Can One Change Set Off Many More?

You met climate change in Grade 10 as two lists: its causes (greenhouse gases, cutting forests, changes in how much sunlight the Earth reflects) and its effects (melting snow, glacial lakes, drought and floods) Grade 10: Grade 10: Causes of climate change. But in the real Earth, an effect is often also a cause. Melting snow is an effect of warming, and also a cause of more warming. When an effect comes back round and changes its own cause, the chain of changes has closed into a feedback loop.

There are two kinds, and you know both from daily life.

At a school program, someone holds the microphone too close to the loudspeaker. A small sound goes into the microphone, comes out of the speaker louder, goes back into the microphone, comes out louder still, and in a second there is a terrible howl. Each time round the loop, the change grows. This is an amplifying feedback, also called a positive feedback or a reinforcing loop.

The water tank on your roof has a float valve. As the tank fills, the float rises and slowly closes the valve, so less water flows in. If someone opens a tap and the level falls, the float drops and the valve opens again. Whatever happens, the level is pulled back to the same place. This is a balancing feedback, also called a negative feedback or a stabilizing loop.

How to tell which kind a loop is

Write the loop as a chain of links, each saying how one thing changes the next. A link is same if more of the first thing gives more of the next (more sound in, more sound out). It is opposite if more of the first gives less of the next (a higher float, less water flowing in). Then go once round the loop and count the opposite links:

• an even number of opposite links (0, 2, 4…): the change comes back round pushing the same way. Amplifying.
• an odd number (1, 3…): it comes back pushing the other way. Balancing.

It works like multiplying signs: two minuses make a plus. For the tank: higher level → float closes the valve more (same) → less water flows in (opposite) → the level rises less (same). One opposite link, so the loop is balancing, as you know it is.

The share of the sunlight falling on a surface that it reflects is its albedo, from the Latin albus, white. It is written as a fraction from 0 (reflects nothing, absorbs all) to 1 (reflects all). You saw in Grade 8 that dark, dull surfaces absorb more sunlight than white ones Grade 8: Grade 8: The color of an object and radiation. Albedo puts a number on it:

Surface	Albedo (typical)	Share of sunlight absorbed
fresh snow	0.80–0.90	10–20%
snow-covered sea ice	up to 0.90	10% or more
bare sea ice	0.50–0.70	30–50%
old, dirty snow	about 0.5	about 50%
bare glacier ice	about 0.3–0.4	60–70%
grass and crops	about 0.2–0.25	75–80%
forest	about 0.1–0.15	85–90%
rock, soil and the rubble on glaciers	about 0.1–0.2	80–90%
open sea or a lake	about 0.06	94%

The Earth as a whole has an albedo of about 0.30: it reflects about 30% of the sunlight it gets, as you learned in Grade 10, mostly from clouds, snow and ice.

The ice–albedo loop

Now put the links together. Something warms the Arctic or the Himalaya a little. Some snow and ice melt (warmer → less ice: opposite). With less ice, less of the ground or sea is bright (less ice → lower albedo: same). A lower albedo means more sunlight absorbed (opposite). More sunlight absorbed means more warming (same). Two opposite links: an even number, so the loop is amplifying. This is the ice–albedo feedback. (Grade 10 mentions it in one sentence, under changes in how much sunlight the Earth reflects; here you can see why it is a loop, and below, the evidence that it is happening.)

It works in reverse too. If the Earth cools a little, snow lasts longer and spreads further, the albedo rises, less sunlight is absorbed, and the Earth cools more. An amplifying loop makes any change grow, whichever way it goes.

An idea is not enough: scientists need data to show a feedback is really happening. The best place to look is the Arctic Ocean, round the North Pole. Much of it is covered by sea ice, frozen seawater a few meters thick, which melts back every summer and freezes again in winter. It is smallest in September. Satellites have measured it since late 1978.

Notice how much the points jump about from year to year. The record low, 3.57 million km² in 2012, was followed by 5.21 in 2013. Weather, winds and storms push the ice around, so one year tells you little. That is why scientists look at the trend over many years, not at one year. Even so, every September from 2007 to 2026 had less ice than any September before 2007.

The shrinking ice is the first link. Is the rest of the loop really there? Two more sets of measurements say it is.

