How Science Works | How Small Cloud Droplets Become Falling Rain

Water vapour is gaseous water. An ordinary liquid cloud becomes visible because it contains suspended droplets that interact with light. Condensation supplies liquid water, but producing a droplet is not equivalent to producing a raindrop that survives the journey to the ground.

Consider a small droplet with a radius of 10 micrometres and a larger drop with a radius of 1 millimetre. These are example sizes, not boundaries separating every cloud droplet from every raindrop. A micrometre is one millionth of a metre; a millimetre is one thousandth. The larger radius is 100 times the smaller radius.

That sounds like a modest scaling until we remember that water occupies a volume. For spherical drops of the same density, water mass increases with the cube of radius. The larger drop contains the water equivalent of a million of the smaller droplets. Rain formation must explain this redistribution of mass, not merely a small enlargement of an existing object.

The liquid-water pathway is often called the warm-rain process. Here “warm” identifies growth without requiring an ice stage; it does not mean the rain feels warm on your skin. Rain can also emerge from ice-containing clouds after frozen precipitation melts. NASA’s explanation of the two pathways makes clear that rain does not have one universal origin.

A cloud may contain regions following different histories. Seeing liquid drops at street level cannot, by itself, tell us whether their water previously passed through ice.

Contents · Next chapter

In a rising parcel of air, expansion can lower temperature enough for water vapour to condense onto suitable aerosol particles. Once droplets are present, vapour can continue diffusing towards them and adding liquid water under favourable conditions.

But increasing a droplet’s radius becomes progressively harder through condensation alone. In a simplified diffusion-controlled description, the square of radius grows approximately linearly with time when the surrounding conditions are held fixed. A given increment in radius therefore takes longer for a larger droplet. Real conditions change, so this relationship is a useful model rather than a universal clock.

An original numerical illustration shows the consequence. Suppose a fictional constant environment adds 100 square micrometres to radius squared each time unit. A droplet beginning at radius 10 micrometres has radius squared 100. After one unit it has radius squared 200, giving radius about 14.1 micrometres. After three units it reaches radius squared 400, or radius 20 micrometres.

To reach radius 30 micrometres from the same starting point requires radius squared 900: eight time units. Doubling and tripling radius do not take equal additional intervals. These chosen units are not seconds and do not predict an actual cloud’s development.

This slowing helps explain why the transition from small cloud droplets to efficient collectors is scientifically important. Franklin’s original modelling study examines warm-rain growth and turbulence in precisely this context. Knowing that condensation and collection both occur is easier than predicting when a particular cloud will begin producing drizzle.

Contents · Previous chapter · Next chapter

A tiny water droplet already experiences gravity. It does not wait to cross a special mass threshold before acquiring weight. What differs with size is its motion through the surrounding air.

As a droplet settles relative to air, drag opposes its motion. Its settling speed depends on its size and on the flow around it. Small droplets have low settling speeds; within the cloud-to-rain size range, larger drops generally settle faster. Air itself is also moving, often upward in a developing cloud.

Keep two reference frames separate. A drop can move downward relative to air while still moving upward relative to the ground because the air carrying it rises faster. Conversely, a downdraft can help carry a drop downward. Statements that a cloud “holds up” droplets compress these distinct motions into one picture.

For a deliberately hypothetical example, take upward motion as positive. If air rises at 0.8 metres per second and a drop settles through it at 0.3 metres per second, the drop moves upward at 0.5 metres per second relative to the ground. If its relative settling speed were instead 1.2 metres per second in the same air, its ground-relative motion would be downward at 0.4 metres per second.

These are chosen velocities, not a size-to-speed calibration. The calculation isolates the idea that ground motion combines air motion and motion relative to air. It also shows why a photograph of a drop’s position cannot reveal all the forces or velocities involved.

NOAA’s precipitation overview introduces the connection between different droplet sizes, different fall speeds and growth by merging. For a precise explanation, keep gravity, drag and moving air in the picture throughout.

Contents · Previous chapter · Next chapter

Imagine two droplets settling in perfectly still air. If one falls faster and their paths are suitably aligned, it can catch up with the other. If they have identical settling speeds and remain separated, differential settling alone does not bring them together.

This is why the size distribution matters. A cloud containing a range of droplet sizes offers opportunities that an idealised collection of identical droplets does not. The distribution includes how many drops occupy each size range, not just an average size.

A collector’s radius also affects the region in which another droplet’s centre could approach closely enough for contact. For spherical drops with radii R and r, their surfaces meet when their centres are separated by R + r. The corresponding geometric target area is π(R + r)².

Multiplying this area by their relative speed gives an idealised volume swept per unit time. Multiplying again by the number of potential partners per unit volume estimates how many geometric encounters a collector could have per unit time. This is a model of opportunities, not yet a count of successful mergers.

