

When the budget is out of balance

If the absorbed sunlight exactly equals the outgoing heat, temperature holds steady. If something makes Earth absorb more than it radiates, by adding a gas that traps heat, or by lowering albedo when bright ice melts, there is a net energy imbalance. The planet then warms, and as it warms it radiates more strongly, until the extra outgoing heat closes the gap and balance is restored at a new, higher temperature.

Careful satellite measurements show Earth is currently absorbing more energy than it emits. The IPCC's Sixth Assessment Report assessed this Earth energy imbalance at about 0.79 watts per square meter for 2006 to 2018, with a likely range of roughly 0.5 to 1.1. Later satellite analyses indicate the imbalance has continued to grow since then, so treat any single figure as a snapshot rather than a constant and check the current CERES record for an up-to-date value.

Put 0.79 watts per square meter next to the 240 watts flowing each way and you see how small the surplus is: about one third of one percent of the traffic. Yet spread over Earth's 5.1 x 10^14 square meters and sustained year after year, it amounts to an energy accumulation equivalent to several Hiroshima-sized bombs every second. Small rates applied to enormous areas over long times produce large totals, which is exactly the stock-and-flow logic of Lesson 1. Over 90 percent of that accumulating energy ends up in the ocean.

Key idea: Earth's energy imbalance is small in percentage terms, assessed near 0.79 W/m2 for 2006-2018 and rising since, but sustained over a planet it accumulates enormous quantities of heat, mostly in the ocean.

Common misconceptions

• "Earth's internal heat drives the climate." No. Geothermal heat reaching the surface averages under 0.1 watt per square meter, thousands of times weaker than sunlight.
• "A high-albedo surface has no temperature." No. High albedo just means it reflects most sunlight and absorbs little, so it heats up less, not that it is somehow frozen at zero.
• "If Earth absorbs extra energy it warms forever." No. Warming raises the outgoing heat, which eventually rebalances the budget at a higher temperature, unless the forcing keeps growing.
• "The energy imbalance is too small to matter." No. A fraction of a watt per square meter applied to the whole planet for decades is what has warmed the ocean and melted the ice.
• "Changes in the Sun explain the current imbalance." No. Satellite records show total solar output has been flat or slightly declining since the 1980s while the imbalance grew.

Recap

• Sunlight is essentially the only energy input; Earth loses energy as infrared heat.
• Averaged over the globe, incoming sunlight is about 340 W/m2: the solar constant of 1,361 divided by four for spherical geometry.
• About 100 W/m2 is reflected and about 240 W/m2 absorbed and re-radiated.
• Albedo is the fraction of sunlight reflected; Earth's average is about 0.30, and absorbed fraction equals 1 minus albedo.
• Balancing 240 W/m2 with the Stefan-Boltzmann law gives minus 18 degrees C, so the greenhouse effect supplies the missing 33 degrees C.
• Earth now absorbs more than it emits, an imbalance near 0.79 W/m2 for 2006-2018 and growing, with over 90 percent of the surplus stored in the ocean.

Sources

1. NASA Earth Observatory. (n.d.). Climate and Earth's energy budget. NASA Science. earthobservatory.nasa.gov
2. Forster, P., Storelvmo, T., Armour, K., et al. (2021). Chapter 7: The Earth's energy budget, climate feedbacks, and climate sensitivity. In IPCC AR6 WGI. ipcc.ch
3. NASA Langley Research Center. (n.d.). CERES: Clouds and the Earth's Radiant Energy System. ceres.larc.nasa.gov
4. NOAA Climate.gov ↗. (n.d.). Climate change: Ocean heat content. climate.gov
5. NASA Earth Observatory. (n.d.). Aerosols: Tiny particles, big impact. NASA Science. earthobservatory.nasa.gov
6. IPCC. (2021). Summary for policymakers. In Climate change 2021: The physical science basis (AR6 WGI). ipcc.ch
7. NASA Science. (n.d.). Earth indicators: Ocean warming. climate.nasa.gov

Commercial airliners cruise near 11 kilometers, and at that height roughly three quarters of the atmosphere's entire mass is already beneath the aircraft. Passengers looking down at the weather from above are, in a real sense, looking down on nearly the whole climate system. The gases that keep this planet habitable occupy a shell so shallow that a routine flight rises above most of it.

The big picture

The atmosphere is a thin shell of gas held to Earth by gravity, so thin that if the planet were the size of an apple, the air that matters would be no thicker than the apple's skin. Yet that skin keeps the surface warm and breathable.

Two facts drive this lesson. First, the air is mostly nitrogen and oxygen, but the tiny leftover fraction of greenhouse gases does the climate work. Second, the atmosphere is layered by temperature, and the bottom layer, where we live, is where nearly all weather and climate happen.

What the air is made of

By volume, dry air is about 78 percent nitrogen and 21 percent oxygen. Argon fills roughly another 0.9 percent, and carbon dioxide is only about 0.04 percent, expressed more conveniently as parts per million (ppm): 425 ppm means 425 CO₂ molecules for every million molecules of air.

Those trace amounts matter enormously. The small fraction of greenhouse gases, carbon dioxide, methane, nitrous oxide, and above all water vapor, is what keeps the surface from freezing. It is like a pinch of yeast in a loaf: a tiny ingredient by weight that changes the whole result. Water vapor is the wild card, highly variable from near zero in a cold desert to several percent in the humid tropics.

Key idea: Air is mostly nitrogen and oxygen, but the trace greenhouse gases, tiny in amount, control the surface temperature.

The trace gases, in numbers

Precision matters here, because the whole climate argument rests on how much these concentrations have changed. NOAA's Global Monitoring Laboratory measures them continuously from a global network of sampling sites, and reports the following globally averaged annual means for 2025.

Gas	2025 global mean	Pre-industrial (about 1750)	Increase
Carbon dioxide (CO2)	425.6 ppm	about 278 ppm	about 53 percent
Methane (CH4)	1,936 ppb	about 722 ppb	about 168 percent
Nitrous oxide (N2O)	339 ppb	about 270 ppb	about 25 percent

Two cautions about reading such a table. First, these values climb every year, so always check the observation date and consult the current NOAA record rather than a remembered number. The Mauna Loa station, which sees a strong seasonal cycle, recorded a monthly mean of about 432 ppm at its May 2026 peak, higher than the global annual figure. Second, ppb means parts per billion; methane is a thousand times scarcer than CO₂ by count, yet molecule for molecule it traps far more heat over a 20-year window, so scarcity does not imply insignificance.

Key idea: Measured globally, CO₂ reached about 425.6 ppm in 2025 against a pre-industrial 278 ppm, with methane up about 168 percent and nitrous oxide about 25 percent.

Well-mixed gases versus variable ones

Atmospheric constituents fall into two behaviourally distinct groups, and the distinction determines how each one is measured and modelled. A well-mixed gas persists long enough to be distributed almost uniformly around the planet before it is removed. Carbon dioxide, methane, and nitrous oxide all qualify, which is why a measurement taken in the middle of the Pacific characterises the entire globe to within a few parts per million.

Water vapor is the opposite case. Its atmospheric residence time is roughly nine days, far too short for global mixing, so its concentration is set locally by temperature: warm air can hold much more vapor than cold air. Consequently humidity is not an independent variable that humans can dial up or down. It adjusts to whatever temperature the long-lived gases establish, which is precisely why water vapor functions as a feedback rather than a forcing.

Key idea: Long-lived gases like CO₂ mix uniformly worldwide and act as forcings, whereas water vapor lasts about nine days and simply tracks temperature, making it a feedback.

Layers of the atmosphere

The atmosphere is stacked into layers defined by how temperature changes with height, like floors in a building where the temperature behaves differently on each floor.

• The troposphere reaches from the surface up to roughly 10 to 15 km. It holds about three quarters of the atmosphere's mass and nearly all its water. Temperature falls with altitude here, which is why mountaintops are cold, and this is the floor where weather happens.
• The stratosphere extends to about 50 km and contains the ozone layer, which soaks up harmful ultraviolet light. Unusually, this layer warms with height because the ozone absorbs sunlight up top.
• The mesosphere reaches to about 85 km; it is the coldest layer and where most meteors burn up.
• The thermosphere extends hundreds of kilometers higher; the air is so thin it barely counts as air, yet it is very hot, and auroras glow here.

Layer	Approx. top	Key feature
Troposphere	10 to 15 km	Weather; most mass and water
Stratosphere	about 50 km	Ozone layer; warms with height
Mesosphere	about 85 km	Meteors burn up; coldest layer
Thermosphere	about 600 km	Auroras; extremely thin air

Key idea: The atmosphere is layered by its temperature profile, and the troposphere at the bottom holds most of the mass, the water, and the weather.

Pressure, density, and why the air runs out

The atmosphere has no sharp upper edge; it simply thins until it is indistinguishable from space. Because each layer of air must support the weight of everything above it, pressure decreases roughly exponentially with altitude, halving about every 5.5 kilometers in the lower atmosphere.

Work that halving rule through. At sea level, pressure is about 1,013 hectopascals. At 5.5 km, roughly the summit height of Kilimanjaro, it has fallen to about half, near 500 hectopascals, which is why climbers there take in half the oxygen per breath. At 11 km, cruising altitude, it has halved twice to about 250 hectopascals, meaning three quarters of the atmosphere's mass lies below. By 32 km, six halvings have occurred and roughly 99 percent of the mass is beneath you.

This exponential thinning explains a fact that sounds paradoxical: the thermosphere is extremely hot, with individual molecules moving at temperatures of hundreds or thousands of degrees, yet a thermometer placed there would read cold. Temperature measures the energy per molecule, while heat transfer depends on how many molecules there are to collide with. At those altitudes there are almost none.

Key idea: Atmospheric pressure halves roughly every 5.5 km, so three quarters of the air lies below aircraft cruising altitude and 99 percent below about 32 km.

Why each layer has the temperature profile it does

The layers are defined by whether temperature rises or falls with height, and each pattern has a physical cause. In the troposphere, the air is heated principally from below, because sunlight passes through the atmosphere and warms the ground, which then heats the air in contact with it. Temperature therefore declines with altitude at an average rate near 6.5 degrees Celsius per kilometer, called the environmental lapse rate.

Apply it. If the surface is 15 degrees Celsius and an aircraft climbs to 10 kilometers, the expected outside temperature is 15 minus 6.5 times 10, which equals minus 50 degrees Celsius, close to what airline displays actually report. Because warm air near the ground is buoyant relative to the colder air above it, the troposphere overturns readily, and that vertical churning is precisely what generates clouds and storms. The name comes from the Greek for turning.

Above the tropopause, the profile inverts. In the stratosphere, ozone absorbs ultraviolet sunlight and deposits that energy at altitude, so temperature increases with height. Warm air sitting above cool air is stable and resists overturning, which is why the stratosphere is strongly stratified, why jets prefer to fly just below or within its base, and why volcanic particles injected into it can linger for a year or two instead of raining out in days.

Key idea: The troposphere cools with height at about 6.5 degrees C per kilometer because it is heated from below and overturns, while the stratosphere warms with height because ozone absorbs ultraviolet light, making it stable.

Why the troposphere matters most

For climate, the troposphere and lower stratosphere are the main stage. The greenhouse gases and clouds packed into the troposphere govern how much heat escapes to space, so this is where the greenhouse effect largely plays out. Changes in the stratospheric ozone layer, meanwhile, link atmospheric chemistry to both ultraviolet exposure and climate, which is why later lessons treat the ozone hole and greenhouse warming as related but distinct problems.

Key idea: Because most air, water, and greenhouse gas sit in the troposphere, that lowest layer controls weather and most of the climate.

Common misconceptions

• "Because CO₂ is only 0.04 percent of the air, it cannot matter." Amount is not the same as effect. A trace gas that strongly absorbs infrared can control temperature even at a few hundred ppm.
• "The ozone layer and the greenhouse effect are the same thing." No. Ozone in the stratosphere blocks ultraviolet light; the greenhouse effect involves different gases trapping infrared heat in the troposphere.
• "It gets hotter the higher you go, closer to the Sun." No. In the troposphere it gets colder with height; the closeness to the Sun is negligible compared with how the layers are heated.
• "Water vapor is the main greenhouse gas, so CO₂ is irrelevant." No. Water vapor is abundant but its concentration is controlled by temperature, so it acts as an amplifier of the warming that CO₂ and other long-lived gases initiate.
• "Methane is negligible because it is measured in parts per billion." No. Methane is scarce by count but far more effective per molecule, contributing a substantial share of present-day radiative forcing.

Recap

• Dry air is about 78 percent nitrogen and 21 percent oxygen, with CO₂ near 0.04 percent.
• NOAA measured global mean CO₂ at about 425.6 ppm in 2025, against roughly 278 ppm before industrialization.
• Trace greenhouse gases, including highly variable water vapor, keep the surface warm.
• Pressure halves about every 5.5 km, so roughly 99 percent of the atmosphere lies below 32 km.
• The layers, from the ground up, are troposphere, stratosphere, mesosphere, thermosphere.
• The troposphere cools with height at about 6.5 degrees C per kilometer and holds most mass, water, and weather.
• The stratosphere warms with height because its ozone absorbs ultraviolet light, which makes it stable.

Sources

1. NOAA. (n.d.). Layers of the atmosphere. JetStream. noaa.gov
2. UCAR Center for Science Education. (n.d.). Layers of the atmosphere. scied.ucar.edu
3. NOAA Global Monitoring Laboratory. (2026). Trends in atmospheric carbon dioxide (global annual mean 425.64 ppm for 2025; Mauna Loa monthly means through mid-2026). gml.noaa.gov
4. NOAA Global Monitoring Laboratory. (2026). Trends in atmospheric methane (global annual mean 1,935.7 ppb for 2025). gml.noaa.gov
5. NOAA Global Monitoring Laboratory. (2026). Trends in atmospheric nitrous oxide (global annual mean 338.9 ppb for 2025). gml.noaa.gov
6. U.S. Environmental Protection Agency. (n.d.). Overview of greenhouse gases. epa.gov
7. Gulev, S. K., Thorne, P. W., Ahn, J., et al. (2021). Chapter 2: Changing state of the climate system. In IPCC AR6 WGI (pre-industrial greenhouse gas concentrations). ipcc.ch

In 1856 an American scientist named Eunice Newton Foote filled glass cylinders with different gases, set them in sunlight, and recorded which grew warmest. The cylinder of carbon dioxide outperformed all the others. Three years later, working with far more sensitive apparatus in London, John Tyndall demonstrated that carbon dioxide and water vapor absorb infrared radiation while nitrogen and oxygen do not. By 1896 Svante Arrhenius had calculated how much the planet would warm if carbon dioxide doubled. The basic physics in this lesson is over 160 years old and has survived every attempt to overturn it.

The big picture

The greenhouse effect is the reason Earth's surface averages a comfortable 15 degrees Celsius instead of a frozen minus 18 degrees Celsius. Certain gases act like a blanket that lets sunlight in but slows heat from escaping, so the surface stays warmer than bare sunlight alone would make it.

This lesson explains the mechanism step by step, why only a few trace gases do the job, and the crucial difference between the natural greenhouse effect that made life possible and the enhanced effect humans are now adding.

The mechanism, step by step

Follow the energy in four steps.

1. Sunlight arrives mostly as visible and near-visible light, called shortwave radiation. The atmosphere is largely transparent to it, so it passes through and warms the surface.
2. The warmed surface radiates energy back upward. Because Earth is far cooler than the Sun, it radiates in the longwave (infrared) band, the invisible heat you feel from a warm road at night.
3. Greenhouse gas molecules absorb much of that outgoing infrared and re-emit it in all directions, including back down toward the surface.
4. The downward share slows the escape of heat to space, so the lower atmosphere and surface settle at a warmer temperature.

Picture a blanket on a cold night. The blanket does not create heat; it just slows the heat leaving your body, so you end up warmer. Greenhouse gases do the same for the planet.

Key idea: Greenhouse gases let shortwave sunlight in but absorb outgoing longwave heat and radiate some back down, warming the surface like a blanket.

The trick is in the wavelengths

Why do carbon dioxide, water vapor, and methane trap heat while nitrogen and oxygen do not, even though N₂ and O₂ are over 99 percent of the air? The answer is molecular. Greenhouse gas molecules are built so they can vibrate and bend in ways that absorb infrared light, at specific wavelengths called their absorption bands. Nitrogen and oxygen are simple, symmetric two-atom molecules that cannot flex in those infrared-absorbing ways, so they let the heat pass. It is the trace gases that count, not the bulk of the air.

The label greenhouse is imperfect, because a real garden greenhouse mostly works by blocking warm air from rising away, not by trapping infrared. But the essential idea holds: the atmosphere lets light in and impedes heat from leaving.

Key idea: Only molecules that can vibrate to absorb infrared act as greenhouse gases, which is why trace CO₂ matters and abundant nitrogen and oxygen do not.

The molecular detail

The rule can be stated precisely. A molecule absorbs infrared radiation only if the vibration involved changes its electric dipole moment, meaning the vibration must shift the balance of positive and negative charge within the molecule. Nitrogen and oxygen consist of two identical atoms, so any stretching motion remains perfectly symmetric and no dipole appears. They are effectively transparent to infrared.

Carbon dioxide is linear and symmetric at rest, yet it possesses vibrations that break that symmetry. Its bending mode, in which the molecule flexes out of a straight line, produces a strong absorption band centred near 15 micrometers. That wavelength sits squarely within the range at which Earth radiates its heat, which is the specific coincidence that makes carbon dioxide so consequential. Water vapor, a bent molecule with a permanent dipole, absorbs across a broad set of infrared bands, while methane and nitrous oxide each absorb in regions where the atmosphere would otherwise be relatively transparent.

Key idea: A gas absorbs infrared only if a vibration changes its charge distribution, and CO₂'s bending mode near 15 micrometers falls right where Earth emits most strongly.

The better mechanism: raising the emission height

The blanket analogy is serviceable, but the rigorous explanation is more interesting and answers several common objections at once. Space sees only the infrared radiation that escapes from the topmost layers where the atmosphere becomes transparent. Because temperature falls with altitude in the troposphere, the height at which radiation escapes determines how cold the emitting layer is, and therefore how much energy leaves.

Adding carbon dioxide makes the atmosphere slightly more opaque at those wavelengths, so photons escape from a higher and consequently colder level. A colder emitter radiates less. Outgoing energy therefore drops below incoming energy, an imbalance is created, and the entire troposphere must warm until the new emission level reaches the temperature required to restore 240 watts per square meter leaving the planet.

This framing explains why the stratosphere cools while the surface warms, a pattern discussed in Lesson 11 as a fingerprint of greenhouse warming rather than solar change. It also explains why adding greenhouse gases works even though the lower atmosphere already absorbs nearly all the infrared in CO₂'s strongest bands: what matters is not absorption near the ground but the temperature at the altitude from which radiation finally escapes.

Key idea: Adding CO₂ pushes the level at which heat escapes to space higher and colder, reducing outgoing radiation until the surface and troposphere warm enough to compensate.

Radiative forcing, quantified

Radiative forcing is the standard way of measuring how strongly something pushes the energy budget, expressed in watts per square meter. The IPCC's Sixth Assessment Report assesses the effective radiative forcing from doubling carbon dioxide at 3.93 watts per square meter, with a very likely range of 3.12 to 4.74.

Crucially, the relationship between concentration and forcing is logarithmic, not linear: each doubling adds roughly the same forcing. Going from 278 to 556 ppm delivers about 3.93 watts per square meter, and going from 556 to 1,112 ppm delivers roughly another 3.93. This is why emissions must fall to zero rather than merely level off, since holding a constant high concentration holds a constant forcing.

Work an example with real numbers. Carbon dioxide rose from about 278 ppm before industrialisation to 425.6 ppm in 2025. The ratio is 425.6 divided by 278, which is 1.531. Taking natural logarithms, the number of doublings is ln(1.531) divided by ln(2), or 0.426 divided by 0.693, which is 0.61 doublings. Multiplying 0.61 by 3.93 gives about 2.4 watts per square meter of forcing from CO₂ alone. For comparison, AR6 assessed CO₂ forcing at 2.16 watts per square meter for 1750 to 2019, when concentrations stood near 410 ppm, and total human-caused forcing from all agents at 2.72 watts per square meter with a range of 1.96 to 3.48.