• The Arctic got darker. Satellites also measure how much sunlight is reflected back to space. In 2014 three scientists in California, Kristina Pistone, Ian Eisenman and V. Ramanathan, showed that the Arctic’s albedo fell from 0.52 to 0.48 between 1979 and 2011, as the ice gave way to open sea. That meant the Arctic was taking in about 6.4 W more sunlight on every square meter, averaged over the whole year. Clouds hadn’t changed enough to explain it; the lost ice did.
• The Arctic warmed fastest. Since 1979 the Arctic has warmed nearly four times as fast as the Earth as a whole (Mika Rantanen and others, 2022). The ice–albedo loop is one of the main reasons, together with others in the air above the Arctic.

So the chain holds, link by link: less ice (measured), a darker Arctic (measured), more sunlight absorbed (measured), more warming (measured), and the warming is greatest exactly where the ice is going. That is how a scientist makes the claim that one change set off the others.

In the Himalaya

The same loop runs on Nepal’s glaciers. When winter snow melts off a glacier in spring, it uncovers older, dirtier snow and gray ice, which absorb far more sunlight. Many of Nepal’s glaciers are covered by rock rubble near their ends, and where it is thin, the dark rubble warms and melts the ice under it faster. (Where it is thicker than a few centimeters, it does the opposite: it shades the ice like a blanket.) And where meltwater collects into a lake, the dark water absorbs most of the sunlight that falls on it, warms, and eats into the ice cliff at its edge, which breaks off in blocks. The lake grows, so it absorbs more sunlight, so it grows faster.

Imja Tsho, below Island Peak in the Everest region, was a few small ponds on the glacier in 1962, about 0.03 km² of water. By 2013 it was a lake of 1.35 km², with much of the glacier’s loss happening where its ice broke off into the lake Grade 10: Grade 10: Glacial lake outburst floods.

If the ice–albedo loop makes every warming bigger, why hasn’t the Earth’s temperature run away, up or down, long ago? Because there is a stronger loop pulling the other way.

Every warm object gives off heat as infrared radiation, and the hotter it is, the more it gives off: a hot iron radiates far more than a warm one. The Earth is the same. If it warms, it gives off more infrared to space, so it loses energy faster, which cools it back. Warmer → more heat given off (same) → less energy kept (opposite) → cooler (same). One opposite link: balancing. This loop is the reason the Earth settles at a steady temperature at all, and it is far stronger than any other: for each degree of warming, the whole Earth gives off about 3.2 W more on every square meter.

Other loops add to it or take from it:

• Water vapor (amplifying). Warmer air takes up more water vapor, about 7% more for every °C, and water vapor is a greenhouse gas. It is the strongest of the amplifying loops.
• Ice–albedo (amplifying), as you have seen.
• Clouds (both). Low, thick clouds reflect sunlight and cool; high, thin clouds hold heat in and warm. How clouds will change is the hardest part of the climate to pin down, but the evidence so far says their total effect amplifies warming a little.
• Frozen ground (amplifying). The ground of the far north, and of the high Himalaya and Tibet, stays frozen all year: permafrost. The permafrost of the far north holds the remains of plants that never rotted: more than one and a half times as much carbon as there is in the air today. As it thaws, microbes rot them and release carbon dioxide and methane, which warm the Earth more.

How much does a loop multiply a change?

Suppose something pushes the Earth’s temperature up by an amount we call the push. The amplifying loops then add more warming: if they add a fraction f of the push in the first round, that extra warming goes round the loop too, and adds f of itself, and so on, each round smaller than the last. The total is

$$\text{total change} = \text{push} \times (1 + f + f^2 + f^3 + \dots) = \frac{\text{push}}{1 - f}$$

The fraction f is the loop’s feedback strength. If f is 0, there is no feedback; if f is between 0 and 1, the change grows but settles at a bigger value; if f is negative, the loop is balancing and the change ends up smaller than the push. And if f reached 1 or more, each round would be as big as the last or bigger, and the change would never settle: a runaway.

Notice that at large f, a small change in f makes a big change in the total: at f = 0.6 the total is 3.0 °C, at 0.7 it is 4.0 °C, at 0.8 it is 6.0 °C. That is why scientists work so hard to measure the cloud feedback: a small uncertainty in f means a large uncertainty in the warming.

Feedback loops don’t only run between ice and air. One of the most important runs between plants, water and soil, and you can see it on Nepal’s hills.

Stand in a forest during monsoon rain. The rain hits leaves first, and drips down gently. Under the trees, a spongy layer of fallen leaves and the roots of plants hold the soil open, and the water soaks in (it infiltrates). It flows slowly through the ground, and comes out weeks later in springs and streams. Now stand on a bare, grazed slope in the same rain. The drops hit the bare soil hard, knock its grains loose and seal its surface. The water can’t soak in fast enough, so it runs off over the surface, gathering into little channels that carry the soil away.