The thought experiment identifies three independently useful questions. How large is the encounter target? How quickly do the two populations approach one another? How many possible partners are present? Increasing one factor does not guarantee rain if another is very small.

Cloud air is turbulent rather than perfectly still. Eddies can alter relative velocities and where droplets cluster. Equal-size droplets therefore need not be incapable of meeting in a real cloud. The identical-speed example diagnoses the limitations of a differential-settling-only model; it is not a claim that real clouds contain no other source of relative motion.

Contents · Previous chapter · Next chapter

A geometrical encounter does not ensure physical contact. Air flows around a moving drop, and a smaller neighbour may follow that flow around it. The trajectories are not equivalent to solid balls moving through empty space.

Researchers distinguish collision efficiency, describing how effectively geometric encounter opportunities lead to contact, from coalescence efficiency, describing how often actual contacts lead to merging. Their product is commonly used as a collection efficiency. The precise definitions must be checked when comparing studies, because a quoted “efficiency” may refer to only one stage.

The distinction can be understood through a hypothetical observation log. Suppose a model predicts 100 geometric encounter opportunities. If 40 produce contact and 30 of those contacts produce a single merged drop, collision efficiency is 40/100 = 0.40. Coalescence efficiency is 30/40 = 0.75. The combined successful fraction is 0.40 × 0.75 = 0.30, or 30%.

These invented numbers teach conditional counting. They are not measured efficiencies for cloud droplets. Depending on sizes and conditions, coalescence after contact can be very likely; that does not make the distinction unnecessary. Larger-drop collisions can also produce separation or breakup rather than permanent growth.

Wang and colleagues’ primary study of turbulent collision efficiency explicitly separates approach, aerodynamic interaction and coalescence. This staged description helps avoid a misleading illustration in which every droplet drawn inside a collector’s path is automatically absorbed.

The same logic matters outside meteorology. Counting opportunities, contacts and successful outcomes separately reveals where a process is being limited. An unexplained reduction in successful collection could originate at more than one stage.

Contents · Previous chapter · Next chapter

Draw eight equal circles to represent eight spherical droplets, each of radius r. Draw an arrow from them to one larger circle. Label the arrow “all eight merge; no evaporation or breakup”. What radius should the final circle have?

The initial total water volume is eight times 4πr³/3. If the final radius is R, conservation of volume requires 4πR³/3 = 8 × 4πr³/3. Cancelling the common factor leaves R³ = 8r³, so R = 2r.

Eight equal droplets make one drop with twice the radius, not eight times the radius. In a two-dimensional drawing, that final circle has four times the area of one small circle. Do not add the drawn areas as though the sketch were a flat puddle: the circles stand for three-dimensional spheres.

Now draw just two equal droplets merging. Their combined radius is the cube root of two times r, approximately 1.26r. A single equal-size merger enlarges radius by about 26%, even though it doubles mass. This is why droplet number, radius and water mass cannot be used interchangeably.

Return to the 10-micrometre and 1-millimetre example. Since 1 millimetre is 1,000 micrometres, the radius ratio is 1,000/10 = 100. The volume ratio is 100³ = 1,000,000. That does not mean one collector must undergo exactly a million encounters. It could collect drops that have already grown or merged, and condensation can add water too.

Finally, imagine a sealed mathematical box where merging is the only process. Droplet number decreases as mergers occur, but total liquid mass stays constant. A model that increases total water merely because drops combine has violated its own bookkeeping. Actual clouds exchange water with their surroundings, so their full budget also includes condensation, evaporation, transport and precipitation leaving the region.

Contents · Previous chapter · Next chapter

Once some drops grow larger, they may settle faster relative to smaller droplets and present larger encounter targets. Successful collection can therefore support further collection. The population becomes increasingly unequal: a minority can hold a substantial fraction of the liquid water.

This does not imply unlimited runaway growth. Potential partners are depleted or transported away, drops can leave the cloud, and breakup can return water to smaller sizes. Turbulence and evaporation also change the conditions during the process. A local tendency to accelerate growth is not a promise that every collector becomes enormous.

A useful comparison keeps total water fixed. Population A has many similar small droplets. Population B redistributes the same water into a mixture including fewer larger collectors. They have equal liquid mass but different numbers, size distributions and potential encounter rates. A measurement of water content alone cannot tell these populations apart.

This explains why aerosols complicate rainfall arguments. Changing the number and properties of particles on which droplets form can change the initial distribution. But a statement such as “more particles always mean less rain” ignores subsequent air motion, mixing, cloud lifetime and other environmental differences. A controlled comparison must specify what else has been held constant.