Key idea: Doubling CO₂ forces the climate by about 3.93 W/m2, the effect is logarithmic in concentration, and the rise to 425.6 ppm corresponds to roughly 2.4 W/m2 from CO₂ alone.

Natural versus enhanced greenhouse effect

The natural greenhouse effect has kept Earth habitable for billions of years; without it, liquid water and life as we know it would be impossible. What is new is the enhanced greenhouse effect. By burning fossil fuels and clearing land, humans have raised atmospheric CO₂ from about 278 ppm before industrialization to a global annual mean of 425.6 ppm in 2025, along with large increases in methane and nitrous oxide.

More greenhouse gas means a thicker blanket: more outgoing infrared absorbed, a larger energy imbalance, and a warmer surface. The physics linking CO₂ to warming was worked out in the 1800s and confirmed countless times since. It is not in scientific doubt.

It is worth separating two questions that often get conflated. Whether added greenhouse gases warm the planet is settled physics, verified in laboratories, by satellites measuring the infrared spectrum leaving Earth, and by the historical record. How much warming results from a given amount of gas is a quantitative question with a genuine uncertainty range, because it depends on feedbacks. Uncertainty about magnitude is not uncertainty about direction, and Lesson 12 examines exactly where the remaining uncertainty lies.

A small worked example

The numbers show how much the blanket does. With no greenhouse gases, Earth's average surface temperature would be about minus 18 degrees Celsius. With the natural greenhouse effect, it is about plus 15 degrees Celsius. That is a warming of 15 minus (minus 18), which equals 33 degrees Celsius of extra warmth, entirely from trace gases. Human activity is now adding to that baseline, which is why even a degree or two of further warming is significant.

Common misconceptions

• "The greenhouse effect is bad." The natural greenhouse effect is essential; without it Earth would be frozen. The concern is the extra, human-enhanced warming on top.
• "Greenhouse gases block incoming sunlight." No. They are largely transparent to incoming shortwave sunlight; they absorb outgoing longwave heat.
• "Since nitrogen is 78 percent of the air, it must be the main greenhouse gas." No. Nitrogen cannot absorb infrared effectively; the trace gases do the trapping.
• "CO₂'s absorption bands are already saturated, so more cannot matter." No. Warming works by raising the altitude from which radiation escapes to a colder level, and that mechanism continues to operate however opaque the lower atmosphere already is.
• "Doubling CO₂ again would double the warming." No. Forcing is logarithmic, so each doubling adds roughly the same 3.93 W/m2 rather than twice as much.

Recap

• Sunlight comes in as shortwave; Earth radiates heat back as longwave infrared.
• Greenhouse gases absorb that infrared and re-emit some downward, warming the surface.
• Only vibrations that change a molecule's charge distribution absorb infrared, so trace gases matter and N₂/O₂ do not.
• Rigorously, added CO₂ raises the emission height to a colder level, cutting outgoing radiation until the surface warms.
• Doubling CO₂ gives about 3.93 W/m2 of forcing, and the effect is logarithmic in concentration.
• The natural greenhouse effect adds about 33 degrees Celsius and makes Earth habitable.
• Humans have enhanced it by raising CO₂ from about 278 ppm to 425.6 ppm in 2025.

Sources

1. NASA Science. (n.d.). What is the greenhouse effect? Climate Change FAQ. science.nasa.gov
2. Forster, P., Storelvmo, T., Armour, K., et al. (2021). Chapter 7: The Earth's energy budget, climate feedbacks, and climate sensitivity. In IPCC AR6 WGI (ERF for doubled CO₂ = 3.93 W/m2). ipcc.ch
3. IPCC. (2021). Summary for policymakers. In Climate change 2021: The physical science basis (AR6 WGI; total anthropogenic forcing 2.72 W/m2 for 1750-2019). ipcc.ch
4. NASA Science. (n.d.). Causes of climate change. science.nasa.gov
5. NASA Science. (n.d.). Steamy relationships: How atmospheric water vapor amplifies Earth's greenhouse effect. science.nasa.gov
6. NOAA Global Monitoring Laboratory. (2026). The NOAA Annual Greenhouse Gas Index (AGGI). gml.noaa.gov
7. NOAA Climate.gov ↗. (n.d.). Climate change: Annual greenhouse gas index. climate.gov

Every carbon atom in your body has been somewhere remarkable. It may have spent a hundred million years locked in limestone, a decade dissolved in the North Atlantic, a summer inside a maize leaf in Iowa, and a fraction of a second passing through a diesel engine. Carbon does not disappear; it relocates. The whole of climate science depends on knowing where the carbon is, how fast it moves, and what happens when people move it faster than nature can move it back.

The big picture

Carbon is the backbone of life and a master control on climate. The carbon cycle is the constant movement of carbon among places where it is stored, the air, the oceans, living things and soils, and the rocks and fossil fuels underground.

The key to the whole climate problem lives here. There is a fast cycle that recycles carbon in years to centuries and a slow cycle that recycles it over millions of years. Humans have taken carbon from the slow cycle and dumped it into the fast one, faster than nature can put it back.

The reservoirs

Carbon sits in reservoirs (also called pools), think of them as tanks of different sizes, and moves between them by fluxes (flows). The main reservoirs are:

• The atmosphere, holding carbon as CO₂ and methane. This tank is small but is the one we change most easily.
• The oceans, by far the largest fast-exchanging pool, holding vast amounts of dissolved carbon.
• The land biosphere and soils, the carbon locked in living matter and decaying organics.
• The rocks and fossil fuels underground, an enormous, very slowly cycling store.

Key idea: Carbon is stored in reservoirs (air, ocean, land, rock) and moves between them as fluxes; the ocean is the biggest fast pool.

How big is each tank?

Carbon is conventionally counted in gigatonnes of carbon, abbreviated GtC, where one gigatonne is a billion tonnes. Because policy documents often use gigatonnes of carbon dioxide instead, keep the conversion handy: one GtC equals 3.664 GtCO₂. A second useful constant is that adding 2.124 GtC to the atmosphere raises its concentration by one part per million.

Reservoir	Approximate size (GtC)	Turnover
Atmosphere	about 900	years to millennia
Vegetation	about 450	years to decades
Soils (including permafrost)	about 1,700 or more	decades to millennia
Ocean (dissolved carbon)	about 37,000	decades to millennia
Fossil fuel reserves	roughly 1,000 or more	millions of years
Carbonate rocks and sediments	tens of millions	tens of millions of years

Check the arithmetic on the atmosphere. At the 2025 global mean of 425.6 ppm, the atmospheric stock is 425.6 multiplied by 2.124, which is about 904 GtC. Before industrialisation, at 278 ppm, it held about 590 GtC. Humans have therefore added roughly 314 GtC to the air, an increase of over half, while the ocean holds forty times as much and the rocks hold more than all the other reservoirs combined by a wide margin.

Key idea: The atmosphere holds only about 900 GtC against roughly 37,000 GtC in the ocean, so a flux that barely dents the ocean can transform the air.

Fast cycle: life breathing

On timescales of days to centuries, carbon moves quickly, and the main pump is life itself. Photosynthesis pulls CO₂ out of the air and builds it into plants; respiration and decomposition burn that carbon back to CO₂ and return it. The ocean absorbs CO₂ at its cold surface and releases it where waters are warm. You can think of the planet as breathing: in through photosynthesis, out through respiration.

Before industrialization, these enormous flows were roughly balanced. What the land and ocean took up each year, they gave back, which held atmospheric CO₂ near 280 ppm for thousands of years.

Key idea: The fast carbon cycle is life breathing, photosynthesis in and respiration out, and it was nearly balanced before humans intervened.

Slow cycle: rock and fossils

On timescales of millions of years, a slow cycle grinds along. Weathering of rock by rainwater removes CO₂ from the air, and volcanoes belch it back. Over geological time, a small share of organic carbon got buried and compressed into fossil fuels, coal, oil, and gas, and into carbonate rocks, locking carbon away for eons. Normally this slow cycle just trickles.

The slow cycle also contains the planet's own thermostat, operating over hundreds of thousands of years. Rainwater absorbs CO₂ and becomes weakly acidic; that acid dissolves silicate rock; the dissolved products wash to the sea and are eventually deposited as carbonate sediment. Because the chemical reactions run faster when the climate is warmer and wetter, a warmer world removes CO₂ faster, which cools things back down. This silicate weathering feedback is a stabilising loop of exactly the kind Lesson 1 described. It is also far too slow to help us, operating over hundreds of thousands of years rather than the decades over which humans are changing the atmosphere.

Key idea: The slow carbon cycle exchanges carbon between rocks and air over millions of years, buried the carbon now stored as fossil fuels, and includes a silicate weathering thermostat far too slow to offset human emissions.

The human perturbation

Here is the crux. Burning fossil fuels reaches into the slow reservoir and dumps its carbon into the fast one, transferring carbon that took millions of years to accumulate into the atmosphere in mere decades. That is far faster than the land and ocean can soak it up. Each year, ocean and land sinks absorb roughly half of what we emit; the rest accumulates in the air, which is why CO₂ keeps climbing.

A small worked example makes the buildup concrete. If humans emit 10 units of carbon in a year and natural sinks absorb about 5 (roughly half), then 10 minus 5, or 5 units, stay in the atmosphere that year. Repeat that year after year and the atmospheric stock keeps rising, which is exactly the trend measured since the Industrial Revolution. And because a substantial fraction of emitted CO₂ lingers for centuries, the carbon cycle explains both why CO₂ is rising and why it will stay high long after emissions fall.

Key idea: Fossil-fuel burning moves slow-cycle carbon into the fast cycle faster than sinks absorb it, so about half stays airborne and CO₂ accumulates.

The global carbon budget, with real numbers

The Global Carbon Project publishes an annual audit of every term in this ledger. Averaged over the decade 2014 to 2023, the accounting ran as follows, in gigatonnes of carbon per year.

Term	GtC per year	Share of emissions
Fossil fuel and cement emissions	about 9.6	-
Land-use change emissions	about 1.2	-
Growth in the atmosphere	5.2	48 percent
Ocean sink	2.9	26 percent
Land sink	3.2	30 percent

For the single year 2023, total emissions reached 11.1 GtC (40.6 GtCO₂), atmospheric growth was 5.9 GtC, the ocean took up 2.9 GtC, and the land sink fell to only 2.3 GtC, its weakest since 2015, largely because an El Nino stressed tropical vegetation and Canadian wildfires burned intensely. That single year illustrates a critical vulnerability: the sinks are not guaranteed. They are living and chemical systems that respond to climate, and a warmer, drier land surface absorbs less.

Convert the atmospheric growth into concentration to see the connection with Lesson 4. Dividing 5.9 GtC by 2.124 GtC per ppm gives about 2.8 ppm, and the measured increase in 2023 was indeed 2.79 ppm. The observed rise in the Keeling curve, the emissions inventory, and the ocean chemistry all agree, which is one reason confidence in this budget is high.

Key idea: Over 2014-2023 humans emitted roughly 10.8 GtC a year; the atmosphere kept 48 percent while the ocean took 26 percent and the land 30 percent, and the land sink weakens in hot, dry years.

The long tail of a CO₂ molecule

Ask how long a pulse of carbon dioxide stays in the air and you will not get a single number, because the removal happens in stages. Roughly half of an emitted pulse is drawn down within a few decades, mostly by the surface ocean and growing vegetation. The next portion takes centuries as the deep ocean mixes and slowly buffers the added acid. A residual fraction, on the order of one fifth to one quarter, persists for many thousands of years until silicate weathering finally removes it.

This staged behaviour is why comparing CO₂ to ordinary pollutants misleads. Soot cleared from a city improves the air within days. Carbon dioxide emitted today will still be influencing the climate when the children born this century have great-grandchildren. It is also why the arithmetic of Lesson 1 applies so directly: a stock with a very slow drain requires the inflow to reach zero, not merely to shrink, before it stops growing.

Key idea: A CO₂ pulse is removed in stages, with roughly half gone in decades but a fifth or more lingering for millennia, so emissions today commit the climate for a very long time.

Common misconceptions

• "Plants and the ocean will just absorb whatever we emit." They absorb only about half; the remainder builds up in the atmosphere.
• "CO₂ we emit is gone within a few years." A large fraction persists for centuries, which is why stabilizing the climate requires net-zero emissions.
• "Volcanoes emit more CO₂ than humans." No. Human fossil-fuel emissions are roughly a hundred times larger than volcanic emissions each year.
• "The natural fluxes are far bigger than ours, so we cannot matter." Natural exchange is indeed larger, but it runs both ways and nearly cancels. Human emissions are a one-way addition, and it is the net imbalance that accumulates.
• "The sinks will keep taking half of what we emit." Not guaranteed. The land sink collapsed by more than a third in 2023 under El Nino heat and wildfire, and warming tends to weaken both sinks.

Recap

• The carbon cycle moves carbon among air, ocean, land and life, and rocks.
• The atmosphere holds about 900 GtC, the ocean about 37,000 GtC, and 2.124 GtC equals 1 ppm.
• The fast cycle (photosynthesis and respiration) was balanced before industrialization.
• The slow cycle buried carbon as fossil fuels and contains a silicate weathering thermostat far too slow to help.
• Over 2014-2023 the atmosphere kept 48 percent of emissions, the ocean took 26 percent, and the land 30 percent.
• Burning fossil fuels jumps slow-cycle carbon into the fast cycle far faster than sinks can absorb it.
• A CO₂ pulse clears in stages, with a substantial fraction persisting for millennia.

Sources

1. NASA Earth Observatory. (n.d.). The carbon cycle. NASA Science. earthobservatory.nasa.gov
2. Friedlingstein, P., O'Sullivan, M., Jones, M. W., et al. (2025). Global Carbon Budget 2024. Earth System Science Data, 17, 965-1039. essd.copernicus.org
3. Global Carbon Project. (n.d.). Global Carbon Budget. globalcarbonbudget.org
4. Canadell, J. G., Monteiro, P. M. S., Costa, M. H., et al. (2021). Chapter 5: Global carbon and other biogeochemical cycles and feedbacks. In IPCC AR6 WGI. ipcc.ch
5. NOAA Global Monitoring Laboratory. (2026). Trends in atmospheric carbon dioxide. gml.noaa.gov
6. NASA Earth Observatory. (n.d.). The ocean's carbon balance. NASA Science. earthobservatory.nasa.gov
7. Our World in Data. (n.d.). CO₂ emissions. ourworldindata.org

Roughly half of the nitrogen atoms in your body passed through a steel reactor before they reached your plate. The Haber-Bosch process, industrialised just before the First World War, takes inert nitrogen gas out of the air and forces it to combine with hydrogen under enormous pressure to make ammonia. It is the single reason global population could grow past about four billion. It is also the reason rivers from the Mississippi to the Yangtze carry more nitrogen than any ecosystem downstream evolved to handle.

The big picture

Carbon is not the only element that cycles. Every essential nutrient moves through the air, water, rock, and living things in what is called a biogeochemical cycle, bio for life, geo for earth, chemical for the substances involved.

Two of these cycles, water and nitrogen, matter most for this course. The water cycle moves heat and moisture around the planet and gets more energetic as the world warms. The nitrogen cycle was gently balanced for millions of years until humans roughly doubled it to grow food, with pollution as a side effect.

The water cycle

The water cycle moves water among ocean, atmosphere, land, and ice through evaporation (liquid to vapor), condensation into clouds, precipitation (rain and snow), runoff, and infiltration into the ground. But it does far more than deliver rain.

Evaporating water absorbs heat, and that heat is released again when the vapor condenses. So the water cycle acts like a planetary conveyor for energy, carrying heat from the hot tropics toward the cooler poles and powering storms along the way. Think of sweat cooling your skin as it evaporates, then imagine that heat reappearing wherever the moisture later condenses.

Warming turns up the intensity. A warmer atmosphere holds about 7 percent more water vapor per degree Celsius. More available moisture means heavier downpours where it does rain, while the same extra warmth dries soils faster and can deepen droughts between rains. The water cycle does not just shift; it speeds up.

Key idea: The water cycle transports huge amounts of energy, not just water, and a warmer atmosphere intensifies both heavy rain and drought.

Latent heat, and why evaporation is an energy transfer

The energy claim deserves a number. Converting one kilogram of liquid water into vapor at ordinary temperatures requires about 2.45 million joules, an amount called the latent heat of vaporisation. That energy is not destroyed; it is stored in the vapor and released again wherever condensation occurs, which may be a thousand kilometers away and eight kilometers up.

Globally, this latent heat flux carries roughly 80 watts per square meter from the surface upward, a third of the 240 watts per square meter that Lesson 3 showed Earth absorbs. Evaporation is therefore not a minor side process; it is one of the principal ways the surface sheds energy. It also explains why hurricanes exist at all. A tropical cyclone is essentially a machine that converts the latent heat stored in warm ocean water into the kinetic energy of wind, which is why the strongest storms form over the warmest seas.

Key idea: Evaporating a kilogram of water stores about 2.45 million joules that are released on condensation, so the water cycle moves roughly 80 W/m2 of energy and powers storms.

The Clausius-Clapeyron relation, worked through

The 7 percent figure quoted above is not an empirical rule of thumb but a consequence of thermodynamics known as the Clausius-Clapeyron relation, which describes how the maximum water vapor content of air rises with temperature. Near present-day surface temperatures, the saturation vapor pressure increases by about 7 percent for each degree Celsius of warming.

Compound it and the effect grows. If the world warms by 3 degrees Celsius, the moisture-holding capacity rises by 1.07 multiplied by itself three times, which equals about 1.23, or a 23 percent increase. The IPCC assesses that extreme daily rainfall intensifies at roughly this rate, about 7 percent per degree of global warming, even though total annual precipitation rises much more slowly, at closer to 1 to 3 percent per degree.

That mismatch is the source of a genuinely counterintuitive result. Because total rainfall grows more slowly than the intensity of individual downpours, the extra water must be delivered in fewer, heavier events with longer dry intervals between them. A warmer world is therefore expected to be both wetter when it rains and drier between rains, which is exactly the pattern of flood-and-drought whiplash discussed in Lesson 14.

Key idea: Clausius-Clapeyron gives about 7 percent more moisture per degree C, and because total rainfall rises more slowly, the extra water arrives as heavier downpours separated by longer dry spells.

The nitrogen cycle

Nitrogen gas is 78 percent of the air, yet most life cannot use it directly, because the N₂ molecule is locked tight, like a nutrient sealed in a jar no organism can open. Nitrogen fixation is the process that opens the jar: specialized bacteria, and lightning, convert inert N₂ into usable forms such as ammonia and nitrate. Other microbes later return nitrogen to the air through denitrification, closing the loop.

For most of history, the scarcity of usable nitrogen limited how much plants could grow. Then, about a century ago, humans invented the industrial fixation of nitrogen to make fertilizer, and roughly doubled the amount of reactive nitrogen entering ecosystems. That fed billions of people, one of the most consequential inventions in history, but the excess does not vanish.

Fertilizer that washes off fields triggers eutrophication: the extra nutrients feed explosive algae blooms, the algae die and decompose, and the decomposers strip oxygen from the water. The result is a dead zone, an area so oxygen-poor that fish and shellfish cannot survive.

Key idea: Nitrogen fixation makes inert N₂ usable; humans doubled the flow to grow food, and the runoff causes eutrophication and dead zones.

How a dead zone forms, step by step

Eutrophication follows a reliable sequence, and understanding it makes the problem far easier to reason about.

1. Fertiliser, manure, or sewage delivers nitrogen and phosphorus into a river, which carries them to a lake or coastal sea.
2. Algae and cyanobacteria, previously limited by nutrient scarcity, multiply explosively into a bloom that can be visible from orbit.
3. The bloom shades out submerged plants below it, and within days to weeks the algae die.
4. Bacteria decompose the enormous mass of dead algae, and that decomposition consumes dissolved oxygen.
5. Where the water column is stratified, so that warm fresh surface water floats over cold salty deep water, no oxygen can mix down from above. Bottom oxygen falls below about 2 milligrams per liter, the threshold called hypoxia, and mobile animals flee while immobile ones suffocate.

The northern Gulf of Mexico hosts the best-documented example, fed by drainage from the Mississippi basin. Its hypoxic zone typically covers thousands of square miles each summer, and reducing it has proved difficult because the nutrients come from millions of dispersed farm fields rather than from any single pipe that could be closed.

