A thunderstorm starts as a bubble of air that ends up warmer than its surroundings and keeps rising. Three pieces of physics explain most of what happens next: buoyancy read off a sounding, latent heat that keeps the bubble warm as its vapor condenses, and a cold pool that lets storms rebuild themselves in long lines.
Forecasters read storm potential from a sounding: the air temperature at each height, measured by a weather balloon. Imagine lifting a small parcel of surface air. It expands and cools as it rises. Wherever it is warmer than the air around it, it floats upward on its own, and the area between the two curves measures the energy it gains.
How to read this: left is colder, up is higher. The green line is the real air; the gold line is a bubble of surface air pushed upward. Where gold is right of green (red shading) the bubble is warmer than its surroundings and floats up on its own.
Try this: click Stable, then Explosive, and watch the red area grow.
Below cloud base the parcel cools at the dry rate g/cp = . Its water vapor stays fixed, so it saturates at the lifting condensation level, roughly between temperature and dew point. Above that, condensing vapor releases heat and the parcel cools more slowly: about in warm, humid air and in the cold air higher up.
Model: an undiluted parcel lifted from 1000 hPa, with all condensed water dropped at once (a pseudo-adiabat). Ice, the weight of water and virtual temperature are ignored. The environment has one lapse rate up to a tropopause at or , whichever comes first, and is isothermal above it.
√(2·CAPE) assumes every joule of buoyancy turns into upward speed. Real updrafts mix in drier air from outside, carry the weight of their cloud water and have to push air out of the way, so measured peaks are usually well under this number. The parcel's momentum still carries it past the EL, which is why strong storms show an overshooting top above the flat anvil.
CIN works as a lid. Something has to lift the parcel through it first: a front, a hillside or the outflow of another storm. A strong lid can hold storms off all afternoon while CAPE keeps building.
Warm air can hold far more water vapor than cold air: about , following the Clausius–Clapeyron relation. When a rising parcel cools to its dew point, the extra vapor condenses into cloud drops and hands back the heat it took to evaporate it. Two identical parcels rise from the same spot below. The gold one condenses its vapor; the dashed twin is forbidden to.
How to read this: on the right, two identical bubbles rise together. The gold one turns its water vapor into cloud and the grey dashed one can't. The shaded gold gap between them is the extra warmth the cloud-making releases; the gold ring flashes when condensing begins.
Try this: raise Relative humidity and press Release parcel: cloud base drops and the gold gap opens sooner.
The first line is the Bolton (1980) fit to the Clausius–Clapeyron relation, with T in °C. Divide by pressure to get grams of water per kilogram of dry air: qs ≈ 0.622·es/(p − es), shown on the left curve at 1000 hPa.
The second line is energy bookkeeping. Each gram of vapor that condenses in a kilogram of air warms it by about 2.5 K. The gap between the twins at any height equals Lv/cp times the water condensed so far, computed step by step with p from the hydrostatic equation.
A tropical parcel can carry 20 g/kg of vapor. Condensing most of it is worth about of heating spread over the climb, more than enough to keep it warmer than its surroundings for . That is the fuel tank behind CAPE in Lab 1, and the same engine drives W00 · Hurricanes. Cold, dry air has almost nothing in the tank, which is why winter thunderstorms are rare and shallow.
Rain falling out of a storm cools the air beneath it by evaporation. That cold air sinks and spreads along the ground as a cold pool, and its leading edge, the gust front, shoves warm air upward like a snowplow. New cells form wherever it lifts. Whether that lift goes straight up depends on the wind shear ahead of the line. This is an idealized cross-section, drawn in the frame moving with the gust front.
How to read this: a side view of a line of storms moving right. Cold, rain-cooled air (blue) spreads along the ground and shoves warm air up (amber). Each pale puff is a storm cell; an amber ring marks one just being born. The sliders' Δθ is how much colder the pool is, and ΔU is how much the wind speeds up with height.
Try this: press Set ΔU = C and watch the amber updraft stand straight up.
The cold pool is a density current. Its speed follows from its buoyancy deficit gΔθ/θ acting over its depth H. At the gust front it makes air spin one way. Low-level shear in the warm air ahead spins it the other way. When the two match, the air lifted at the gust front rises straight up and reaches its LFC easily, so cell after cell keeps forming.
This balance argument comes from Rotunno, Klemp and Weisman (1988). The tilt angle drawn here is a toy formula, 55° × (C − ΔU)/(C + ΔU), chosen to show the direction of the effect; the real geometry comes from full simulations.
A single storm rains itself out in under an hour. A line keeps making new cells at its front edge, while older cells drift back and merge into a wide sheet of lighter, steady rain. A mesoscale convective system can cover a whole state and last through the night. Where the line moves slowly, or new cells keep forming over the same ground, rain totals pile up and cause flash floods.
A strong cold pool can also drive a jet of air down from the back of the system to the ground. The line bows outward where it lands. When a bowing system keeps going for hundreds of , its wind damage swath is called a derecho. For the rotating storms that make tornadoes, see W01 · Tornadoes.