Chen and colleagues’ 2016 direct numerical simulations investigated how turbulent conditions and droplet sizes influence collision statistics. Such studies make mechanisms testable under defined conditions. They do not turn one laboratory-scale or numerical result into a universal multiplier for every cloud.

For the reader, the practical lesson is to ask about the whole distribution. An average can conceal exactly the relatively uncommon collectors that matter for the next stage of growth.

Contents · Previous chapter · Next chapter

Watching raindrops merge on a window demonstrates that liquid drops can coalesce. It does not measure the collision rate inside a cloud. Glass changes the forces, constrains motion and provides a surface on which drops accumulate. A familiar demonstration can support an idea without reproducing the atmospheric mechanism.

More informative measurements sample droplet sizes and their local environment within clouds. Researchers want to know where unusually large droplets appear, how numerous they are, what the surrounding size distribution looks like, and how those locations relate to air motion and mixing.

Small and Chuang’s 2008 RICO field study measured shallow cumulus droplets with a phase Doppler interferometer. Drops above 55 micrometres in diameter, termed collision–coalescence initiators in that study, were preferentially found near cloud tops and in clustered environments with distributions shifted towards larger sizes. The authors associated these environments with entrainment and proposed a role for locally enhanced turbulent collection.

Notice three limits. Their size criterion used diameter, whereas our water-budget examples used radius. A location associated with large droplets does not alone establish the entire causal sequence. And the proposed explanation was a hypothesis tested against particular observations, not proof that all rain begins at cloud top.

An informative follow-up comparison would ask whether a model reproduces both the large-droplet abundance and the measured environmental patterns. Matching only the eventual rainfall total could hide compensating errors: too many collectors formed initially but too much subsequent evaporation, for example.

Instruments also have sampling limits. Rare droplets require enough sampled air to estimate their frequency reliably. Distinguishing a true large drop from measurement artefacts matters especially when the scientific question concerns the uncommon tail of a distribution.

Contents · Previous chapter · Next chapter

A drop that exits cloud base may enter drier air and lose water by evaporation. Some precipitation disappears before reaching the surface, producing the falling streaks called virga. A cloud can therefore generate precipitation aloft without delivering rain to the ground at that location.

Ground rainfall depends on several linked stages: producing collectors, sustaining growth, transporting drops out of the cloud, and retaining enough liquid during descent. Winds can also displace falling drops horizontally. The patch of ground below a photographed cloud need not receive everything formed directly overhead.

Think of two otherwise hypothetical descents. In one, the air below cloud base is relatively humid and the path short. In the other, the air is drier and the path longer. Equal drops starting at cloud base need not arrive with equal remaining sizes. Without specifying conditions, “the cloud made large drops” is an incomplete rainfall prediction.

There is also no requirement that every cloudy region rain simultaneously. Different parts of a cloud can have different histories and distributions. Measurements averaged over a large region may smooth out brief, local episodes of efficient collection.

The frontier is not whether water droplets can merge. It is the quantitative problem of predicting distributions and timing in changing air. A 2025 primary research preprint on polydisperse collision kernels explores improved representations of collisions in turbulent droplet populations. As a preprint, it should be read as ongoing research, alongside established observations and peer-reviewed work, rather than as a settled replacement for cloud physics.

Contents · Previous chapter · Next chapter

Why can a liquid cloud exist without raining at the ground? Its droplets may remain small, their movement depends on air motion as well as settling, and drops leaving the cloud may evaporate. Visible liquid water does not establish successful surface precipitation.

Do two equal drops make a drop with twice the radius? No. Under the spherical, equal-density, no-loss assumptions, mass doubles and radius grows by the cube root of two. Eight equal droplets are needed for twice the radius if their water alone supplies the final drop.

Why does a spread of sizes help gravitational collection? Different settling speeds permit overtaking, while larger drops offer different encounter geometry. Turbulence can add relative motion and change spatial clustering, so differential settling is only one component of actual cloud encounters.

What would make a convincing scientific explanation stronger than a cartoon? A water budget that balances, separate treatment of encounters and mergers, measured size distributions, explicit air-motion conditions, and predictions that can fail when tested. The explanation should also identify what it cannot infer from a window, photograph or surface rain gauge alone.

A safe learning exercise is to use the paper diagram in Chapter 6 and recompute the final radius after 27 equal droplets merge. The cube root of 27 is three, so the resulting radius is 3r under the same assumptions. Then describe which assumption would fail if half the total water evaporated before the merger was complete. No outdoor storm observation or equipment is needed.

The central chain is now visible: condensation creates liquid droplets; differences in motion create encounters; some encounters become collisions and mergers; the size distribution evolves; and only some resulting drops survive to the ground. Return to Atmospheric Science to place that small-scale chain inside the larger moving atmosphere.

Contents · Previous chapter · Continue to Atmospheric Science