Notice that warming makes this worse through two independent routes. Warmer water holds less dissolved oxygen to begin with, and stronger surface warming intensifies the stratification that prevents oxygen from mixing downward. Nutrient pollution and climate change are separate problems that happen to reinforce each other.

Key idea: Dead zones form when nutrient-fed algal blooms die and their decomposition strips oxygen from stratified bottom water, and warming worsens both the oxygen supply and the stratification.

Nitrogen's climate connection

The nitrogen cycle is not merely a water quality issue. One of its products, nitrous oxide, is released by soil microbes processing surplus fertiliser, and it is a potent greenhouse gas: molecule for molecule it traps far more heat than carbon dioxide and persists in the atmosphere for over a century. NOAA measured its global mean at about 339 parts per billion in 2025, roughly 25 percent above the pre-industrial value of about 270.

Nitrous oxide is also, since the phase-out of chlorofluorocarbons discussed in Lesson 16, among the most significant remaining destroyers of stratospheric ozone. A single agricultural practice therefore touches food supply, water quality, climate forcing, and the ozone layer at once. That interconnection is the whole point of studying biogeochemical cycles rather than isolated pollutants.

Key idea: Surplus fertiliser nitrogen releases nitrous oxide, a long-lived greenhouse gas now near 339 ppb that also depletes stratospheric ozone, linking agriculture to climate and air chemistry.

The pattern repeats: phosphorus

The phosphorus cycle behaves similarly. Mined phosphate fertilizer boosts crops, but the runoff into rivers and lakes drives the same over-enrichment and oxygen loss. The general lesson is simple and important: nutrient cycles that stayed roughly balanced for millennia are now heavily loaded by human activity, and the excess does not disappear. It reappears downstream as pollution. These couplings return in the pollution module.

Key idea: Human loading of nutrient cycles like nitrogen and phosphorus does not vanish; it reappears downstream as water pollution.

Common misconceptions

• "The water cycle only affects whether it rains." No. It also moves enormous amounts of heat around the planet and powers storms.
• "Since air is 78 percent nitrogen, plants have all the nitrogen they need." No. Most life cannot use N₂ gas directly; it must first be fixed into usable forms.
• "Fertilizer runoff makes water greener and healthier." No. The algae blooms it causes decay and remove oxygen, creating dead zones that kill aquatic life.
• "A dead zone means the water is poisoned." Not usually. The water is simply short of dissolved oxygen because bacteria consumed it while decomposing dead algae.
• "More rain everywhere is the main effect of a warmer water cycle." No. Total rainfall rises slowly while individual downpours intensify at about 7 percent per degree, so wet extremes and dry spells both sharpen.

Recap

• A biogeochemical cycle moves an element through life, water, air, and rock.
• The water cycle transports heat as well as moisture, carrying roughly 80 W/m2 as latent heat.
• Clausius-Clapeyron gives about 7 percent more water vapor per degree Celsius, compounding to about 23 percent for 3 degrees.
• Total rainfall rises more slowly than downpour intensity, so wet and dry extremes both intensify.
• Nitrogen fixation converts inert N₂ into forms life can use, and Haber-Bosch roughly doubled the global flow.
• Runoff drives blooms whose decay strips oxygen, producing hypoxic dead zones below about 2 mg of oxygen per liter.
• Surplus nitrogen also releases nitrous oxide, a greenhouse gas near 339 ppb that depletes stratospheric ozone.

Sources

1. U.S. Geological Survey. (n.d.). Water cycle. Water Science School. usgs.gov
2. Douville, H., Raghavan, K., Renwick, J., et al. (2021). Chapter 8: Water cycle changes. In IPCC AR6 WGI. ipcc.ch
3. Fowler, D., Coyle, M., Skiba, U., et al. (2013). The global nitrogen cycle in the twenty-first century. Philosophical Transactions of the Royal Society B, 368(1621). pmc.ncbi.nlm.nih.gov
4. Erisman, J. W., Sutton, M. A., Galloway, J., Klimont, Z., & Winiwarter, W. (2008). How a century of ammonia synthesis changed the world. Nature Geoscience, 1, 636-639. nature.com
5. U.S. Environmental Protection Agency. (n.d.). Nutrient pollution. epa.gov
6. NOAA National Ocean Service. (n.d.). Hypoxia. oceanservice.noaa.gov
7. NOAA Global Monitoring Laboratory. (2026). Trends in atmospheric nitrous oxide. gml.noaa.gov

A casino owner cannot tell you what the next spin of the roulette wheel will produce, and does not care. What the owner knows with great confidence is the distribution of thousands of spins, and that knowledge is enough to build a business on. Meteorologists are in the first position and climate scientists in the second. The distinction is not a matter of one being harder than the other; it is a matter of asking different questions about the same physical system.

The big picture

People constantly mix up two related ideas. Weather is what the atmosphere is doing right now, over hours to days. Climate is the long-run pattern of that weather, averaged over decades. A handy way to remember it: weather is like your mood on a given day, while climate is like your personality over the years.

This distinction is the antidote to a common bad argument, that a cold snap disproves global warming. It does not, and this lesson explains why, along with the natural rhythms that ride on top of the long-term trend.

Two timescales

Weather is the state of the atmosphere at a place over hours to days: today's temperature, this afternoon's thunderstorm, next week's cold snap. Climate is the statistics of that weather over long periods, conventionally 30 years or more, capturing the averages, the extremes, and the range you can expect for a place and season. A 30-year average used as a baseline is called a climate normal. As the saying goes, climate is what you expect and weather is what you get.

Key idea: Weather is the atmosphere over hours to days; climate is its long-term statistics over about 30 years or more.

Why thirty years, and not three?

The thirty-year convention was adopted by the World Meteorological Organization and is not arbitrary. It is long enough to average out year-to-year noise and the main multi-year oscillations, yet short enough to remain relevant to the present. Statisticians describe the choice as a signal-to-noise problem.

Put numbers on it. The year-to-year scatter in global mean temperature, measured as a standard deviation around the trend, is roughly 0.1 degrees Celsius. The underlying warming trend is currently near 0.2 degrees Celsius per decade. Over three years the trend produces about 0.06 degrees of change, smaller than the noise, so nothing can be concluded. Over thirty years the trend produces about 0.6 degrees, roughly six times the noise, and the signal is unmistakable.

The same arithmetic explains why local records need even longer. At a single weather station the year-to-year scatter can be 1 degree Celsius or more, ten times the global figure, because local weather is far noisier than the planetary average. Detecting a trend at one location therefore takes decades longer than detecting the same trend globally, which is precisely why global datasets, not personal memory, are the appropriate evidence.

Key idea: Thirty years is chosen because a 0.2 degrees C per decade trend needs that long to rise clearly above roughly 0.1 degrees C of year-to-year noise, and local noise is ten times larger still.

Why one cold day proves nothing

Weather is noisy and variable, so a single cold day, or even a cold winter, tells you nothing about the long-term trend. It is like judging a school's average height from one tall student: the individual says little about the group. Climate is detected by averaging out the day-to-day noise over decades and large areas.

When scientists say the planet has warmed, they are describing a shift in that long-term global average. The IPCC assessed global surface temperature for 2011 to 2020 as about 1.09 degrees Celsius above the 1850 to 1900 baseline, with a likely range of 0.95 to 1.20. Any given day, anywhere, can still be unusually cold, just as a warming personality can still have an occasional bad mood.

Key idea: A single cold day is weather noise and says nothing about the decades-long climate trend, which is only visible after averaging.

Who measures the global average, and how

No single thermometer reports the planet's temperature. Several independent teams construct global records from land stations, ships, drifting buoys, and satellites, each applying its own corrections for changes in instruments, station moves, and gaps in coverage. The principal datasets include NASA's GISTEMP, NOAA's NOAAGlobalTemp, the UK Met Office and University of East Anglia's HadCRUT5, Berkeley Earth, and the ECMWF reanalysis known as ERA5.

Their independence is the point. Different teams, different methods, different correction schemes, and yet their year-to-year curves lie almost on top of one another. Records are also reported as anomalies, meaning departures from a reference period, rather than as absolute temperatures. Anomalies travel much further than absolute readings, because a cold snap chills a whole region at once even though the absolute temperature at a mountaintop and a nearby valley differ greatly. That property is what allows a limited network of stations to characterise the whole globe.

Key idea: Several independent teams build the global record from different data and methods, report it as anomalies rather than absolute temperatures, and arrive at closely matching results.

How to read the graph

The wiggly line is yearly weather; it jumps up and down every year. The straight red line is the climate trend, the average direction once the wiggles are smoothed out. Notice that even while the trend rises, some individual years dip below earlier ones. That is the whole point: the trend can climb steadily even though not every year is warmer than the last.

Natural variability

The climate also has natural rhythms that ride on top of the long-term trend. The best known is El Nino and La Nina, a see-saw of Pacific Ocean temperatures that nudges weather worldwide every few years, warming the globe a bit during El Nino and cooling it during La Nina. This is natural variability: it makes some years warmer or cooler than the trend alone would predict.

But these are oscillations around a rising baseline, not the cause of the century-scale warming. Separating the signal (the long-term trend) from the noise (natural variability) is a core skill in climate science, and it is why scientists insist on decades of data before drawing conclusions.

Key idea: Natural cycles like El Nino make individual years warmer or cooler, but they ride on top of the long-term trend rather than causing it.

Reading the record years correctly

Recent headline figures make an excellent exercise in distinguishing weather from climate. The World Meteorological Organization confirmed 2024 as the warmest year in the instrumental record at about 1.55 degrees Celsius above the 1850 to 1900 average, and reported 2025 as the second or third warmest at about 1.43 degrees Celsius, with 2015 to 2025 the warmest eleven years on record.

Two things follow, and both are commonly garbled. First, 2025 being cooler than 2024 is not evidence that warming stopped. It reflects the swing from El Nino conditions toward neutral or La Nina ones, exactly the natural variability described above, superimposed on a rising baseline. Second, a single year exceeding 1.5 degrees Celsius is not the same as the world having crossed the 1.5 degree threshold discussed in the Paris Agreement. That threshold refers to a multi-decade average of human-caused warming, which remains below the single-year peaks. Long-term warming is currently estimated to be running near 1.3 to 1.4 degrees Celsius.

The honest summary is therefore layered. Individual years bounce around by a couple of tenths of a degree because of ENSO and volcanic activity. The decadal average climbs steadily. The threshold in international agreements refers to the decadal average, not the bounce. Anyone quoting one year as proof of anything, in either direction, is confusing weather with climate.

Key idea: 2024 was the warmest year at about 1.55 degrees C and 2025 about 1.43 degrees C, but crossing 1.5 in one year is not the same as crossing the long-term threshold, which tracks the multi-decade average.

What attribution can and cannot claim

A related confusion concerns individual events. It is scientifically untenable to say that climate change "caused" a particular hurricane, because that storm would have had some probability of occurring in any climate. What the field of extreme event attribution can do is compare the likelihood of an event in the observed world with its likelihood in simulations of a world without human emissions.

The result is expressed as a change in odds or intensity, not as a yes-or-no verdict. A heatwave might be assessed as thirty times more likely, or 2 degrees Celsius hotter, than it would have been without human influence. Confidence in such statements is generally high for heat extremes, moderate for heavy rainfall, and lower for individual tropical cyclones and droughts, where natural variability is larger and the physical links are more complex. Lesson 14 develops this further.

Key idea: Attribution science changes the odds rather than assigning blame for single events, with high confidence for heat extremes and lower confidence for storms and droughts.

Common misconceptions

• "A record cold week disproves global warming." No. Warming is a long-term average; individual cold spells are expected weather around a rising trend.
• "Weather and climate are the same thing." No. Weather is the daily state; climate is the multi-decade pattern.
• "El Nino is causing global warming." No. El Nino and La Nina are short-term oscillations around the trend, not its cause.
• "A cooler year than last year means warming has paused." No. Year-to-year noise of about 0.1 degrees C easily masks a decadal trend of 0.2 degrees C; only multi-year averages settle the question.
• "One year above 1.5 degrees C means we have breached the Paris limit." No. That limit refers to long-term human-caused warming, not a single year boosted by El Nino.

Recap

• Weather is hours to days; climate is 30 years or more of weather statistics.
• A climate normal is a 30-year baseline average, chosen so the trend clearly exceeds the noise.
• Global year-to-year noise is about 0.1 degrees C against a trend near 0.2 degrees C per decade; local noise is far larger.
• A single cold day is noise and cannot disprove a long-term warming trend.
• The IPCC assessed 2011-2020 as about 1.09 degrees C above 1850-1900; WMO reported 2024 at about 1.55 and 2025 at about 1.43.
• El Nino and La Nina are natural variability riding on the trend.
• Event attribution reports changes in likelihood or intensity, not a verdict that one storm was caused by warming.

Sources

1. NOAA Climate.gov ↗. (n.d.). What's the difference between global warming and climate change? climate.gov
2. IPCC. (2021). Summary for policymakers. In Climate change 2021: The physical science basis (AR6 WGI; 1.09 degrees C for 2011-2020 relative to 1850-1900). ipcc.ch
3. World Meteorological Organization. (2026). State of the Global Climate 2025 (2025 about 1.43 degrees C above 1850-1900; 2015-2025 the warmest 11 years). wmo.int
4. World Meteorological Organization. (2025). WMO confirms 2024 as warmest year on record at about 1.55 degrees C above pre-industrial level. wmo.int
5. NOAA Climate.gov ↗. (n.d.). El Nino and La Nina (El Nino-Southern Oscillation). climate.gov
6. NOAA Climate.gov ↗. (n.d.). Climate change: Global temperature. climate.gov
7. Seneviratne, S. I., Zhang, X., Adnan, M., et al. (2021). Chapter 11: Weather and climate extreme events in a changing climate. In IPCC AR6 WGI (event attribution methods and confidence levels). ipcc.ch

In 1992 a shipping container tumbled off a freighter in the North Pacific and released some 28,000 plastic bath toys into the sea. Oceanographers seized the opportunity. Over the following decades the ducks and turtles washed ashore in Alaska, Hawaii, Japan, and eventually the North Atlantic, and each landfall put a data point on the map of how the ocean actually moves. An accidental spill became one of the more charming validations of circulation models.

The big picture

The tropics receive far more sunlight than the poles, so the planet is forever trying to even things out by moving heat from the equator toward the poles. That transport happens two ways, through the atmosphere (winds and storms) and through the oceans (currents), and together they set the climate of every region on Earth.

This lesson traces how uneven heating drives global circulation, how the ocean's slow deep loop works, and why the ocean, with its enormous capacity to store heat, is the true governor of the climate.

Atmospheric circulation

Warm tropical air rises, drifts toward the poles high up, sinks back down around 30 degrees latitude (creating the great deserts there), and returns along the surface. These looping patterns are called circulation cells, and they are like a pot of water on a stove: heated fluid rises in the middle, spreads out, cools, and sinks at the edges.

Earth's rotation bends the moving air, an apparent sideways push called the Coriolis effect. This deflection organizes the winds into reliable belts, the trade winds near the tropics and the westerlies in the mid-latitudes, which steer weather systems and push on the ocean surface.

Key idea: Uneven heating sets up circulation cells, and Earth's rotation (the Coriolis effect) bends them into steady wind belts.

Three cells per hemisphere, and the climates they build

The circulation resolves into three cells in each hemisphere, and their boundaries are written across the map of world climates.

• The Hadley cell runs from the equator to roughly 30 degrees latitude. Air rises where the trade winds converge, in a band called the intertropical convergence zone, cooling as it climbs and dumping heavy rain. That rising branch is why tropical rainforests sit near the equator.
• The descending branch of the Hadley cell, near 30 degrees, brings dry air downward that warms and absorbs moisture as it sinks. Trace that latitude around the globe and you find the Sahara, the Arabian, the Kalahari, the Atacama, and the Australian deserts. The great deserts are not a coincidence; they are the exhaust of a circulation cell.
• The Ferrel cell occupies the mid-latitudes, roughly 30 to 60 degrees, and is driven indirectly by the cells on either side. Its surface winds are the prevailing westerlies, and its poleward edge is where warm and cold air masses collide along the polar front, generating the travelling storms of temperate latitudes.
• The polar cell completes the pattern, with cold dense air sinking over the poles and flowing outward at the surface.

High in the atmosphere, at the boundaries between cells, the temperature contrast drives fast ribbons of wind called jet streams. The polar jet steers mid-latitude weather systems, and because it is powered by the temperature difference between the Arctic and the mid-latitudes, researchers are actively investigating whether rapid Arctic warming is altering its behaviour. That question remains genuinely unsettled, with competing evidence and no consensus yet.

Key idea: Three cells per hemisphere set the pattern of world climates, with rainforests under the rising branch near the equator and the great deserts under the sinking branch near 30 degrees.

Coriolis, more precisely

The Coriolis effect is not a force in the ordinary sense but a consequence of describing motion on a rotating sphere. Anything moving freely across Earth's surface appears to curve: to the right in the Northern Hemisphere and to the left in the Southern. The deflection is strongest at the poles and falls to zero exactly at the equator, which is why hurricanes never form within a few degrees of it.

Combine that deflection with wind stress on the sea and you get the ocean's gyres, the great circular current systems that fill each major basin, rotating clockwise in the north and counterclockwise in the south. Their western edges carry intense, narrow, fast currents such as the Gulf Stream and the Kuroshio, while their interiors are slow and, in the case of the North Pacific and North Atlantic, are where floating debris accumulates.

Key idea: The Coriolis deflection is rightward in the north, leftward in the south, and zero at the equator, and it organises wind-driven flow into basin-scale gyres with fast western boundary currents.

Ocean circulation

The winds drag on the sea and drive surface currents like the Gulf Stream, which carries tropical warmth up into the North Atlantic and keeps northwestern Europe far milder than its latitude would suggest.

Beneath the wind-driven surface runs a slow global loop called the thermohaline circulation, sometimes nicknamed the ocean conveyor belt. It is driven by differences in water density, and density depends on temperature (thermo) and salinity (haline): cold, salty water is denser and sinks. In the North Atlantic and near Antarctica, cold salty water plunges to the deep sea, creeps along the bottom, and rises again elsewhere in a journey that takes roughly a thousand years. Where deep water rises back up is called upwelling, and it brings nutrients that feed rich fisheries.

Key idea: Winds drive warm surface currents like the Gulf Stream, while density differences drive the deep, slow thermohaline conveyor.

The AMOC, and how to talk about its risk

The Atlantic portion of that conveyor is called the Atlantic Meridional Overturning Circulation, or AMOC, and it has become one of the most discussed and most misreported topics in climate science. The concern is straightforward in principle: the sinking that drives it depends on surface water becoming dense enough, and both warming and the addition of fresh meltwater from Greenland make surface water less dense.

Precision matters here, so consider what the IPCC actually assessed. The AMOC is very likely to weaken over the twenty-first century under all emission scenarios. However, an abrupt collapse before 2100 is assessed as unlikely, with only medium confidence attached to that judgement, and the assessment explicitly notes that if such a collapse occurred it would produce abrupt shifts in regional weather and the water cycle. Some studies published since have argued the risk is higher than that, and the question is genuinely active.

Notice the shape of this uncertainty, because it recurs throughout climate science. The direction of change is well established: weakening. The magnitude and timing are uncertain. The possibility of a threshold response is real but not well quantified. Reporting that treats a low-probability, high-impact possibility as either a certainty or a fantasy misrepresents the science in both directions.

Key idea: The AMOC is very likely to weaken this century, while an abrupt collapse before 2100 is assessed as unlikely with only medium confidence, an example of confident direction paired with uncertain magnitude.

Why the ocean rules the climate

Water has an enormous heat capacity, meaning it takes a lot of energy to change its temperature, the reason a coastal city has milder swings than an inland desert. The IPCC assessed that ocean heat uptake accounted for about 91 percent of the change in the global energy inventory between 1971 and 2018, with the remainder shared among melting ice, warming land, and the atmosphere. That is why sea level rises and ice keeps melting even in years when the air temperature wobbles: most of the extra energy is going into the water, not the air.

The size of the reservoir explains the modest temperature change. Gram for gram, water absorbs roughly four times as much energy per degree as air, and the ocean outweighs the atmosphere by a factor near 270. Warming the top two kilometers of the ocean by a few hundredths of a degree therefore stores an amount of energy that would have warmed the atmosphere dramatically. Ocean heat content is consequently a far steadier indicator of the planet's energy imbalance than surface air temperature, and it has set new records in recent years without the year-to-year swings that ENSO imposes on the air.