Measurements on test plots in Nepal’s Middle Hills show how big the difference is. Undisturbed forest and grassland lost less than 0.1 t of soil per hectare a year. Degraded sal forest lost 3 to 10 t. Bare slopes lost more than 15 t. The plots also showed that the plants close to the ground, and the fallen leaves, mattered more than the trees above: a forest whose undergrowth has been grazed and swept away loses soil even under a full canopy.

Now follow the loop. Fewer plants → more runoff (opposite) → more soil washed away (same) → thinner soil, holding less water and fewer nutrients (opposite) → fewer plants can grow (same)… Two opposite links, an even number: amplifying. Once a slope has lost enough of its cover and soil, it can go on losing them even if nobody cuts or grazes there any more. And the change spreads to other systems: the washed-off soil fills rivers and reservoirs downstream, springs fed by water that used to soak in dry up in the dry season, and floods come faster after rain.

There is good news too: the loop also runs backwards. In the 1970s and 80s, many of Nepal’s hills were losing their forests fast. When forests were handed to the communities beside them, the cover came back on many hills, and with it the undergrowth and fallen leaves that hold the soil and let rain soak in Grade 10: Grade 10: Measures to control climate change. An amplifying loop can be turned round, if something pushes hard enough the other way.

When a wetland dries

Wetlands and lakes can have a loop of their own. Their water evaporates and keeps the air above them humid, and some of that water comes back nearby as dew, mist and rain. If a wetland shrinks, less water evaporates, the air gets drier, less water comes back, and the wetland shrinks more: less wetland → less evaporation → less humidity → less dew and rain → less wetland. Every link is “same”: zero opposite links, so the loop is amplifying. (Careful: if you write “drier air” instead of “less humidity”, that link becomes “opposite” and the next one does too, and the count is still even. The answer can’t depend on how you name things, and it doesn’t.)

The biggest example is the Aral Sea in Central Asia, once the world’s fourth-largest lake. From the 1960s, most of the water of the two rivers that fed it was taken off into canals to water cotton fields. The lake shrank to a small part of its size. Around where it was, the summers have become hotter and the winters colder: without the water to warm the air in winter and cool it in summer, the land swings between extremes. Salty dust now blows off the dry lake bed onto the fields around.

A dam is a change to one small part of the Earth’s surface: a wall across a river. Follow what it sets off.

1. Upstream: the river stops, and drops its load. Behind the dam, the river’s water slows to almost nothing in the reservoir, so the sand and silt it was carrying settles to the bottom. Reservoirs slowly fill up. Nepal’s Kulekhani reservoir (Indra Sarobar, in Makwanpur), finished in 1982, could hold 85.3 million m³ when it was new. In July 1993 a huge cloudburst over its hills, with more than 500 mm of rain in a day at one rain gauge, washed in enormous amounts of soil and rock, and by 2010 the reservoir held only 64.9 million m³. Check dams have since been built upstream to catch sediment before it arrives.
2. Around the reservoir: the ground fills with water. Water from the reservoir seeps into the rock and soil around it, so the groundwater rises, and wells nearby may fill more easily. (In the wrong rocks, it can also make slopes slide.)
3. Downstream: hungry water. The water let out below the dam is clear: it has left its load behind. A river that can carry sand but has none picks it up from its own bed and banks, so below many dams the riverbed is dug deeper, and its banks collapse. Engineers call it hungry water. Here there is a balancing loop: as the river washes out the fine sand, the bed is left covered with coarse stones it can’t move, and the digging slows down.
4. At the coast: the delta shrinks. A river’s delta is built from the sediment it brings to the sea, while waves wash sediment away. With the sediment held back by dams, the waves win. After Egypt finished the Aswan High Dam in 1970, the sediment the Nile carried to the Mediterranean fell from more than 100 million tonnes a year to almost none, and in places the coast of its delta retreated by more than 100 m a year before sea walls were built.

Not every chain of changes is a loop. A dam starving a delta of sand is a chain: one change causes another, which causes another, but nothing comes back round to change the dam. A loop needs the last change to act on the first. Both matter, and in the real Earth they are tangled together.

• Why Has the Climate Changed Before? The ice ages came and went as the Earth’s orbit slowly changed, with feedbacks like Croll’s turning small changes into big ones.
• Where Does the Carbon Go? Following carbon through the air, the sea, rocks and living things, and how the carbon cycle feeds back on the climate.
• What Does Water Do to Rock? How water breaks rock, dissolves it and carries it away: the erosion and sediment in this topic, from the rock’s side.