The ocean also absorbs about a quarter of human CO₂ emissions. This makes it both a great moderator of climate and a system under growing stress, since the added heat and CO₂ alter currents, oxygen levels, and seawater chemistry, topics we take up in later modules.

Key idea: Water's huge heat capacity lets the ocean absorb over 90 percent of the extra heat, so it moderates climate while quietly storing most of global warming's energy.

A small worked example

Consider two years of global warming's energy. If the ocean takes up about 90 percent of the excess heat and everything else (air, land, ice) shares the remaining 10 percent, then for every 100 units of extra energy trapped, about 90 go into the sea and only about 10 are left to warm the atmosphere, melt ice, and heat the land combined. That lopsided split is exactly why air temperature alone understates how much energy the planet is accumulating.

Common misconceptions

• "Winds and currents exist for no particular reason." No. They are driven by the temperature difference between the hot tropics and cold poles.
• "The deep ocean conveyor is driven by wind." No. Surface currents are wind-driven, but the deep thermohaline circulation is driven by density from temperature and salinity.
• "Global warming shows up mostly in the air." No. About 91 percent of the extra heat goes into the ocean, not the atmosphere.
• "The Gulf Stream is about to shut off." No. The wind-driven Gulf Stream will persist as long as the winds do. The concern is a weakening of the density-driven overturning it feeds, which is very likely but assessed as unlikely to collapse abruptly before 2100.
• "Deserts occur wherever it happens to be hot." No. The great subtropical deserts sit under the descending branch of the Hadley cell near 30 degrees latitude, where sinking air suppresses rainfall.

Recap

• Uneven solar heating drives heat from the tropics to the poles.
• Three cells per hemisphere plus the Coriolis effect create steady wind belts, tropical rainforests, and subtropical deserts.
• Coriolis deflection is rightward in the north, leftward in the south, and zero at the equator, organising ocean gyres.
• Wind-driven surface currents like the Gulf Stream carry warmth poleward.
• The thermohaline conveyor is driven by density differences in temperature and salinity.
• The AMOC is very likely to weaken this century; abrupt collapse before 2100 is assessed as unlikely with medium confidence.
• Water's large heat capacity let the ocean absorb about 91 percent of the extra energy over 1971-2018.

Sources

1. NOAA National Ocean Service. (n.d.). Currents. Ocean Service Education. oceanservice.noaa.gov
2. Fox-Kemper, B., Hewitt, H. T., Xiao, C., et al. (2021). Chapter 9: Ocean, cryosphere and sea level change. In IPCC AR6 WGI (AMOC assessment; ocean heat uptake). ipcc.ch
3. Forster, P., Storelvmo, T., Armour, K., et al. (2021). Chapter 7: The Earth's energy budget, climate feedbacks, and climate sensitivity. In IPCC AR6 WGI (ocean took about 91 percent of the energy inventory change, 1971-2018). ipcc.ch
4. NOAA. (n.d.). Ocean currents. NOAA Education. noaa.gov
5. NOAA Climate.gov ↗. (n.d.). Climate change: Ocean heat content. climate.gov
6. NASA Science. (n.d.). Earth indicators: Ocean warming. climate.nasa.gov
7. NOAA. (n.d.). The atmosphere. JetStream (global circulation and jet streams). noaa.gov

Three kilometers beneath a research station on the East Antarctic plateau, a drill has been cutting cylinders of ice for decades. Each cylinder is a slice of the past, and the deepest of them contain snow that fell before our species existed. Crack one open in a cold laboratory and you release air that last touched the atmosphere hundreds of thousands of years ago. It is the closest thing science has to a time machine, and it settled the question of whether today's carbon dioxide is unusual.

The big picture

Thermometers only go back a couple of centuries, so to see the deeper past scientists read nature's own records, called proxies. A proxy is like a diary written by the environment: tree rings, ocean mud, coral bands, and ancient ice each preserve clues about the conditions when they formed.

These records do two jobs. They show how far today's carbon dioxide sits outside the natural range, and they let scientists identify the natural drivers of past climate change, and then rule those drivers out as the cause of the recent warming.

How proxies record the past

A proxy is a natural recorder that preserves a signal of former conditions, letting us reconstruct climate before instruments existed (the field of paleoclimate). Several proxies work together:

• Tree rings record year-by-year growth; wide rings mean good growing years, narrow rings mean stress like drought.
• Ocean and lake sediments pile up in layers holding the shells of tiny organisms whose chemistry reflects the temperature of the water they lived in.
• Corals lay down annual bands, like tree rings underwater.
• Ice cores, cylinders of ancient ice drilled from Antarctica and Greenland, trap bubbles of old air and preserve annual snow layers.

Key idea: Proxies such as tree rings, sediments, corals, and ice cores are nature's diaries, letting us reconstruct climate long before thermometers.

How a proxy becomes a thermometer

A proxy is useless until it is calibrated, and the calibration is what makes paleoclimatology a quantitative science rather than a collection of anecdotes. The most important tool is oxygen isotope analysis. Ordinary water contains a mixture of oxygen-16 and the heavier oxygen-18. Lighter molecules evaporate slightly more readily and heavier ones condense slightly more readily, and both preferences depend on temperature, so the ratio of the two isotopes preserved in ice or in a marine shell encodes the conditions when it formed.

Researchers establish the relationship by measuring the isotope ratio in modern samples whose temperature is known independently, then apply that calibrated relationship to older material. The same discipline applies to tree rings, calibrated against instrumental rainfall and temperature records from the same region, and to coral bands, calibrated against measured sea surface temperatures.

Every proxy carries a specific weakness, which is why reconstructions combine several. Tree rings are annually resolved but only reach back a few thousand years and mostly record growing-season conditions. Marine sediments span millions of years but blur decades together because burrowing animals mix the mud. Ice cores are superb for atmospheric composition but exist only where thick ice has survived. Where independent proxies with different weaknesses agree, confidence in the reconstruction is high.

Key idea: Proxies are calibrated against modern instrumental records, chiefly using oxygen isotope ratios, and confidence comes from independent proxies with different weaknesses agreeing.

What ice cores tell us

Ice cores are the star witness because they hold actual samples of the ancient atmosphere. As snow buries and compresses into ice, it seals tiny bubbles of the air of that time, so drilling down and melting the ice releases air from the distant past to be measured directly.

The Antarctic cores drilled by the EPICA project reach back roughly 800,000 years, and the follow-up Beyond EPICA project reached bedrock in 2025 with ice expected to extend the record further still. They show that CO₂ and temperature have marched up and down together through the ice ages, with CO₂ swinging between about 180 ppm in cold glacial periods and about 280 ppm in warm interglacials (an interglacial being a warm gap between ice ages, like the present). Crucially, the 2025 global mean of 425.6 ppm is far above anything in that entire 800,000-year record, and it rose in about a century rather than over millennia. The IPCC concluded with high confidence that CO₂ is now higher than at any point in at least 2 million years.

Key idea: Ice cores sample ancient air directly and show CO₂ stayed between about 180 and 280 ppm for 800,000 years, making today's 425.6 ppm higher than at any time in at least 2 million years.

A small worked example

How far outside the natural range is today's CO₂? The natural glacial-to-interglacial swing spanned 280 minus 180, which is 100 ppm, and that swing took thousands of years. Today's level of 425.6 ppm sits about 146 ppm above the top of that natural range, a jump larger than the entire ice-age swing, achieved in roughly a century.

Now compare the rates, which matters even more than the levels. Coming out of the last ice age, CO₂ rose by roughly 100 ppm over about 10,000 years, an average of 0.01 ppm per year. Over the decade 2014 to 2023 the measured increase averaged about 2.5 ppm per year. Dividing 2.5 by 0.01 gives a factor of roughly 250. Even taking the fastest natural intervals rather than the average, the modern rise is on the order of a hundred times faster than anything the ice cores record. That comparison is why scientists call the modern change a spike rather than a cycle, and why ecosystems that survived past changes may struggle with this one: it is not only the destination but the speed of travel.

Natural drivers of past change

Climate changed naturally long before humans, driven by three main forces:

• Milankovitch cycles: slow, regular variations in Earth's orbit and the tilt of its axis that change how sunlight is distributed over the globe and pace the ice ages over tens of thousands of years.
• Volcanic forcing: large eruptions inject reflective particles into the stratosphere that can cool the planet for a year or two.
• Solar changes: small variations in the Sun's output nudge temperature a little.

Understanding these natural drivers is essential, because it lets scientists show that none of them can account for the recent warming, whereas the measured rise in greenhouse gases can. Ruling out the alternatives is a central part of how the human cause was established.

Key idea: Milankovitch cycles, volcanoes, and solar changes drove past climate shifts, but none of them can explain the recent warming, which greenhouse gases do.

Did CO₂ lead or lag the ice ages?

Careful reading of the ice cores shows that at the end of past glacial periods, temperature in Antarctica began rising several centuries before CO₂ did. This is sometimes presented as evidence that carbon dioxide cannot drive temperature. It is a genuine observation and a serious misreading of it.

The sequence makes sense once you separate the trigger from the amplifier. Milankovitch cycles change the distribution of sunlight but supply far too little energy on their own to end an ice age. The extra summer sunlight in the north begins the melt, warming oceans then release dissolved CO₂, and the added greenhouse gas amplifies and globalises the warming. Roughly speaking, the orbital nudge starts the process and greenhouse feedback does most of the work. Analyses of globally distributed proxies find that global temperature and CO₂ rose largely in step, with CO₂ leading global temperature even where it lagged local Antarctic temperature.

Note the logical structure, because it generalises. A factor that acted as a feedback in one situation can act as a forcing in another. Nothing in the ice-core record implies that adding carbon dioxide directly, as humans are doing now, would fail to warm the planet. It confirms the opposite: a modest CO₂ change of about 100 ppm helped drive the several-degree swings between glacial and interglacial worlds.

Key idea: CO₂ lagged local Antarctic temperature by centuries at glacial terminations because orbital changes triggered the warming and CO₂ amplified it, which demonstrates rather than undermines its power.

Deeper time: when CO₂ was last this high

Ice cores stop at hundreds of thousands of years, but other proxies reach much further. Boron isotopes in fossil shells and the density of pores on fossil leaves both track ancient CO₂, and marine sediments preserve temperature and sea level. Two intervals are especially instructive.

The mid-Pliocene, roughly 3 million years ago, is the most recent period with CO₂ in the neighbourhood of today's level. Global temperature then was several degrees warmer than pre-industrial, and sea level stood many meters higher, because the ice sheets were smaller. That world is not a forecast, since the Pliocene had millennia to equilibrate while we have had a century, but it indicates the direction in which the slow parts of the system are heading.

Further back, the Paleocene-Eocene Thermal Maximum about 56 million years ago saw a massive natural release of carbon and global warming of roughly 5 degrees Celsius, accompanied by ocean acidification and substantial extinction among deep-sea organisms. It is often cited as the closest natural analogue to today, with one crucial difference: that carbon was released over thousands of years, whereas the modern release is happening over roughly two centuries.

Key idea: The mid-Pliocene, the last time CO₂ approached today's level, was several degrees warmer with sea level many meters higher, and even the abrupt PETM released its carbon far more slowly than humans are doing now.

Common misconceptions

• "Climate has always changed, so today's change must be natural too." That past change was natural does not make present change natural; the known natural drivers cannot account for the recent warming, but greenhouse gases can.
• "Ice cores are made of frozen CO₂." No. They are frozen water; the CO₂ is measured from tiny bubbles of trapped ancient air.
• "Today's CO₂ is normal for Earth's history." No. It is far above the entire 800,000-year ice-core range, higher than at any time in at least 2 million years, and rising roughly a hundred times faster than natural swings.
• "CO₂ lagged temperature in the ice cores, so it cannot cause warming." No. Orbital changes triggered deglaciation and CO₂ amplified it; globally, CO₂ and temperature rose largely together.
• "Proxies are guesswork." No. They are calibrated against instrumental measurements, and independent proxies with different weaknesses are required to agree.

Recap

• Proxies are natural recorders used to reconstruct past climate (paleoclimate), calibrated against modern instruments.
• Oxygen isotope ratios are the workhorse proxy, because evaporation and condensation sort isotopes by temperature.
• Ice cores trap ancient air, letting us measure past atmospheric composition directly.
• Over 800,000 years, CO₂ cycled between about 180 and 280 ppm.
• Today's 425.6 ppm is higher than at any time in at least 2 million years and is rising about a hundred times faster.
• CO₂ acted as an amplifier at glacial terminations, lagging local Antarctic temperature but driving global change.
• Milankovitch cycles, volcanoes, and solar changes drove past shifts but not the recent warming.

Sources

1. NOAA National Centers for Environmental Information. (n.d.). Paleoclimatology. ncei.noaa.gov
2. NOAA National Centers for Environmental Information. (n.d.). Ice core paleoclimatology. ncei.noaa.gov
3. Luthi, D., Le Floch, M., Bereiter, B., et al. (2008). High-resolution carbon dioxide concentration record 650,000-800,000 years before present. Nature, 453, 379-382. nature.com
4. Gulev, S. K., Thorne, P. W., Ahn, J., et al. (2021). Chapter 2: Changing state of the climate system. In IPCC AR6 WGI (CO₂ higher than at any time in at least 2 million years). ipcc.ch
5. NASA Science. (n.d.). Milankovitch (orbital) cycles and their role in Earth's climate. science.nasa.gov
6. NASA Earth Observatory. (n.d.). Paleoclimatology: Explaining the evidence. NASA Science. earthobservatory.nasa.gov
7. Beyond EPICA. (n.d.). Beyond EPICA: Oldest Ice (deep Antarctic drilling project). beyondepica.eu

Detectives distrust a case built on one witness. They look for corroboration from sources that could not have colluded: a fingerprint, a receipt, a security camera, a stranger's memory. If all of them point the same way, the case is strong not because any one is perfect but because their errors are unrelated. Climate science works the same way, and this lesson assembles the witnesses.

The big picture

The conclusion that Earth is warming does not rest on any single dataset. It comes from many independent lines of evidence, separate instruments and indicators that all point the same way, like several witnesses who never met telling the same story.

Knowing the planet warmed is one question; knowing why is another, called attribution. Several distinct fingerprints point specifically at greenhouse gases rather than the Sun, and that is why the world's scientific bodies now regard human-caused warming as settled.

Many independent lines of evidence

The evidence for warming comes from unrelated sources that agree:

• Surface thermometers on land and sea show a global average rise of about 1.09 degrees Celsius for 2011 to 2020 relative to 1850 to 1900.
• Satellites independently confirm warming of the lower atmosphere.
• The oceans are measurably warmer and are rising, both because water expands as it warms (thermal expansion) and because land ice is melting.
• Glaciers are retreating on every continent, and Arctic sea ice has shrunk dramatically.
• The Greenland and Antarctic ice sheets are losing mass.
• Growing seasons have lengthened and species ranges have shifted poleward and upslope.

When physical instruments and living things independently agree, the signal is robust. It would take an implausible coincidence for all of them to be wrong in the same direction.

Key idea: Warming is confirmed by many independent lines of evidence, thermometers, satellites, oceans, ice, and biology, that all point the same way.

The indicators, with numbers attached

Vague statements about change are hard to check, so it is worth pinning each indicator to a measured quantity, a source, and a date. All of these figures are revised as observations accumulate, so treat them as a snapshot and consult the linked monitoring pages for current values.

Indicator	Observed change	Source
Global surface temperature	about +1.09 degrees C, 2011-2020 vs 1850-1900	IPCC AR6 WGI
Warmest year on record	2024, about 1.55 degrees C above 1850-1900	WMO
Global mean sea level	about +21 to 24 cm since 1880	NOAA Climate.gov ↗
September Arctic sea ice	shrinking about 12.2 percent per decade	NASA / NSIDC
Greenland ice sheet	losing roughly 264 gigatonnes per year since 2002	NASA GRACE and GRACE-FO
Antarctic ice sheet	losing roughly 135 gigatonnes per year since 2002	NASA GRACE and GRACE-FO
Atmospheric CO2	425.6 ppm global mean in 2025, from about 278 ppm	NOAA GML

Consider what would have to be true for this to be an artefact. Satellites weighing the gravitational pull of ice sheets, tide gauges and radar altimeters measuring sea level, thermometers in ships and buoys, and biologists recording earlier flowering dates all use different instruments, different physics, and different research communities. A shared error is not credible.

Key idea: Each indicator has a measured value from a distinct instrument and community, and no single error could produce agreement across gravity satellites, tide gauges, thermometers, and biological records.

Attribution: the human fingerprint

How do we know greenhouse gases, not the Sun, are responsible? The clues are specific patterns that only one cause can produce, much as a detective distinguishes suspects by their fingerprints.

• Stratospheric cooling: if a brighter Sun were the cause, every layer of the atmosphere would warm. Instead, the lower atmosphere warms while the stratosphere cools, exactly the pattern expected when greenhouse gases trap heat below.
• Nights are warming faster than days, and winters faster than summers, again the greenhouse signature rather than the solar one.
• The extra CO₂ carries an isotopic signature, a ratio of carbon isotopes, showing it comes from ancient plant-derived fossil carbon rather than volcanoes or the ocean.
• The amount of extra CO₂ in the air matches the known quantity of fossil fuel humans have burned.

Climate models can only reproduce the observed warming when human emissions are included; natural factors alone fall far short. That match is the quantitative core of attribution.

Key idea: Fingerprints like stratospheric cooling, faster night warming, and the fossil isotopic signature point specifically to greenhouse gases, not the Sun.

How attribution is quantified

Attribution is not an impression; it is a statistical procedure with published uncertainty ranges. Researchers run climate models twice, once with all known influences and once with natural influences alone, and compare the resulting spatial and vertical patterns with observations. The technique is called optimal fingerprinting, and it asks how much of each candidate pattern must be present to reproduce what the instruments actually recorded.

The IPCC's Sixth Assessment Report presents the result as a set of contributions to the 1.09 degrees Celsius of observed warming for 2011 to 2020. Human influence contributed a best estimate of 1.07 degrees Celsius, with a likely range of 0.8 to 1.3. Well-mixed greenhouse gases alone contributed about 1.0 to 2.0 degrees, partly offset by aerosols contributing about minus 0.8 to 0.0. Natural drivers, solar and volcanic combined, contributed between minus 0.1 and plus 0.1 degrees Celsius, and internal variability likewise between minus 0.2 and plus 0.2.

Read those numbers carefully. The human contribution is essentially the entire observed warming, and the natural contribution is indistinguishable from zero. It is on this basis that the report states, in unusually direct language for a consensus document, that it is unequivocal that human influence has warmed the atmosphere, ocean, and land.

Key idea: Optimal fingerprinting attributes a best estimate of 1.07 degrees C of the observed 1.09 degrees C to human influence, with natural drivers contributing between minus 0.1 and plus 0.1.

Measuring the greenhouse effect directly

One further line of evidence deserves separate mention, because it removes any need to rely on models at all. Instruments have measured the greenhouse effect strengthening in real time.

Satellites comparing the infrared spectrum leaving Earth in the 1970s with the same spectrum decades later found that less energy escapes precisely at the wavelengths absorbed by carbon dioxide and methane, and by the amount predicted. Complementing this from below, surface spectrometers in Oklahoma and Alaska recorded the downward infrared radiation from the sky over an eleven-year period and detected an increase attributable specifically to rising carbon dioxide, again matching prediction.

These experiments close the loop. The physics of infrared absorption is demonstrated in the laboratory, the resulting reduction in escaping heat is measured from orbit, the resulting increase in downward heat is measured at the ground, and the accumulated energy is measured in the ocean. Every link in the causal chain has been observed, not merely inferred.

Key idea: Satellites have measured less infrared escaping at CO₂'s wavelengths and ground instruments have measured more infrared arriving, confirming the mechanism by direct observation.

The scientific consensus

As a result, the world's scientific bodies, and the Intergovernmental Panel on Climate Change (IPCC) that synthesizes the research, conclude it is unequivocal that human influence has warmed the atmosphere, ocean, and land. This is not a matter of opinion or a single model; it is a convergent conclusion from physics, chemistry, and a wall of observations.

The word consensus is often misunderstood, so it is worth being exact about what has been measured. A 2021 survey of the peer-reviewed literature examined a large sample of papers published since 2012 and found agreement with human-caused warming exceeding 99 percent among papers that expressed a position. Earlier surveys of the literature and of publishing climate scientists returned figures in the 90 to 97 percent range using different samples and methods.

Crucially, consensus is a description of where the evidence has led, not an argument in itself. Scientists did not vote on the greenhouse effect. The agreement exists because independent researchers, using different data and often hoping to find something new, kept arriving at the same conclusion. The IPCC does not conduct research; it assesses thousands of published studies, and its reports are reviewed line by line, including by governments that have every incentive to contest inconvenient findings.

Key idea: The IPCC concludes human influence on warming is unequivocal, and over 99 percent of position-taking papers in the recent literature agree, because independent lines of evidence converged rather than because anyone voted.

What would change scientists' minds

A useful test of any scientific claim is to ask what evidence would overturn it. Here the answer is clear and short. If the stratosphere were warming rather than cooling, the greenhouse explanation would be in serious trouble. If the extra carbon in the air carried the isotope signature of volcanoes rather than of ancient plants, the source would be wrong. If satellites found more heat escaping at CO₂'s wavelengths, the mechanism would fail. If the ocean were losing heat while the air warmed, the energy accounting would not close.

Every one of these tests has been run, and every one came out the other way. That is why the conclusion is stated so strongly. It is not that the possibility of being wrong was never entertained; it is that the specific ways of being wrong were checked.

Key idea: The human-cause conclusion is testable, and each way it could have failed, from stratospheric warming to a volcanic isotope signature, has been checked and ruled out.

Common misconceptions

• "It's just the Sun." No. A brighter Sun would warm all atmospheric layers; instead the stratosphere cools, which fingerprints greenhouse gases.
• "The evidence is one adjusted thermometer record." No. Satellites, oceans, ice, and biology give independent confirmation.
• "Scientists are evenly divided on the cause." No. Over 99 percent of position-taking papers in the recent peer-reviewed literature agree humans are the main cause.
• "Attribution rests entirely on computer models." No. Satellites and ground spectrometers have directly measured the change in infrared radiation at the specific wavelengths greenhouse gases absorb.
• "Natural variability could account for the warming." No. The IPCC assessed the combined contribution of solar and volcanic drivers as between minus 0.1 and plus 0.1 degrees Celsius, against 1.09 degrees observed.

Recap

• Many independent indicators, instruments and biology alike, confirm warming.
• Measured changes include 1.09 degrees C of warming, 21-24 cm of sea-level rise, Arctic sea ice down 12.2 percent per decade, and Greenland losing about 264 gigatonnes a year.
• Attribution means determining the cause of the observed change, using optimal fingerprinting with published uncertainty ranges.
• Human influence contributed a best estimate of 1.07 degrees C; natural drivers contributed between minus 0.1 and plus 0.1.
• Stratospheric cooling, faster night warming, and the fossil isotopic signature fingerprint greenhouse gases, not the Sun.
• Satellites and surface instruments have directly measured the greenhouse effect strengthening.
• The IPCC finds human influence on the climate is unequivocal, and over 99 percent of relevant papers agree.

Sources

1. IPCC. (2021). Summary for policymakers. In Climate change 2021: The physical science basis (AR6 WGI; human contribution 1.07 degrees C, natural minus 0.1 to plus 0.1). ipcc.ch
2. Eyring, V., Gillett, N. P., Achuta Rao, K. M., et al. (2021). Chapter 3: Human influence on the climate system. In IPCC AR6 WGI. ipcc.ch
3. NASA Science. (n.d.). Evidence: How do we know climate change is real? science.nasa.gov
4. NASA Science. (n.d.). Scientific consensus: Earth's climate is warming. science.nasa.gov
5. Lynas, M., Houlton, B. Z., & Perry, S. (2021). Greater than 99% consensus on human caused climate change in the peer-reviewed scientific literature. Environmental Research Letters, 16(11), 114005. iopscience.iop.org
6. NASA Science. (n.d.). Earth indicators: Ice sheets (GRACE and GRACE-FO mass loss rates). climate.nasa.gov
7. NASA Science. (n.d.). Earth indicators: Arctic sea ice minimum extent (12.2 percent per decade). climate.nasa.gov

In 1967, working at a laboratory in Princeton with a computer far less capable than a modern phone, Syukuro Manabe and Richard Wetherald published a calculation of how the atmosphere's temperature would respond to doubled carbon dioxide. They obtained about 2 degrees Celsius. More than half a century of vastly more powerful models, satellite observations, and paleoclimate reconstructions have refined that estimate without overturning it, and in 2021 the work earned a share of the Nobel Prize in Physics.

The big picture

A climate model is a computer program that represents the atmosphere, ocean, land, and ice as a three-dimensional grid and applies the laws of physics to compute how the climate evolves. Models are not crystal balls or curve-fits; they are physics simulations, much like the ones used to design aircraft.

The direct warming from more CO₂ is modest on its own. What decides how much warming we actually get are feedbacks, the amplifying and damping loops introduced in Module 1. This lesson explains how models are tested, which feedbacks matter, and how to read a projection honestly.

What a climate model is, and how it is tested

A climate model divides the planet into a grid of boxes and, in each box, enforces conservation of energy, mass, and momentum, the same physics that governs a boiling pot or a weather system. It then steps forward in time to see how temperature, winds, and currents evolve.

Models are tested by hindcasting, which means running them on the past and checking the result against what actually happened. Run backward, good models reproduce the seasonal cycle, the brief cooling after large volcanic eruptions, past ice ages, and the observed warming of the last century. That track record is why their projections are taken seriously, while their imperfections are openly quantified rather than hidden.

Key idea: Climate models are physics simulations tested by hindcasting, reproducing known climate history before being trusted with the future.

Inside the machine: grids, timesteps, and parameterisation

A modern global model divides the atmosphere into cells typically 25 to 100 kilometers across and several dozen layers deep, with a matching ocean grid, and advances the whole thing in steps of minutes for simulated centuries. That resolution is the central engineering constraint of the field, and it creates a specific problem.

Many processes that matter are smaller than a grid cell. An individual cumulus cloud is a few kilometers across; a raindrop is millimeters. Such processes cannot be simulated directly, so they are represented by parameterisations, physically motivated formulas that predict the average behaviour within a cell from the conditions in that cell. Convection, cloud formation, and turbulence are all handled this way.

Parameterisation is where honest uncertainty lives. It is not a fudge, since the formulas are constrained by field campaigns, laboratory work, and high-resolution simulations of small regions, but different reasonable choices produce different results. This is precisely why clouds dominate the spread across models, and why increasing computer power, which allows finer grids and fewer parameterisations, remains a genuine scientific priority rather than a mere convenience.

Key idea: Models resolve grid cells tens of kilometers across, so clouds and convection must be parameterised, and those parameterisation choices are the main source of disagreement between models.

Feedbacks decide how much warming

Recall the microphone-squeal image of a positive feedback. Several such loops set the final amount of warming:

• The water-vapor feedback is the strongest amplifier: warming lets air hold more water vapor, which is itself a greenhouse gas, which warms further.
• The ice-albedo feedback adds more: melting bright ice exposes dark ocean and land that absorb more sunlight.
• Clouds are the biggest uncertainty, because they can both cool (by reflecting sunlight) and warm (by trapping heat), and their net effect is hard to pin down.
• Permafrost thaw is a dangerous slow feedback that can release stored carbon as CO₂ and methane, adding still more warming.

Because amplifying feedbacks dominate, the total warming expected from a doubling of CO₂, called the climate sensitivity, is likely around 3 degrees Celsius, with a range that reflects the feedback uncertainties, especially clouds.

Key idea: Feedbacks, led by water vapor and ice-albedo, amplify the modest direct warming, giving a climate sensitivity of about 3 degrees Celsius for doubled CO₂.

The feedback ledger

Feedbacks are quantified as a change in the planet's net energy flow per degree of warming, in watts per square meter per degree Celsius. Negative values stabilise, positive values amplify. The IPCC's assessed central values are approximately as follows.

Feedback	Strength (W/m2 per degree C)	Effect
Planck response	about -3.2	Stabilising: a warmer planet radiates more
Water vapor plus lapse rate	about +1.3	Amplifying, and the largest positive term
Surface albedo	about +0.35	Amplifying: melting ice exposes dark surfaces
Clouds	about +0.4, plus or minus roughly 0.35	Amplifying on balance, and by far the least certain
Net total	about -1.2	Stabilising overall, so the system settles

Two conclusions fall out of this table. First, the net feedback is negative, which is why Earth's climate settles at a new equilibrium rather than running away. Second, the uncertainty on clouds alone is comparable in size to the entire albedo feedback, which is why narrowing cloud uncertainty has done more than anything else to narrow the sensitivity range.

Key idea: The Planck response of about -3.2 W/m2 per degree C stabilises the system, water vapor adds about +1.3, and cloud uncertainty of roughly plus or minus 0.35 dominates the remaining spread.

Climate sensitivity: the number and its honest range

Two related quantities are used, and confusing them causes trouble. Equilibrium climate sensitivity is the eventual warming once the whole system, including the slow deep ocean, has adjusted to doubled CO₂. Transient climate response is the warming at the moment of doubling under a steady 1 percent annual increase, which is closer to what this century will actually experience.

The Sixth Assessment Report gives a best estimate for equilibrium sensitivity of 3 degrees Celsius, a likely range of 2.5 to 4, and a very likely range of 2 to 5. For the transient response, the best estimate is 1.8 degrees Celsius with a likely range of 1.4 to 2.2. These ranges are narrower than in the previous assessment, and the narrowing came from combining three independent constraints: process understanding of individual feedbacks, the historical warming record, and paleoclimate reconstructions of past warm and cold states.

Be careful how you interpret the range. It does not mean that scientists cannot tell whether adding CO₂ warms the planet; the entire range lies well above zero. It means the size of the response is uncertain by roughly a factor of two, which matters enormously for policy but not at all for the direction. Uncertainty is also not automatically reassuring: values at the high end of the range would be considerably worse than the central estimate.

Key idea: Equilibrium climate sensitivity has a best estimate of 3 degrees C with a likely range of 2.5 to 4, so the magnitude is uncertain by about a factor of two while the direction is not in question.

Projections, not prophecies

Models are run under different emissions scenarios, assumed future paths of greenhouse-gas emissions, because the biggest unknown is human choice, not physics. The current generation are called Shared Socioeconomic Pathways, and the IPCC reports best-estimate warming for 2081 to 2100 relative to 1850 to 1900 as follows.

Scenario	Description	Best estimate	Very likely range
SSP1-1.9	Very low emissions, net zero around 2050	1.4 degrees C	1.0 to 1.8
SSP1-2.6	Low emissions, net zero after 2050	1.8 degrees C	1.3 to 2.4
SSP2-4.5	Intermediate emissions	2.7 degrees C	2.1 to 3.5
SSP3-7.0	High emissions, roughly doubling by 2100	3.6 degrees C	2.8 to 4.6
SSP5-8.5	Very high emissions	4.4 degrees C	3.3 to 5.7

Reading a projection therefore means reading its scenario and its range, not a single number. Work the comparison: the gap between the lowest and highest pathways is 4.4 minus 1.4, which is 3 degrees Celsius. For scale, the difference between the depth of the last ice age and the pre-industrial world was roughly 5 to 6 degrees Celsius globally. Human choices over this century therefore span more than half an ice age of climate change.

Note also that the scenario ranges overlap the sensitivity uncertainty. Even under a single emissions path, the outcome is a distribution rather than a point, because the physics response is itself uncertain. The models agree on the essentials: more greenhouse gas means more warming, and the amount depends chiefly on how much we emit.

Key idea: Projections are conditional on scenarios, ranging from a best estimate of 1.4 degrees C under the lowest pathway to 4.4 under the highest, a spread of 3 degrees driven by human choices.

How have past projections performed?

A projection made decades ago can now be checked against what happened, which is the fairest available test. Studies comparing climate model projections published between the 1970s and 2000s with subsequent observations found that most were skilful: the majority projected warming rates statistically consistent with what was later measured, once the actual emissions and forcings that occurred were accounted for.

That last qualification matters. A model asked to project warming under an assumed emissions path cannot be blamed if the world followed a different path. When historical projections are evaluated against the forcings that actually materialised, their temperature predictions hold up well. Early models did have real limitations, notably in ocean heat uptake and aerosols, and the field has corrected them. But the record does not support the claim that models systematically exaggerate.

Key idea: Projections published from the 1970s onward have mostly proved skilful when evaluated against the emissions that actually occurred, which is the appropriate test.

Common misconceptions

• "Models are just curve-fits to past data." No. They are built from physical laws and validated by reproducing climate they were not tuned to, like volcanic cooling and ice ages.
• "A single projection number is the prediction." No. Projections come as scenario-dependent ranges, because future emissions are a human choice.
• "Feedbacks are too uncertain to say anything." No. Water vapor and ice-albedo are well understood; the main uncertainty is clouds, and it is quantified.
• "The uncertainty range means the science is unsettled." No. The whole range lies well above zero warming. The uncertainty concerns magnitude, not direction, and high-end values are worse, not safer.
• "Models have consistently over-predicted warming." No. Evaluated against the emissions that actually occurred, most projections since the 1970s match subsequent observations.

Recap

• Climate models are physics simulations on a 3D grid, not curve-fits.
• Grid cells tens of kilometers wide force clouds and convection to be parameterised, the main source of model spread.
• Hindcasting tests them against known climate history.
• The Planck response of about -3.2 W/m2 per degree C stabilises; water vapor adds about +1.3 and clouds about +0.4 with large uncertainty.
• Equilibrium climate sensitivity has a best estimate of 3 degrees C, likely 2.5 to 4, very likely 2 to 5.
• Projections range from about 1.4 degrees C under SSP1-1.9 to 4.4 under SSP5-8.5 for 2081-2100.
• Past projections have mostly proved skilful when judged against the forcings that actually occurred.

Sources

1. Forster, P., Storelvmo, T., Armour, K., et al. (2021). Chapter 7: The Earth's energy budget, climate feedbacks, and climate sensitivity. In IPCC AR6 WGI (feedback parameters; ECS 3 degrees C, likely 2.5-4). ipcc.ch
2. Lee, J.-Y., Marotzke, J., Bala, G., et al. (2021). Chapter 4: Future global climate: Scenario-based projections and near-term information. In IPCC AR6 WGI. ipcc.ch
3. IPCC. (2021). Summary for policymakers. In Climate change 2021: The physical science basis (AR6 WGI; SSP warming table for 2081-2100). ipcc.ch
4. NOAA Geophysical Fluid Dynamics Laboratory. (n.d.). Climate modeling. gfdl.noaa.gov
5. The Nobel Foundation. (2021). The Nobel Prize in Physics 2021 (Manabe and Hasselmann, for physical modelling of Earth's climate). nobelprize.org
6. Carbon Brief. (2018). Explainer: How scientists estimate climate sensitivity. carbonbrief.org
7. Schuur, E. A. G., McGuire, A. D., Schadel, C., et al. (2015). Climate change and the permafrost carbon feedback. Nature, 520, 171-179. nature.com

The Netherlands has been arguing with the sea for eight hundred years and currently keeps about a quarter of its territory below sea level behind dikes, dunes, and storm barriers. It works, at enormous and permanent expense, for a wealthy country with excellent engineering and gentle geology. Most of the world's threatened coastline has none of those advantages. Understanding what sea level is doing, and why, is therefore not an abstract exercise for a large fraction of humanity.

The big picture

Global mean sea level has risen roughly 21 to 24 centimeters since 1880, and the rate is accelerating. Two things cause it: warmer water takes up more room, and melting land ice adds new water to the sea.

The single most important idea in this lesson is a distinction people often get backwards. Melting floating sea ice does not raise sea level, but melting land ice does. Getting that right explains why the Greenland and Antarctic ice sheets, not the Arctic ice pack, are the ones to watch.

The two causes

Sea-level rise has two dominant drivers:

• Thermal expansion: as the ocean absorbs heat, the water expands and takes up more volume. Water, like most things, swells a little when it warms, and spread across an entire ocean that adds up.
• Melting land ice: water added from shrinking mountain glaciers and, increasingly, the vast Greenland and Antarctic ice sheets.

Key idea: Sea level rises mainly from thermal expansion of warming water plus water added by melting land ice.

The rate, and its acceleration

Sea-level rise is one of the most carefully monitored quantities in Earth science, measured by coastal tide gauges since the nineteenth century and by satellite radar altimeters continuously since 1993. The two records agree where they overlap, and together they show not merely rise but acceleration.

Period	Average rate
1901 to 1971	about 1.3 mm per year
1971 to 2006	about 1.9 mm per year
2006 to 2018	about 3.7 mm per year

Work out what the acceleration means. The rate for 2006 to 2018 is roughly 2.8 times the rate for the first seven decades of the twentieth century. Sustained at 3.7 mm per year, a century would add about 37 centimeters; but since the rate is still climbing, that figure is a floor rather than a forecast. By 2023 global mean sea level stood 101.4 mm above its 1993 level, the highest annual value in the satellite record.

Note the units. Sea-level rise is quoted in millimeters per year because it is genuinely slow at human timescales, and that slowness is itself the problem: it is fast enough to inundate infrastructure built to last a century, and slow enough that no single year forces a response.

Key idea: The rate rose from about 1.3 mm per year in 1901-1971 to about 3.7 mm per year in 2006-2018, so sea-level rise is accelerating rather than merely continuing.

Who contributes what

Because the causes are physically distinct, researchers keep a budget for sea level much as they do for carbon, and the terms must sum to the observed total. For the recent satellite era the contributions break down roughly as follows.

• Thermal expansion of the warming ocean supplies on the order of a third to a half of the total. It dominated during the twentieth century and remains large.
• Mountain glaciers outside the great ice sheets contribute a comparable share despite holding far less ice, because they are small, warm, and melting fast.
• The Greenland ice sheet has become a major contributor, losing roughly 264 gigatonnes a year since 2002.
• The Antarctic ice sheet contributes less at present, around 135 gigatonnes a year, but holds by far the largest long-term potential.
• Changes in land water storage, chiefly groundwater pumped up and eventually delivered to the sea, add a small positive term.

The trend across these terms is the important part. Early in the record, expansion of warming water dominated. Increasingly, the ice sheet terms are growing fastest, which matters because an ice sheet, unlike thermal expansion, can in principle respond nonlinearly.

Key idea: Thermal expansion, glaciers, Greenland, Antarctica, and land water storage each contribute, and the ice sheet terms are the ones growing fastest.

A crucial distinction: sea ice versus land ice

Here is the point that trips people up. Melting sea ice, the floating ice of the Arctic pack, does not raise sea level, because floating ice already displaces its own weight in water. This is the same reason a melting ice cube does not overflow a full glass: the cube was already pushing aside exactly the volume of water it will become.

Melting land ice, glaciers and ice sheets resting on rock, is different. That ice was sitting on land, held out of the ocean, so when it melts it adds brand-new water to the sea and does raise the level. This is why the Greenland and Antarctic ice sheets, which hold enough water to raise seas by many meters over long times, are watched so closely. Sea ice loss still matters for climate, though, through the ice-albedo feedback, it just does not directly change sea level.

Key idea: Floating sea ice already displaces its weight, so its melt does not raise sea level; land ice adds new water and does.

The ice-cube test, worked through

Fill a glass to the brim with water and floating ice cubes. When the cubes melt, the glass does not overflow, because each floating cube was already displacing a volume of water equal to the meltwater it produces. Now imagine instead balancing an ice cube on the rim of the glass, out of the water. When that one melts, its water drips in and the glass overflows. The floating cube is Arctic sea ice; the rim cube is land ice. Same experiment, opposite result, and it is the whole reason land ice dominates sea-level rise.

Consequences for coasts

Rising seas do their damage gradually and in surges. A higher baseline means ordinary storms push water further inland, so storm surge flooding that once was rare becomes routine, the flood you used to see once a century arrives every few years. Salt water intrudes into freshwater aquifers and farmland. Low-lying cities, river deltas, and small island nations are most exposed, and hundreds of millions of people live within a meter or two of high tide.

Even if emissions stopped today, the ocean and ice respond slowly, so some further rise is already locked in. That is a key reason adaptation is unavoidable alongside cutting emissions.

Key idea: A higher baseline turns rare coastal floods into frequent ones, and because the response is slow, some further rise is already committed.

Projections, and where the uncertainty really lies

The IPCC's assessed likely ranges for global mean sea-level rise by 2100, relative to 1995 to 2014, run from about 0.28 to 0.55 meters under the very low emissions pathway to about 0.63 to 1.01 meters under the very high one. Those are the ranges the report considers likely, meaning at least a two-in-three chance.

The report then adds an unusual and important caveat. Because the physics of rapid ice sheet collapse, particularly the possible failure of tall ice cliffs in West Antarctica once their buttressing shelves are gone, is poorly understood, higher outcomes approaching 2 meters by 2100 cannot be ruled out, though they are assessed with low confidence. This is a textbook case of deep uncertainty: not a probability that can be quoted, but a recognised gap in physical understanding.

Beyond 2100 the picture is clearer in direction and starker in magnitude. Sea level will continue rising for centuries to millennia regardless of what happens to emissions, because the ocean and ice sheets are still adjusting to warming that has already occurred. Over two thousand years, the committed rise is on the order of 2 to 3 meters if warming is limited to 1.5 degrees Celsius, and considerably more at higher warming. Coastal planning therefore has to deal in centuries, not decades.

Key idea: Likely rise by 2100 spans roughly 0.28 to 1.01 meters depending on emissions, higher outcomes near 2 meters cannot be excluded with low confidence, and rise continues for millennia regardless.

Sea level is not the same everywhere

Global mean sea level is a useful average and a poor guide to any particular shoreline, because at least three effects make local change differ from the global figure.

First, land moves. Parts of Scandinavia and Canada are still rising as the crust rebounds from the weight of vanished ice-age glaciers, so relative sea level there can even fall. Conversely, river deltas compact under their own sediment and subside further when groundwater or oil is extracted; parts of Jakarta, the Mississippi delta, and coastal Bangladesh are sinking several times faster than the sea is rising.

Second, and counterintuitively, an ice sheet exerts gravitational attraction on the ocean around it. When Greenland loses mass, that attraction weakens and sea level within roughly 2,000 kilometers of Greenland actually falls, while the displaced water piles up in distant oceans. Melting in Greenland therefore raises sea level most in the southern hemisphere, and Antarctic melting raises it most in the north. These patterns are called sea-level fingerprints, and they let researchers infer which ice sheet is responsible from the geographic pattern of rise.

Third, ocean circulation and wind patterns pile water unevenly. A weakening AMOC, for instance, is expected to raise sea level along the northeastern coast of North America above the global average.

Key idea: Local sea-level change differs from the global mean because land rises or subsides, ice sheets exert gravitational fingerprints, and currents redistribute water.

Why sea ice still matters

Sea ice does not raise sea level when it melts, but dismissing it would be a mistake, because it does something arguably more consequential. Ice reflects most of the sunlight striking it while open water absorbs most of it, so losing summer sea ice converts a bright surface into a dark one and adds energy to the system.

This ice-albedo feedback is a principal reason the Arctic has warmed several times faster than the global average, a phenomenon called Arctic amplification. Warmer Arctic conditions in turn thaw permafrost, releasing carbon, and reduce the temperature contrast that drives the polar jet stream. Sea ice loss is therefore a climate story rather than a sea-level story, and September Arctic extent is shrinking at about 12.2 percent per decade.

Key idea: Melting sea ice changes no sea level but replaces a bright surface with a dark one, driving Arctic amplification through the ice-albedo feedback.

Common misconceptions

• "Melting Arctic sea ice will flood the coasts." No. Floating ice already displaces its weight, so its melt does not raise sea level; land ice does.
• "Sea-level rise is only about melting ice." No. Thermal expansion of warming water is a comparably large cause.
• "If we stop emitting, seas stop rising immediately." No. The oceans and ice respond slowly, so some further rise is already locked in.
• "Sea level rises by the same amount everywhere." No. Land subsidence or rebound, gravitational fingerprints from shrinking ice sheets, and ocean currents make local change differ substantially from the global mean.
• "Sea ice loss is harmless because it does not raise sea level." No. It replaces a bright reflecting surface with dark water, amplifying Arctic warming.

Recap

• Sea level has risen roughly 21 to 24 cm since 1880 and the rate is accelerating.
• The rate rose from about 1.3 mm per year in 1901-1971 to about 3.7 mm per year in 2006-2018.
• The two main causes are thermal expansion and melting land ice, with glaciers, Greenland, and Antarctica each contributing.
• Floating sea ice melt does not raise sea level; land ice melt does.
• Likely rise by 2100 spans about 0.28 to 1.01 meters by scenario, with higher outcomes not excluded at low confidence.
• Local sea level differs from the global mean because of subsidence, gravitational fingerprints, and currents.
• Higher baselines worsen storm surge, and rise continues for centuries to millennia after emissions stop.

Sources

1. Fox-Kemper, B., Hewitt, H. T., Xiao, C., et al. (2021). Chapter 9: Ocean, cryosphere and sea level change. In IPCC AR6 WGI (rates by period; projections; deep uncertainty). ipcc.ch
2. NOAA Climate.gov ↗. (n.d.). Climate change: Global sea level (21-24 cm since 1880; 101.4 mm above 1993 in 2023). climate.gov
3. NASA. (n.d.). Sea Level Change Portal: Overview of global sea level. sealevel.nasa.gov
4. NOAA National Ocean Service. (n.d.). Is sea level rising? oceanservice.noaa.gov
5. Edwards, T. L., Nowicki, S., Marzeion, B., et al. (2021). Projected land ice contributions to twenty-first-century sea level rise. Nature, 593, 74-82. nature.com
6. NASA Science. (n.d.). Earth indicators: Ice sheets (Greenland about 264 Gt/yr, Antarctica about 135 Gt/yr since 2002). climate.nasa.gov
7. National Snow and Ice Data Center. (n.d.). Ice sheets. nsidc.org

In late June 2021 the town of Lytton, in British Columbia, recorded 49.6 degrees Celsius, a temperature never before measured anywhere in Canada and hotter than any reading ever taken in Las Vegas. The next day the town burned down. Within a fortnight, an international team had published an analysis concluding the heatwave would have been virtually impossible without human-caused warming. That turnaround, from event to quantified attribution in days, is one of the more remarkable developments in modern climate science.

The big picture

Climate change does not usually invent brand-new kinds of weather; it loads the dice, shifting the odds and intensity of extremes we already know. A warmer world rolls sixes more often, meaning more record heat, heavier downpours, and deeper droughts.

The ocean, meanwhile, faces a second CO₂ problem that has nothing to do with temperature: acidification. Together, heat and changing chemistry put coral reefs, the rainforests of the sea, under a double stress.

Loading the dice on extremes

Because the whole distribution of temperatures shifts warmer, record-breaking heat waves become far more frequent and severe. The rest follows from basic physics:

• Warmer air holds more moisture, so when it does rain it can rain harder, raising the risk of heavy downpours and floods.
• The same warmth increases evaporation, drying soils faster, which can deepen droughts and lengthen fire seasons between rains.
• Tropical cyclones are not necessarily more numerous, but a warmer ocean provides more energy, so the strongest storms tend to intensify and drop more rain.

Key idea: Warming shifts the odds and intensity of existing extremes, making heat waves, heavy rain, and drought more likely rather than creating new weather types.

How loading the dice works

Picture the bell curve of daily temperatures. Shift the whole curve a couple of degrees warmer and two things happen at the hot end: days that used to be rare records become common, and entirely new record highs, off the old chart, start to appear. Cold days still occur, they have just become rarer. This is why warming can be real even though you still get cold snaps, and why the biggest changes show up in the extremes, not just the average.

Quantify the shift and the effect becomes vivid. Take a temperature distribution with a standard deviation of about 1 degree Celsius and define an extreme as anything more than three standard deviations above the mean, which occurs about once in 700 days. Now shift the whole curve warmer by 1 degree. The old three-sigma threshold is now only two sigma away, and events beyond it occur roughly once in 44 days. A one-degree shift in the average turned a once-in-two-years event into a monthly one. Extremes respond far more dramatically than averages, which is why a seemingly modest global change produces disproportionate damage.

Key idea: A small shift in the average temperature causes a large increase in record hot days, because the whole distribution moves.

What attribution science can and cannot claim

Since the mid-2000s, researchers have developed formal methods for assessing the influence of climate change on individual events. The standard approach compares two large ensembles of simulations, one representing the world as it is and one representing a counterfactual world without human emissions, and reports how the probability or intensity of the event differs between them.

Results are reported as a fraction of attributable risk or as a change in odds. A statement like "this heatwave was made at least ten times more likely" is a claim about probability distributions, verifiable against the ensembles, and quite different from the unscientific claim that climate change caused the event.

Confidence varies systematically by event type, and the IPCC assessed the following.

• Heat extremes: it is virtually certain that hot extremes have become more frequent and more intense across most land regions since the 1950s, and human influence is the main driver. Attribution confidence here is high.
• Heavy precipitation: the frequency and intensity of heavy rainfall have likely increased over most land regions with sufficient data, with human influence the likely main driver. Confidence is moderate to high.
• Agricultural and ecological drought: increases are detected in some regions, driven substantially by greater evaporative demand in warmer air. Confidence is regional rather than global.
• Tropical cyclones: it is likely that the proportion of intense category 3 to 5 storms has increased over the past four decades, while the total number is not clearly trending. Attribution for any single storm remains harder.

This graded structure is worth internalising. Someone who claims every disaster proves climate change is overstating the evidence; someone who claims attribution is impossible is decades out of date. The truthful position is that the science reports changed odds, with well-documented confidence that is high for heat, moderate for rainfall, and lower for storms and droughts.

Key idea: Attribution compares the real world with simulations of a world without emissions and reports changed odds, with high confidence for heat extremes and lower confidence for cyclones and droughts.

Ocean acidification: the other CO₂ problem

The ocean absorbs roughly a quarter of the CO₂ we emit, which slows warming but changes seawater chemistry. Dissolved CO₂ reacts with water to form carbonic acid, lowering the ocean's pH (the scale of acidity, where lower means more acidic) in a process called ocean acidification. Surface waters are already measurably more acidic than in pre-industrial times.

More acidic water makes it harder for corals, oysters, and many plankton to build their calcium carbonate shells and skeletons, the way acid slowly eats at chalk. Because these shell-builders sit at the base of marine food webs, the effect ripples upward. This is a chemistry problem driven by CO₂ itself, separate from and additional to warming.

Key idea: Absorbed CO₂ forms carbonic acid and lowers ocean pH, making it harder for shell-building life to form calcium carbonate, a problem distinct from warming.

Putting a number on acidification

Average surface ocean pH has fallen from roughly 8.2 before industrialisation to about 8.1 today, a change that sounds trivial until you remember that pH is a logarithmic scale. Each whole unit represents a tenfold change in hydrogen ion concentration.

Work it out. A drop of 0.1 pH units corresponds to multiplying the hydrogen ion concentration by 10 raised to the power 0.1, which equals about 1.26. A drop of 0.11 units gives about 1.29. In other words, surface seawater now holds roughly 26 to 30 percent more hydrogen ions than it did before industrialisation. Describing that as a small change would be a serious misreading of the arithmetic.

The chemistry then propagates. Added hydrogen ions combine with carbonate ions to form bicarbonate, which reduces the carbonate available for shell building. Researchers track this through the saturation state of aragonite, the particular form of calcium carbonate that corals and many molluscs use. Where saturation falls below one, existing shells begin to dissolve. Cold water holds more dissolved CO₂, so polar and deep waters are crossing that threshold first, and shell-forming plankton in the Southern Ocean are among the earliest organisms affected.

Key idea: Ocean pH has fallen about 0.1 units, which on a logarithmic scale means roughly 26 to 30 percent more hydrogen ions, reducing the carbonate available for shells.

Coral reefs under a double stress

Corals feel both blows at once. When water gets too warm, corals expel the colorful algae that feed them and turn ghostly white in a coral bleaching event; prolonged bleaching kills them. A marine heat wave, a spell of unusually warm ocean water, is the usual trigger. At the same time, acidification weakens their skeletons from the chemistry side. Reefs shelter about a quarter of all marine species and protect coastlines from waves, so their loss is felt widely, and widespread bleaching events have already struck reefs worldwide during hot years.

The mechanism is worth stating precisely, because it explains why the damage tracks temperature so tightly. Reef-building corals live in partnership with single-celled algae that live inside their tissues and supply most of their food through photosynthesis. When water stays roughly 1 degree Celsius above the usual summer maximum for several weeks, that partnership breaks down and the coral expels its algae. NOAA quantifies the stress in degree heating weeks, the accumulated excess temperature over a rolling twelve-week window: about 4 degree heating weeks typically triggers significant bleaching, and 8 or more brings widespread mortality.

NOAA confirmed in April 2024 that the world had entered its fourth recorded global bleaching event, with heat stress severe enough to cause bleaching documented across reefs in every major ocean basin. Recovery is possible if temperatures fall and the interval before the next event is long enough, but reef-building corals grow slowly, and the intervals between mass bleaching events have been shortening.

Key idea: Corals suffer a double stress, bleaching from marine heat waves and skeleton weakening from acidification, threatening ecosystems that shelter a quarter of marine species.

Common misconceptions

• "Climate change creates entirely new kinds of weather." Mostly no. It shifts the odds and strength of extremes that already exist.
• "Ocean acidification is just warming by another name." No. Acidification is a chemistry change from dissolved CO₂, separate from and additional to warming.
• "Coral bleaching means the coral is already dead." Not immediately. Bleached coral has expelled its algae and is stressed; it can recover if conditions improve, but prolonged bleaching kills it.
• "A 0.1 drop in pH is a negligible change." No. pH is logarithmic, so a 0.1 unit fall means roughly 26 percent more hydrogen ions.
• "Scientists cannot say anything about a specific storm or heatwave." No. Attribution studies can quantify how much more likely or intense an event became, with confidence that is high for heat and lower for cyclones.

Recap

• Warming loads the dice, intensifying heat waves, heavy rain, and drought.
• A small shift in the average sharply increases record hot days, because extremes sit in the tail of the distribution.
• Attribution reports changed odds, with virtual certainty for more frequent hot extremes and lower confidence for individual storms.
• Absorbed CO₂ forms carbonic acid, lowering ocean pH by about 0.1 units, or 26 to 30 percent more hydrogen ions.
• Acidification reduces carbonate saturation and hinders calcium carbonate shells and skeletons.
• Bleaching is measured in degree heating weeks, with about 4 triggering bleaching and 8 or more causing mortality.
• Corals face a double stress: bleaching from heat and weaker skeletons from acid, and a fourth global bleaching event was confirmed in 2024.

Sources

1. Seneviratne, S. I., Zhang, X., Adnan, M., et al. (2021). Chapter 11: Weather and climate extreme events in a changing climate. In IPCC AR6 WGI. ipcc.ch
2. NOAA National Ocean Service. (n.d.). What is ocean acidification? oceanservice.noaa.gov
3. NOAA Pacific Marine Environmental Laboratory. (n.d.). Ocean acidification. pmel.noaa.gov
4. NOAA. (2024). NOAA confirms 4th global coral bleaching event. noaa.gov
5. NOAA Coral Reef Watch. (n.d.). Satellite coral bleaching heat stress monitoring (degree heating weeks and bleaching alert areas). coralreefwatch.noaa.gov
6. NOAA National Ocean Service. (n.d.). What is coral bleaching? oceanservice.noaa.gov
7. Fox-Kemper, B., Hewitt, H. T., Xiao, C., et al. (2021). Chapter 9: Ocean, cryosphere and sea level change. In IPCC AR6 WGI (marine heatwaves and ocean chemistry). ipcc.ch

In 1914 a passenger pigeon named Martha died in the Cincinnati Zoo. A century earlier her species had been the most abundant bird in North America, moving in flocks that darkened the sky for hours and were estimated in the billions. Abundance turned out to be no protection at all. The passenger pigeon is a reminder that ecological collapse can happen to species that seem inexhaustible, and that it can happen fast.

The big picture

Biodiversity is the variety of life, and it is not a luxury. Functioning ecosystems quietly provide the pollination, clean water, fertile soil, and flood control that human life depends on. Losing biodiversity is like pulling rivets from an airplane: a few may not matter, but keep pulling and the structure fails.

Species are now vanishing far faster than the natural rate. This lesson defines biodiversity, lists the drivers of its decline, and shows how climate change acts as a threat multiplier on top of the others.

What biodiversity is, and why it matters

Biodiversity spans three levels: the genetic diversity within a species, the number and variety of species, and the diversity of whole ecosystems. All three matter for resilience.

Healthy ecosystems deliver ecosystem services, the benefits people obtain from nature, including pollination of crops, purification of water, formation of fertile soil, regulation of climate and floods, timber and fisheries, and cultural and recreational value. More diverse ecosystems tend to be more productive and better able to absorb shocks, just as a varied investment portfolio weathers a downturn better than a single stock.

Key idea: Biodiversity is variety at the genetic, species, and ecosystem levels, and it underpins the ecosystem services that human life depends on.

The four kinds of ecosystem service

Economists and ecologists conventionally sort ecosystem services into four categories, and the classification is useful because it exposes which benefits markets price and which they ignore.

• Provisioning services are the tangible products: food, fresh water, timber, fibre, and medicines. These generally have market prices, so they are visible in economic accounts.
• Regulating services include pollination, water purification, flood control, pest regulation, and climate regulation through carbon storage. These rarely have prices, which is precisely why they are often destroyed inadvertently.
• Supporting services are the underlying processes that make the others possible: soil formation, nutrient cycling, and primary production.
• Cultural services cover recreation, aesthetic and spiritual value, and the sense of place that landscapes provide.

Pollination illustrates the accounting problem. Roughly three quarters of the world's leading food crops depend to some degree on animal pollination, and global assessments value that service in the hundreds of billions of dollars a year. Yet no farmer pays a wild bee. When an unpriced service disappears, the loss shows up later as a cost that someone must cover, whether by hand-pollinating orchards or by importing food.

Key idea: Services divide into provisioning, regulating, supporting, and cultural, and the unpriced regulating services like pollination are the ones most easily destroyed by accident.

Why biodiversity is declining

Species are going extinct far faster than the natural background rate, so much so that scientists speak of a possible sixth mass extinction. The dominant drivers, roughly in order of impact so far, are:

Driver	Example
Habitat loss	Clearing rainforest for agriculture
Overexploitation	Overfishing collapsing fish stocks
Invasive species	Introduced predators wiping out island birds
Pollution	Pesticides and nutrient runoff harming wildlife
Climate change	Warming pushing species beyond their ranges

Habitat loss, converting forests and wetlands to farms and cities, has been the leading cause. Overexploitation (overfishing, hunting, logging), invasive species carried by human transport, pollution, and increasingly climate change follow. These pressures often combine and reinforce one another rather than acting alone.

Key idea: Habitat loss has been the leading driver of biodiversity loss, with overexploitation, invasive species, pollution, and climate change adding to it.

How fast, exactly?

Claims about extinction rates need a baseline, and paleontologists supply one. The background extinction rate, inferred from the fossil record, is on the order of 0.1 to 1 extinctions per million species per year. Modern rates estimated from documented extinctions are tens to hundreds of times higher, and the IPBES Global Assessment concluded that the current rate is at least tens to hundreds of times above the average of the past 10 million years and is accelerating.

That assessment also produced the figure most often quoted: around 1 million animal and plant species are threatened with extinction, many within decades. It reached that number by extrapolating from well-studied groups to less-studied ones, which is why it is an estimate with genuine uncertainty rather than a count.

The direct count comes from the IUCN Red List, which has now assessed more than 175,900 species and classifies over 49,500 of them as threatened. Note what that comparison tells you: the Red List has assessed only a small fraction of the roughly 2 million described species, and a far smaller fraction of the estimated total, so the count is a floor rather than a total. Insects, fungi, and marine invertebrates are especially under-assessed.

Key idea: Extinctions are running tens to hundreds of times above the fossil background rate, with over 49,500 of the 175,900 species assessed by the IUCN classed as threatened and roughly a million estimated to be at risk overall.

Climate as a threat multiplier

Climate change acts as a threat multiplier, a factor that intensifies other existing pressures rather than acting alone. As temperatures shift, species must move poleward or upslope to track suitable conditions, but many cannot move fast enough or are blocked by cities and farmland.

Timing can also break down. A phenological mismatch happens when interacting species fall out of sync, for example a flower blooming before its pollinator emerges, so neither gets what it needs. Mountain and polar species can run out of cooler habitat entirely, with nowhere higher or farther to go. Layered on top of habitat loss and pollution, climate change can push already stressed populations past the brink. Protecting biodiversity therefore means both cutting emissions and conserving and connecting habitat so species have somewhere to move.

Key idea: Climate change multiplies other threats by forcing species to move, breaking the timing between them, and erasing cool refuges, pushing stressed populations over the edge.

Extinction risk as a function of warming

The IPCC's Working Group II assessed how extinction risk scales with global temperature, and the results are worth quoting because they show both the trend and the honest width of the estimates. Of the terrestrial species studied, the percentage facing very high extinction risk rises from roughly 3 to 14 percent at 1.5 degrees Celsius of warming, to roughly 3 to 29 percent at 3 degrees, and to roughly 3 to 48 percent at 5 degrees.

Two features of those ranges repay attention. The lower bound stays near 3 percent across all warming levels, reflecting studies with conservative assumptions about species' ability to adapt and move. The upper bound climbs steeply, reflecting studies that assume limited dispersal. The width of the range is a genuine expression of what is not known, and the trend across warming levels is unambiguous even so.

Certain systems have low thresholds. The report assesses that warm-water coral reefs would decline by 70 to 90 percent at 1.5 degrees Celsius of warming and by over 99 percent at 2 degrees. Polar and mountain-top ecosystems face similar constraints, because their species simply run out of colder places to go.

Key idea: Very high extinction risk rises from roughly 3-14 percent of studied species at 1.5 degrees C to 3-48 percent at 5 degrees, and warm-water coral reefs decline 70-90 percent even at 1.5.

What conservation can actually do

The response has to address the drivers in proportion to their size, which means habitat is the first priority and climate the fastest-growing one. Several approaches have measurable track records.

• Protecting and connecting habitat. Protected areas work, but isolated reserves become traps when the climate shifts beneath them. Corridors that link reserves let populations move, which is why connectivity has become as important a design goal as area.
• Restoration. Rebuilding wetlands, forests, and reefs can recover both biodiversity and carbon storage at once, though a restored system rarely matches an intact one.
• Managing exploitation. Fisheries with enforced catch limits have recovered repeatedly, demonstrating that overexploitation is among the most reversible drivers.
• Controlling invasive species. Island eradication programmes have produced some of conservation's clearest successes, since islands have defined boundaries.
• Cutting emissions. Because climate change erases the refuges that all other measures depend on, mitigation is itself a conservation strategy.

The honest summary is that biodiversity loss is not one problem but five interacting ones, and that fixing habitat while ignoring climate, or the reverse, addresses only part of the pressure.

Key idea: Effective conservation combines protected and connected habitat, restoration, managed harvests, and invasive control, and none of it holds up unless emissions also fall.

Common misconceptions

• "Biodiversity just means the number of species." No. It also includes genetic variety within species and the diversity of whole ecosystems.
• "Climate change is the single biggest driver of extinctions so far." Not yet. Habitat loss has been the leading driver; climate change is a fast-growing additional pressure.
• "Species can simply move to keep up with warming." Often they cannot, blocked by cities and farmland or lacking cooler habitat to move into.
• "An abundant species is safe." No. The passenger pigeon went from billions to zero in about fifty years, because abundance does not protect against systematic pressure.
• "The IUCN threatened count is the number of species at risk." No. It is a floor, since only about 175,900 of roughly 2 million described species have been assessed, and insects and fungi are barely covered.

Recap

• Biodiversity is variety at the genetic, species, and ecosystem levels.
• Services fall into provisioning, regulating, supporting, and cultural categories, and the unpriced regulating ones are most easily lost.
• Extinctions are running tens to hundreds of times above the fossil background rate.
• Over 49,500 of the 175,900 species assessed by the IUCN are threatened, and IPBES estimates roughly a million at risk overall.
• Habitat loss is the leading driver, with overexploitation, invasive species, pollution, and climate change adding further pressure.
• Very high extinction risk rises from about 3-14 percent of studied species at 1.5 degrees C to 3-48 percent at 5 degrees.
• Climate change is a threat multiplier, so conservation requires protected connected habitat and falling emissions together.

Sources

1. IPBES. (2019). Global assessment report on biodiversity and ecosystem services. Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. ipbes.net
2. IUCN. (2026). The IUCN Red List of Threatened Species (over 175,900 species assessed; over 49,500 threatened). iucnredlist.org
3. IUCN. (n.d.). Background and history of the Red List. iucnredlist.org
4. Parmesan, C., Morecroft, M. D., Trisurat, Y., et al. (2022). Chapter 2: Terrestrial and freshwater ecosystems and their services. In IPCC AR6 WGII. ipcc.ch
5. IPCC. (2022). Summary for policymakers. In Climate change 2022: Impacts, adaptation and vulnerability (AR6 WGII; extinction risk by warming level; coral reef losses). ipcc.ch
6. Bar-On, Y. M., Phillips, R., & Milo, R. (2018). The biomass distribution on Earth. PNAS, 115(25), 6506-6511. pmc.ncbi.nlm.nih.gov
7. NOAA. (n.d.). Climate change impacts. NOAA Education. noaa.gov

For five days in December 1952 a windless fog settled over London and mixed with the smoke of a million coal fires. Visibility fell to a few meters, buses were led by men walking ahead with lanterns, and roughly 4,000 people died within the week, with thousands more in the months that followed. The Great Smog produced Britain's first Clean Air Act four years later, and it established a pattern that recurs throughout this lesson: air pollution is a solvable problem, and solving it saves an extraordinary number of lives.

The big picture

Pollution is the release of harmful substances into the environment. It overlaps with the greenhouse issue but is not the same thing, and keeping the two straight matters. Many air pollutants harm your lungs directly, whereas CO₂ warms the planet without being toxic to breathe.

This lesson separates three often-confused topics: everyday air pollution, the ozone hole (a solved problem, and not the same as global warming), and water pollution including plastics.

Air pollution versus the greenhouse effect

Air pollution includes fine particulate matter, tiny airborne specks from burning fuel that lodge in the lungs, and gases like sulfur and nitrogen oxides. These harm human lungs and hearts and cause millions of premature deaths a year. Nitrogen oxides and volatile compounds also cook in sunlight to form ground-level ozone, or smog. Sulfur and nitrogen emissions produce acid rain, which damages forests, lakes, and buildings.

Here is the key distinction. Greenhouse gases like CO₂ warm the planet but are not directly toxic to breathe, whereas classic air pollutants harm health directly. They often share a source, combustion, so cutting fossil-fuel use yields a double benefit: less warming and cleaner air.

Key idea: CO₂ warms the climate without being a breathing hazard, while particulates and smog harm health directly; both often come from burning fuel, so cutting it helps twice.

The health burden, in numbers

The World Health Organization estimated that ambient outdoor air pollution caused about 4.2 million premature deaths worldwide in 2019, and that ambient and household air pollution together are associated with roughly 6.7 million premature deaths annually. It also reported that in 2019, 99 percent of the world's population lived where WHO air quality guideline levels were not met, and that 89 percent of the outdoor-pollution deaths occurred in low- and middle-income countries.

Set those figures against other causes and the scale becomes clear. Air pollution kills on the order of the annual death toll from tobacco, and it does so without anyone choosing to be exposed. This is the strongest short-term argument for reducing fossil fuel combustion, and it does not depend on any claim about climate at all.

Key idea: WHO estimates about 4.2 million premature deaths a year from outdoor air pollution and 6.7 million from outdoor and household combined, with 99 percent of people breathing air above guideline levels.

Why particle size decides the harm

Particulate matter is classified by diameter, and the classification is a health classification rather than a chemical one. PM10 means particles under 10 micrometers across, and PM2.5 means those under 2.5 micrometers, roughly one thirtieth the width of a human hair.

The distinction matters because the nose and upper airway trap larger particles effectively. Particles below about 2.5 micrometers pass those defences and reach the alveoli, the tiny sacs where gas exchange occurs, and the smallest can cross into the bloodstream. This is why PM2.5 is associated not only with asthma and lung disease but with heart attacks, strokes, and low birth weight. The WHO guideline for annual average PM2.5 was tightened in 2021 to 5 micrograms per cubic meter, a level exceeded across most of the inhabited world.

Key idea: PM2.5 particles are small enough to reach the alveoli and enter the bloodstream, which is why they cause cardiovascular as well as respiratory disease.

The aerosol paradox

Here is an uncomfortable complication that honest treatment requires. The same sulfate particles that damage lungs also reflect sunlight and seed brighter clouds, and they have therefore been cooling the planet. The IPCC assesses aerosol effects as a substantial negative forcing that has offset part of the warming from greenhouse gases.

The consequence is that cleaning up air pollution removes that masking effect and reveals warming that greenhouse gases had already committed. This has been observed following sulfur limits on shipping fuel and the rapid cleanup of Chinese industry. It is not an argument against cleaning the air, since aerosols kill millions of people a year and last only days in the atmosphere while CO₂ lasts for centuries. It is, however, an argument for cutting greenhouse gases at the same time, because otherwise the cleanup unmasks warming without addressing its cause.

Key idea: Sulfate aerosols both harm health and cool the planet, so cleaning the air unmasks warming already committed by greenhouse gases, which strengthens rather than weakens the case for cutting emissions too.

The ozone layer: a solved problem

A separate atmospheric issue is the ozone hole, and it is frequently confused with global warming. Human-made chemicals called chlorofluorocarbons (CFCs), once used in refrigerants and spray cans, drifted up to the stratosphere and destroyed protective ozone, thinning it dramatically over Antarctica and letting more harmful ultraviolet reach the surface.

This is distinct from climate change: the ozone hole is about UV-blocking ozone destroyed by CFCs, not about heat-trapping greenhouse gases. It is also a rare success story. The 1987 Montreal Protocol phased out CFCs worldwide, and the ozone layer is now slowly healing, proof that coordinated global action on an atmospheric problem can work.

The chemistry is worth knowing, because it explains why the damage concentrates over Antarctica. A chlorine atom released from a CFC molecule destroys an ozone molecule and is regenerated, so a single chlorine atom can destroy many thousands of ozone molecules before it is eventually removed. The reaction runs fastest on the surfaces of the ice crystals that form in polar stratospheric clouds, which require the extreme cold of the Antarctic winter vortex. That is why a globally distributed pollutant produces a hole at one pole in one season.

The recovery timetable is now reasonably well characterised. With nearly all controlled substances phased out, the scientific assessments project that stratospheric ozone will return to 1980 values around the middle of this century over most of the globe, later over the Arctic, and around the 2060s over Antarctica. The lag is long because CFC molecules already in the atmosphere persist for decades, a familiar theme from the residence-time discussion in Lesson 1.

Key idea: The ozone hole (from CFCs, blocking UV) is a different problem from greenhouse warming, and the Montreal Protocol is fixing it, with recovery to 1980 values projected around mid-century globally and the 2060s over Antarctica.

Water pollution and plastics

Water pollution comes from two kinds of sources. A point source is a single identifiable outlet, like a factory pipe. A nonpoint source is diffuse and spread out, like farm and street runoff, which is harder to control precisely because it has no single outlet to plug.

Major categories include nutrients (causing the eutrophication and dead zones met earlier), pathogens from sewage, toxic metals and chemicals, and oil spills. A fast-growing concern is plastic pollution: durable plastics accumulate in rivers and oceans, break into tiny fragments called microplastics, and enter food webs. Because pollution crosses borders through air and water, it, like climate change, often requires cooperation to solve.

Key idea: Water pollution comes from point sources (a single pipe) and harder-to-control nonpoint runoff, and plastics that fragment into microplastics are a growing worry.

The scale and fate of plastic

Global plastic production now runs to several hundred million tonnes a year and has grown faster than almost any other material in history. Only a small fraction is recycled; most is landfilled, incinerated, or mismanaged, and millions of tonnes reach rivers and the ocean annually.

What happens next depends on a property that made plastic useful in the first place: it does not readily break down. Sunlight and wave action fragment it mechanically into ever smaller pieces without dismantling the polymer, producing microplastics under 5 millimeters and eventually nanoplastics. These have been found in Arctic sea ice, in deep-ocean sediment, in drinking water, and in human tissue. Research into the health consequences is active and not yet settled, which is itself worth stating plainly rather than overstating in either direction.

Plastic pollution also connects back to climate, since almost all plastic is made from fossil feedstocks and its production and incineration emit CO₂. As with the other topics in this lesson, the apparently separate problems share a root cause in how the world uses fossil carbon.

Key idea: Plastic fragments rather than degrading, spreading microplastics through ecosystems and human tissue, and because it is made from fossil feedstocks it links directly back to the carbon problem.

Common misconceptions

• "The ozone hole causes global warming." No. The ozone hole (from CFCs) and greenhouse warming are separate problems with different causes.
• "CO₂ is a poison you can smell." No. CO₂ drives warming but is not a direct respiratory toxin at atmospheric levels; particulates and smog are the health hazards.
• "Nothing can be done about a global atmospheric problem." The Montreal Protocol shows otherwise: coordinated action healed the ozone layer.
• "Cleaning up air pollution will help slow warming." Partly the reverse. Sulfate aerosols currently cool the planet, so removing them unmasks warming, which is why greenhouse gases must be cut alongside.
• "All airborne particles are equally harmful." No. Particles below 2.5 micrometers penetrate deepest and are linked to heart disease and stroke as well as lung damage.

Recap

• CO₂ warms the planet but is not directly toxic; particulates and smog harm health.
• WHO attributes about 4.2 million premature deaths a year to outdoor air pollution, with 99 percent of people breathing air above guideline levels.
• PM2.5 particles reach the alveoli and the bloodstream, causing cardiovascular as well as respiratory harm.
• Combustion often produces both warming gases and health-damaging pollutants, so cutting fuel use helps twice.
• Sulfate aerosols also cool the planet, so cleaning the air unmasks committed warming.
• The ozone hole (from CFCs) is distinct from greenhouse warming; recovery to 1980 values is projected around mid-century, and the 2060s over Antarctica.
• Water pollution comes from point and nonpoint sources; plastics fragment into microplastics.

Sources

1. World Health Organization. (2024). Ambient (outdoor) air pollution (fact sheet; 4.2 million deaths in 2019; 99 percent above guideline levels). who.int
2. U.S. Environmental Protection Agency. (n.d.). Particulate matter (PM) basics. epa.gov
3. NASA. (n.d.). NASA Ozone Watch. Goddard Space Flight Center. ozonewatch.gsfc.nasa.gov
4. World Meteorological Organization & UN Environment Programme. (2022). Scientific assessment of ozone depletion: 2022. NOAA Chemical Sciences Laboratory. csl.noaa.gov
5. UN Environment Programme. (n.d.). About the Montreal Protocol. OzonAction. unep.org
6. U.S. Environmental Protection Agency. (n.d.). Basic information about nonpoint source (NPS) pollution. epa.gov
7. Szopa, S., Naik, V., Adhikary, B., et al. (2021). Chapter 6: Short-lived climate forcers. In IPCC AR6 WGI (aerosol forcing and air quality links). ipcc.ch

A single barrel of oil contains roughly the energy a fit human labourer could produce in several years of hard work. That extraordinary concentration is why fossil fuels reshaped civilisation within a few generations, and it is also why replacing them is difficult. Any honest account of energy has to respect both facts at once: what these fuels did for us, and what continuing to burn them costs.

The big picture

Energy is where climate policy meets daily life, because burning fossil fuels supplies most of the world's energy and produces most of its CO₂. Solving climate change is, at its core, an energy problem: how to keep the lights on while emitting almost no carbon.

This lesson compares the main energy sources by two questions, do they emit carbon and can they be replenished, then tackles the practical challenge that the cheapest clean sources, solar and wind, do not run on demand.

Fossil fuels

Fossil fuels, coal, oil, and natural gas, are energy-dense, storable, and were cheap, which is why they built the modern world. But they are nonrenewable, meaning finite on human timescales, and their combustion is the primary cause of climate change and much air pollution. Among them, coal emits the most CO₂ per unit of energy and natural gas the least, though gas is still a major emitter.

Quantify the difference between them, because it is often assumed rather than checked. Per unit of heat released, the U.S. Energy Information Administration reports roughly 206 pounds of CO₂ per million British thermal units for bituminous coal, about 161 for diesel fuel, and about 117 for natural gas. Natural gas therefore emits a little under 60 percent as much CO₂ as coal for the same heat, which is a real improvement and nowhere near zero.

That comparison also comes with a caveat. Natural gas is mostly methane, and methane leaking unburned from wells, pipelines, and compressors is itself a powerful greenhouse gas. Leakage rates of even a few percent erode much of the advantage gas holds over coal on a twenty-year timescale, which is why measuring and reducing methane leaks has become a policy priority in its own right.

Key idea: Fossil fuels are energy-dense but nonrenewable, and burning them is the main source of CO₂, with coal the most carbon-intensive and gas about 60 percent of coal before accounting for methane leaks.

Renewable and low-carbon energy

Renewable sources draw on flows nature continually replenishes and emit little or no CO₂ in operation. It helps to picture the difference: a fossil fuel is like a savings account you can only spend down, while a renewable is like income that keeps arriving.

• Solar photovoltaics convert sunlight directly to electricity and have fallen dramatically in cost.
• Wind turbines convert moving air and are now among the cheapest sources of new electricity.
• Hydropower uses flowing water and is a mature, dispatchable source, though it depends on rivers and dams affect ecosystems.
• Geothermal taps Earth's internal heat, and sustainable bioenergy uses plant matter.

Nuclear power is not renewable, since it uses uranium, but it is low-carbon: it produces large, steady electricity with minimal CO₂, at the cost of waste management and high upfront expense. Note the useful category of low-carbon energy, which includes both renewables and nuclear, anything that emits little CO₂ in operation.

Key idea: Renewables run on replenished flows and emit almost no operating CO₂; nuclear is not renewable but is low-carbon, and both count as low-carbon energy.

Source	Renewable?	Operating CO2	Note
Coal	No	Very high	Dispatchable but most polluting
Natural gas	No	High (lower than coal)	Flexible, still a major emitter
Solar	Yes	Near zero	Intermittent; cost has plunged
Wind	Yes	Near zero	Intermittent; very low cost
Hydro	Yes	Low	Dispatchable; alters rivers
Nuclear	No	Very low	Steady baseload; waste and cost

Comparing sources honestly: lifecycle emissions

Calling a source zero-carbon is a shorthand that hides the manufacturing, construction, fuel processing, and decommissioning behind it. The fair comparison is lifecycle emissions, measured in grams of CO₂-equivalent per kilowatt-hour of electricity delivered over the whole life of the plant. The IPCC's assessed median values are approximately as follows.

Source	Median lifecycle emissions (g CO2-eq per kWh)
Coal	about 820
Natural gas (combined cycle)	about 490
Solar photovoltaic (utility scale)	roughly 40 to 50
Hydropower	about 24
Nuclear	about 12
Wind (onshore)	about 11

Read the table carefully, because it settles several arguments. Solar panels do carry manufacturing emissions, but at roughly 40 to 50 grams per kilowatt-hour they emit around one twentieth of what coal does, not a comparable amount. Nuclear and wind are both around 12 grams. And the gap between the fossil and non-fossil groups is an order of magnitude, not a matter of a few percent.

Key idea: On a full lifecycle basis coal emits about 820 g CO₂-eq per kWh against roughly 11 to 50 for wind, nuclear, hydro, and solar, an order-of-magnitude gap.

Capacity factor: why nameplate power misleads

A power plant's advertised capacity is what it produces running flat out, which almost nothing does. The capacity factor is the fraction of that maximum actually achieved over a year, and it varies enormously by technology: roughly 15 to 25 percent for solar photovoltaics, 35 to 45 percent for onshore wind, and around 90 percent for nuclear.

Work an example. A 100 megawatt solar farm at a 20 percent capacity factor delivers 100 times 0.20 times 8,760 hours, which is about 175,000 megawatt-hours a year. A 100 megawatt nuclear reactor at 90 percent delivers 100 times 0.90 times 8,760, or about 788,000 megawatt-hours, roughly four and a half times as much from the same nameplate rating.

This is why headlines about installed capacity can mislead, and why comparisons should use energy delivered rather than capacity installed. It does not make solar uncompetitive; falling costs have more than compensated. It does mean that building a low-carbon grid requires considerably more nameplate capacity than the fossil fleet it replaces, plus the storage and transmission discussed next.

Key idea: Capacity factor converts nameplate power into actual energy delivered, and at roughly 20 percent for solar against 90 percent for nuclear the same rating produces very different output.

Why solar and wind became cheap

The most consequential energy development of the past two decades was not a scientific breakthrough but a manufacturing one. Technologies produced in factories tend to follow a learning curve: each doubling of cumulative production reduces unit cost by a roughly constant percentage. Solar modules have followed such a curve for decades, and their cost has fallen by roughly 90 percent since 2010, with batteries following a similar trajectory.

The mechanism matters for policy, because it means deployment itself drives cost reduction. Subsidies that seemed expensive when solar was costly bought volume, volume drove costs down the learning curve, and the result is that solar and wind are now frequently the cheapest source of new electricity in much of the world. Fossil fuel extraction shows no comparable learning curve, because its cost is dominated by the geology of finding and lifting a resource rather than by repeated manufacture.

Key idea: Solar and battery costs fell roughly 90 percent since 2010 by following manufacturing learning curves, a mechanism fossil extraction does not share.

Intermittency and the grid

Solar and wind are intermittent: the sun sets and the wind lulls, yet electricity must be supplied the instant it is demanded. Managing a grid rich in renewables therefore relies on several tools:

• Energy storage, such as batteries and pumped hydro, to save power for when it is needed.
• Long-distance transmission to average out local weather, since the wind is usually blowing somewhere.
• Flexible demand that can shift to times of plentiful supply.
• A low-carbon dispatchable backbone, such as hydro or nuclear, for the gaps.

None of these obstacles is fundamental; they are engineering and investment challenges, not physical impossibilities. The falling cost of solar, wind, and batteries has made a low-carbon grid increasingly practical.

It helps to distinguish two very different problems that both get called intermittency. The daily one, covering the gap between afternoon solar output and evening demand, is a matter of a few hours and is being solved economically with lithium-ion batteries at large scale. The seasonal one, covering a still, dark week in winter, requires storing energy for days or weeks, and batteries are a poor fit for that duty because the cost scales with the energy stored rather than the power delivered. Candidate answers include hydrogen, long-duration storage chemistries, geographically extensive transmission, and retaining a modest amount of dispatchable low-carbon generation.

Key idea: Intermittency is an engineering challenge, not a fundamental barrier; daily gaps are already handled by batteries while multi-day winter lulls need transmission, long-duration storage, or dispatchable low-carbon backup.

The parts that are genuinely hard

Electricity is the easy sector, and it is worth being candid that other sectors are not. Roughly four categories resist straightforward electrification.

• High-temperature industrial heat. Cement and steel need process temperatures above 1,000 degrees Celsius, and cement additionally releases CO₂ from the limestone itself, independent of the fuel used.
• Long-distance aviation and shipping. Batteries store far less energy per kilogram than jet fuel, so long-haul flight and ocean shipping need alternative fuels rather than electrification.
• Agricultural emissions. Methane from livestock and rice paddies and nitrous oxide from fertiliser come from biology, not combustion.
• Existing infrastructure. Power plants, buildings, and vehicles already built will keep emitting for their remaining lifetimes unless retired early.

Naming these honestly is not pessimism. It clarifies where research and policy attention are most valuable, and it explains why serious decarbonisation plans include carbon capture and carbon removal for residual emissions rather than assuming every source can reach zero.

Key idea: Electricity is the tractable sector, while high-temperature industry, aviation and shipping, agriculture, and existing infrastructure are the genuinely hard parts that justify carbon capture and removal.

Common misconceptions

• "Nuclear power is renewable." No. It uses finite uranium, so it is nonrenewable, though it is low-carbon.
• "Natural gas is clean and carbon-free." No. It emits less CO₂ than coal but is still a major emitter.
• "Intermittency makes a renewable grid impossible." No. It is an engineering challenge solved with storage, transmission, and flexible or dispatchable backup.
• "Solar panels emit as much carbon to build as they save." No. Lifecycle emissions are roughly 40 to 50 grams of CO₂-equivalent per kilowatt-hour against about 820 for coal.
• "Installed capacity tells you how much energy a source supplies." No. Capacity must be multiplied by capacity factor, which is roughly 20 percent for solar and 90 percent for nuclear.

Recap

• Fossil fuels supply most energy and most CO₂; coal is the most carbon-intensive and gas about 60 percent of coal per unit of heat.
• Renewables run on replenished flows and emit almost no operating CO₂.
• Lifecycle emissions run about 820 g CO₂-eq per kWh for coal against roughly 11 to 50 for wind, nuclear, hydro, and solar.
• Capacity factor converts nameplate power into delivered energy and differs greatly between technologies.
• Nuclear is nonrenewable but low-carbon; low-carbon energy includes renewables and nuclear.
• Solar and battery costs fell roughly 90 percent since 2010 by following manufacturing learning curves.
• Storage, transmission, flexible demand, and dispatchable backup balance the grid, while industry, aviation, shipping, and agriculture remain genuinely hard.

Sources

1. U.S. Energy Information Administration. (n.d.). Energy explained. eia.gov
2. U.S. Energy Information Administration. (n.d.). How much carbon dioxide is produced when different fuels are burned? FAQ. eia.gov
3. Clarke, L., Wei, Y.-M., De La Vega Navarro, A., et al. (2022). Chapter 6: Energy systems. In Climate change 2022: Mitigation of climate change (IPCC AR6 WGIII). ipcc.ch
4. Our World in Data. (n.d.). Energy mix. ourworldindata.org
5. Our World in Data. (n.d.). Electricity mix. ourworldindata.org
6. U.S. Energy Information Administration. (n.d.). U.S. energy facts explained. eia.gov
7. IPCC. (2022). Summary for policymakers. In Climate change 2022: Mitigation of climate change (AR6 WGIII). ipcc.ch

There is a curious and useful result buried in the climate physics: peak warming depends almost entirely on the cumulative total of carbon dioxide emitted, not on the path taken to get there. Emit the same total quickly or slowly and you arrive at approximately the same temperature. That near-linear relationship converts an intimidating global problem into a budgeting problem, and budgets, unlike vague aspirations, can be checked.

The big picture

Mitigation means reducing the greenhouse-gas emissions that cause climate change, or removing gases already emitted. The physics gives a clear finish line: because CO₂ lingers for centuries, temperatures roughly stop rising only when emissions reach net zero.

Think of the atmosphere as a bathtub filling with CO₂. Slowing the tap helps, but the water keeps rising until the inflow matches the drain. This lesson explains net zero, which sectors must change, and the policy tools that make it happen.

Why net zero, and what it means

Net zero is the point where any remaining emissions are balanced by an equal amount removed from the atmosphere, so the net addition is zero. Reaching and holding net zero is what it takes to stop the warming from growing, because as long as we add more CO₂ than we remove, the concentration keeps climbing and so does the temperature.

A small worked example makes the bathtub logic concrete. Suppose the world emits 40 units of CO₂ a year and removes 10. The net addition is 40 minus 10, which is 30 units still piling up each year, and warming continues. Cut emissions in half to 20 and the net is 20 minus 10, or 10 units, still positive, so warming still continues, just more slowly. Only when removals equal emissions, net zero, does the pile stop growing.

Key idea: Because CO₂ accumulates, temperatures stabilize only at net zero, when removals balance remaining emissions; merely emitting less still lets warming continue.

The carbon budget, and the arithmetic of a deadline

Because peak warming tracks cumulative emissions, scientists can state a remaining carbon budget: the total additional CO₂ that may be emitted while retaining a given chance of staying below a temperature limit. The IPCC assessed that from the beginning of 2020, holding warming to 1.5 degrees Celsius with a 50 percent chance allowed about 500 gigatonnes of CO₂, and with a 67 percent chance about 400. For 2 degrees Celsius at 67 percent chance the budget was about 1,150 gigatonnes.

Now do the division that makes those numbers concrete. Global CO₂ emissions run at roughly 40 gigatonnes a year. Dividing 500 by 40 gives about 12.5 years of emissions at current rates before the 1.5 degree budget is exhausted, counted from 2020. Several of those years have already passed. The 2 degree budget at 1,150 gigatonnes divided by 40 gives roughly 29 years.

Notice what the budget framing does and does not say. It does not say anything catastrophic happens the instant a threshold is crossed; impacts increase continuously with warming. It does say that delay is expensive in a specific, quantifiable way, because every year of emissions at current rates consumes a fixed and non-renewable share of the remaining allowance. Budgets are also revised as science improves, and they depend on assumptions about non-CO₂ gases, so treat any single figure as an assessed estimate with its own uncertainty.

Key idea: About 500 GtCO₂ remained from 2020 for a 50 percent chance of 1.5 degrees C, roughly 12.5 years at current emissions, which converts delay into a measurable cost.

What must change

Emissions come from several large sectors, and each needs its own solution:

• Electricity: replace fossil generation with solar, wind, hydro, nuclear, and storage.
• Transport: shift to electric vehicles, public transit, cycling, and efficient design; harder for aviation and shipping.
• Industry: decarbonize steel, cement, and chemicals through efficiency, electrification, hydrogen, and capture.
• Buildings: insulate, and switch heating from gas to electric heat pumps.
• Land and agriculture: cut methane and nitrous oxide, reduce deforestation, and restore forests and soils as carbon sinks.

Shifting a sector away from carbon-emitting energy and processes is called decarbonization. A cross-cutting lever is energy efficiency: using less energy for the same service is often the cheapest way to cut emissions, like sealing a drafty house before buying a bigger furnace. Carbon dioxide removal, from reforestation to engineered capture, will be needed to offset the emissions hardest to eliminate.

Sizing the sectors helps set priorities. In 2019, global net anthropogenic greenhouse gas emissions reached about 59 gigatonnes of CO₂-equivalent, roughly 12 percent above 2010 and 54 percent above 1990. The IPCC apportions those emissions across sectors approximately as follows.

Sector	Approximate share of 2019 global GHG emissions
Energy supply	about 34 percent
Industry	about 24 percent
Agriculture, forestry, and other land use	about 22 percent
Transport	about 15 percent
Buildings (direct emissions)	about 6 percent

The shares carry a warning against single-solution thinking. Even eliminating all transport emissions, the sector most visible in daily life, would leave roughly 85 percent of the problem intact. Conversely, energy supply and industry together account for well over half, which is why electricity decarbonisation and industrial process change dominate serious mitigation plans.

Key idea: Every major sector, electricity, transport, industry, buildings, and land, needs its own decarbonization path, with energy supply and industry together making up more than half of the 59 GtCO₂-eq emitted in 2019.

Carbon dioxide removal: necessary, and not a substitute

Because some emissions from agriculture, aviation, and industrial processes will be extremely difficult to eliminate, essentially every pathway to net zero includes some deliberate carbon dioxide removal to balance them. The options divide roughly into biological and engineered approaches.

• Reforestation and improved land management are available now and cheap, but the storage is reversible: a forest can burn or be cut, returning its carbon.
• Bioenergy with carbon capture and storage grows biomass, burns it for energy, and injects the CO₂ underground. It works in principle but competes for land and water at large scale.
• Enhanced weathering spreads crushed silicate rock to accelerate the natural process from Lesson 6.
• Direct air capture uses chemical sorbents to pull CO₂ from ambient air. It is verifiable and land-efficient but currently expensive and energy-hungry, and deployed at a scale thousands of times below what would be needed to matter globally.

Two cautions belong with any discussion of removal. First, the arithmetic: current removal capacity is minuscule compared with roughly 40 gigatonnes of annual CO₂ emissions, so removal cannot substitute for reductions on any near-term timescale. Second, the incentive problem: treating future removal as assured can justify present delay, and if the removal does not materialise at the assumed scale, the emissions have already happened. Removal is best understood as the tool for the residue after deep cuts, not as an alternative to them.

Key idea: Carbon removal is needed to offset genuinely hard-to-eliminate emissions, but current capacity is tiny against roughly 40 GtCO₂ a year, so it complements rather than replaces reductions.

Policy tools

Technology alone is not enough without policy that changes incentives. The main tools are:

• Carbon pricing, a carbon tax or a cap-and-trade market, which makes polluters pay for emissions and harnesses the market to find the cheapest cuts.
• Regulations and standards, which set limits directly, such as vehicle efficiency rules or clean-electricity requirements.
• Subsidies and investment, which speed clean technology down the cost curve, as happened for solar and wind.

Each instrument has documented strengths and weaknesses, and describing them is a matter of evidence rather than political preference. Carbon pricing is economically efficient because it lets emitters find the cheapest cuts themselves, but prices set too low change little, and the costs fall regressively unless revenue is returned to households. Standards deliver certain outcomes in a specific sector and are politically durable, but they cannot easily equalise the cost of abatement across sectors. Subsidies and public investment are effective at driving new technologies down the learning curves described in Lesson 17, but they can persist after they are no longer needed.

International agreements coordinate national efforts. The Paris Agreement of 2015 gathers national pledges toward holding warming well below 2 degrees Celsius, ideally near 1.5. Its architecture is deliberately bottom-up: countries set their own nationally determined contributions and revise them on a five-year cycle, with transparency provisions rather than enforcement. Assessments such as the annual UNEP Emissions Gap Report consistently find that current pledges, even if fully implemented, fall short of the reductions the temperature goals require, and that the gap between pledges and actual policies is larger still.

Most analyses find a mix of these tools works best, since no single instrument covers every sector. Which mix a country should choose involves value judgements about cost, fairness, and the role of government that science can inform but not settle.

Key idea: Carbon pricing, standards, and investment each have documented strengths and trade-offs, and the Paris Agreement coordinates voluntary national pledges that assessments find still fall short of its temperature goals.

Common misconceptions

• "Cutting emissions in half will stop the warming." No. As long as net emissions are positive, CO₂ keeps accumulating and warming continues, only more slowly.
• "Net zero means no one emits anything." No. It means remaining emissions are balanced by an equal amount removed, so the net addition is zero.
• "Technology alone will fix it without policy." No. Policy that changes incentives is needed to deploy the technology at scale.
• "Carbon removal means we do not have to cut emissions." No. Removal capacity today is thousands of times smaller than annual emissions, so it can only handle the residue after deep cuts.
• "Fixing transport would solve most of the problem." No. Transport is about 15 percent of global greenhouse gas emissions; energy supply and industry together exceed half.

Recap

• Mitigation reduces or removes greenhouse gases.
• Peak warming tracks cumulative CO₂, which is why a remaining carbon budget can be stated at all.
• About 500 GtCO₂ remained from 2020 for a 50 percent chance of 1.5 degrees C, roughly 12.5 years at current rates.
• Temperatures stabilize only at net zero, because CO₂ accumulates.
• Global emissions reached about 59 GtCO₂-eq in 2019, led by energy supply at 34 percent and industry at 24 percent.
• Energy efficiency is often the cheapest cut; carbon removal offsets the hardest residual emissions but cannot replace reductions.
• Carbon pricing, standards, investment, and the Paris Agreement drive and coordinate action, though pledges still fall short of the stated goals.

Sources

1. IPCC. (2022). Summary for policymakers. In Climate change 2022: Mitigation of climate change (AR6 WGIII; 59 GtCO₂-eq in 2019; sector shares). ipcc.ch
2. Riahi, K., Schaeffer, R., Arango, J., et al. (2022). Chapter 3: Mitigation pathways compatible with long-term goals. In IPCC AR6 WGIII. ipcc.ch
3. IPCC. (2023). Summary for policymakers. In Climate change 2023: Synthesis report (AR6 SYR; remaining carbon budgets). ipcc.ch
4. United Nations. (n.d.). The Paris Agreement. UN Climate Action. un.org
5. UN Environment Programme. (2025). Emissions Gap Report 2025. unep.org
6. Clarke, L., Wei, Y.-M., De La Vega Navarro, A., et al. (2022). Chapter 6: Energy systems. In IPCC AR6 WGIII. ipcc.ch
7. Our World in Data. (n.d.). CO₂ emissions (annual global and national emissions data). ourworldindata.org

In 2003 a heatwave killed tens of thousands of people across Europe, many of them elderly and living alone in cities with no air conditioning. Within a few years, France and other countries had built heat action plans: registries of vulnerable residents, cooled public spaces, and warnings issued days ahead. When comparable heat returned, the death tolls were far lower. The hazard had not changed. What changed was the preparation, and that is what adaptation means.

The big picture

Because some warming is already locked in by past emissions and the slow response of the oceans and ice, cutting emissions is not enough on its own. Adaptation means adjusting to the climate changes that are already happening or unavoidable, to reduce harm.

A simple way to hold the two ideas together: mitigation treats the cause (the emissions), while adaptation manages the consequences (the impacts). This closing lesson explains adaptation across sectors and weighs individual choices against collective action.

Mitigation versus adaptation

Mitigation reduces greenhouse-gas emissions; adaptation adjusts to the impacts that arrive anyway. Think of a warming planet like a rising river: mitigation is turning off the taps upstream, while adaptation is building your house higher so it survives the water that is already coming.

Both are necessary and complementary, not alternatives. The less we mitigate, the more we will have to adapt, and some impacts exceed what adaptation can handle. So they work as a pair, not a choice between one and the other.

Key idea: Mitigation treats the cause and adaptation manages the consequences; both are needed because some warming is already locked in and unchecked emissions would outpace any adaptation.

What adaptation looks like

Adaptation is practical and specific to each sector:

• Coasts: sea walls, restored wetlands and mangroves, elevated buildings, and in extreme cases managed retreat, deliberately moving people and assets away from areas that cannot be defended.
• Water: efficient irrigation, storage, and reuse to cope with drought and variable rainfall.
• Agriculture: drought- and heat-tolerant crops, shifted planting times, and diversified farming.
• Cities: shade trees and cool roofs to fight heat, and improved drainage for heavier downpours.
• Health and disaster: early-warning systems, heat action plans, and stronger emergency response.

Adaptation is often about building resilience (the capacity to absorb disturbance and recover) and reducing vulnerability (how susceptible a community is to harm). Its costs fall hardest on communities with the fewest resources, which raises questions of fairness at the heart of climate policy.

Key idea: Adaptation means concrete steps, from sea walls to cool roofs to early warnings, that build resilience and reduce vulnerability, with fairness a central concern.

Individual and collective action

Individuals can lower their own carbon footprint, the total emissions caused by a person or activity, by using less energy, driving and flying less, eating lower on the food chain, and reducing waste. These choices matter.

But the largest individual lever is arguably collective, or systemic action: change at the level of policies and institutions that govern large-scale emissions. Voting, advocating, and pushing the governments, employers, and utilities that control energy, transport, and land use reaches far beyond any single household. Climate change is a shared, global problem, and the physics is indifferent to intentions: what counts is the total emissions curve bending down to net zero. That is achievable with today's knowledge and technology; the remaining task is largely one of will, investment, and cooperation.

Key idea: Personal footprint cuts help, but the greatest individual leverage is collective, shaping the policies and institutions that control large-scale emissions.

Common misconceptions

• "Adaptation is the same as mitigation." No. Mitigation cuts emissions (the cause); adaptation reduces harm from impacts (the consequences).
• "If we adapt, we do not need to cut emissions." No. Some impacts exceed what adaptation can handle, so mitigation is still essential.
• "Individual choices are all that matter, or all that is pointless." Neither extreme. Personal cuts help, but collective action on policies and institutions has the greatest leverage.

Recap

• Adaptation adjusts to unavoidable climate impacts to reduce harm.
• Mitigation treats the cause; adaptation manages the consequences; both are needed.
• Adaptation is sector-specific, from sea walls to cool roofs to early-warning systems.
• It builds resilience and reduces vulnerability, with fairness a central issue.
• Individual footprint cuts help, but collective, systemic action has the most leverage.

Sources

1. IPCC, Sixth Assessment Report, Working Group II (2022), Summary for Policymakers on impacts, adaptation, and vulnerability. find source ↗
2. USGCRP, Fourth National Climate Assessment (2018), adaptation chapter. find source ↗
3. NASA Global Climate Change, "Solutions: mitigation and adaptation," climate.nasa.gov ↗.
4. US EPA, "Climate change adaptation resource center," epa.gov ↗.