Skip Navigation

March-April 2010

ResizeResize Text: Original Large XLarge

Weather Queries

We had record cold one night last March in northern Idaho (−5°F at my house). Often when it gets very cold and the sun comes out, I can see fine snowflakes glistening in the air. I believe that this is atmospheric moisture freezing and forming small crystals. Is there a name for this, and is it considered precipitation? I report daily precipitation for the Community Collaborative Rain, Hail, and Snow (CoCoRaHS) Network and need to know.

Bob Wilson
Athol, Idaho

I've seen the phenomenon you describe many times on cold mornings with calm winds, usually around sunrise or soon after, and especially if the sky has recently cleared following overnight snow. I don't think these snow crystals originate in clouds that have since moved on or dissipated. Rather, these crystals form close to the ground in clear air because of intense radiational cooling, sometimes aided by a fresh snow cover.

When the sky is clear and the atmosphere is cold, infrared radiation, always emanating from the Earth's surface, can pass mostly unimpeded through the atmosphere into outer space. At night, or if the sun is low in the sky, this cools the surface and the air in contact with the surface. If the humidity is still high from earlier snowfall and the sky suddenly clears, the cooling is sometimes sufficient to cause the formation of tiny ice crystals on microscopic particles called ice nuclei. The crystals grow directly from the water vapor in the air; there is no freezing of liquid involved.

Caption: Diamond dust reflecting light from near a brilliant sun dog, Boulder, Colorado, 8:00 a.m., January 5, 2004.

Caption: Diamond dust reflecting light from near a brilliant sun dog, Boulder, Colorado, 8:00 a.m., January 5, 2004.

Most likely these are small plate-like hexagonal crystals. They fall with a wobbling motion like a feather or a leaf with their hexagonal faces pointing downward. You can see them when the sun comes out because they act as tiny mirrors, reflecting sunlight toward the eye when their orientation is just right. Because they sparkle in sunlight, these ice crystals are called diamond dust. Their fall speed is very low, just a few centimeters per second. Because they slowly settle to the ground, this qualifies as precipitation. The amount of H2O in these crystals is minute, so you'd report a trace of precipitation. Only in the polar winter, when these crystals can fall for days at a time, might there be enough to be measurable. Weather services report “ice crystals” (IC) in the METAR surface observation. The official definition of IC is “a fall of unbranched ice crystals in the form of needles, columns, or plates.”

If you examine the accompanying photo very carefully, you will notice quite a few very small white dots above, below, and to the left of the brilliant sun dog (technical word parhelion). These minute points of light are reflected from the parhelion by the tiny hexagonal plate crystals floating in the air. It is nearly certain that these are plate crystals because they are responsible for the sun dog in the first place. For a clear explanation of sun dogs and how they form, consult Rainbows, Halos, and Glories, by Robert Greenler, Cambridge University Press, 1980, pp.23–28.

Having grown up in the Deep South, I am familiar with the term “lady rain” for a rainfall that is soft and gentle. In the tropical rainforests of Southeast Asia, most of the rains are anything but soft and gentle. The droplets are often so large that they blast through a cloth umbrella, leaving you covered in a fine spray. When you are caught outside without an umbrella, the drops fall with so much force that they actually sting your exposed skin. What factors determine the differences in size and force of raindrops between temperate and tropical regions?

William M. Fountain
Lexington, Kentucky

I'll address the force of raindrops first, which is equivalent to their fall speed. The fall speed of raindrops in still air depends upon their size: larger raindrops fall faster. At sea level, a raindrop with a diameter of 0.2 millimeters (mm) has a terminal speed of 0.25 m s−1 (meters per second). A drop 2 mm in diameter falls at about 4 m s−1. A very large drop with an 8-mm radius falls at nearly 9 m s−1 and lands with enough force to penetrate a fabric umbrella with a fine spray.

Several factors determine the size of raindrops. Tiny droplets first form on cloud condensation nuclei (CCN), microscopic particles in the air that are wettable, that is, water can spread out on their surface as a thin film, or, in some cases, dissolve the particle. CCN are far more numerous (typically hundreds per cubic centimeter) over land than over the open ocean (often less than 100 per cubic centimeter), because most CCN, particularly smoke from biomass burning, industrial pollution, combustion of fossil fuels, or fine dust raised by the wind, originate over land. Over continents, vapor tends to condense on numerous CCN, forming a large number of small cloud droplets having a narrow range of sizes. Over oceans, competition for CCN is greater; a smaller number of larger droplets condense, having a greater range of sizes.

Condensation alone is too slow to grow droplets large enough to fall from a cloud as rain. In order for raindrops to form, cloud droplets must collide with each other and coalesce. Neither event is guaranteed. First, some droplets must be falling with respect to others so that collisions are possible. Second, even if they collide, two droplets will not necessarily coalesce; sometimes a thin cushion of air between them causes them to bounce apart. A cloud with a wider range of droplet sizes has a greater spread of droplet fall speeds and hence a better chance for the collision and coalescence processes to operate. Thus, small oceanic clouds have a much better chance of producing precipitation than small continental clouds. If you have ever been soaked by what appears to be drizzle in Hawaii but remained comparatively dry in an Atlanta drizzle, you can correctly surmise that the drop size in Hawaii, where the air is clean, is much larger than the drop size in Atlanta, where the air contains many more CCN. The bottom line is that, of two clouds containing the same amount of liquid water, the one containing a smaller number of larger droplets is more likely to produce rain than the one containing a much larger number of smaller droplets.

The arguments above apply to drop sizes in small clouds. To manufacture very large drops, one needs tall clouds and vigorous convection, though not necessarily a thunderstorm.

Aerodynamic forces acting on falling raindrops limit their size and ultimately cause their breakup. A 1-mm diameter drop is very nearly spherical, held to that shape by surface tension. A 2-mm drop falls faster, and the air stream rushing by distorts the shape, flattening the sphere in the vertical direction. A 3-mm drop has the shape of a chestnut, its upper surface convex but its lower surface nearly flat or even slightly concave. By the time a raindrop approaches 5 mm in diameter, aerodynamic forces have hollowed out its bottom, giving it the shape of a jellyfish or a parachute. A toroidal ring of water encircles the bottom perimeter. Most drops don't grow much larger than this, but a very few drops have been measured at more than 8 mm in diameter. The theoretical upper limit for drop size before spontaneous breakup is thought to be about 10 mm.

How might such huge drops form? In a paper titled “Super-Large Raindrops,” Peter Hobbs and Arthur Rangno (in Geophysical Research Letters, Vol. 31, L13102, 2004) speculate that in a polluted atmosphere with large concentrations of CCN, a few giant CCN (diameter greater than 0.02 mm), could grow large droplets very quickly. If these fell through a cloud with exceptionally high concentrations of smaller droplets, they could grow very large through frequent collisions and coalescence. But they documented equally large drops in a clean atmosphere as well, where giant CCN were not found. In this case a broad spectrum of drop sizes (wide variation in fall speed) and exceptionally high cloud water content could produce jumbo-sized raindrops.

I live at 40°N in Boulder, Colorado, where the freezing level is seldom more than 10,000 feet above ground. Hail is thus frequent in our area. Whenever I see very large drops splattering on the pavement at the beginning of a thunderstorm, I worry about hail, and hail often follows. The very large drops often result from hail that has melted just before reaching ground, but the drops have not yet broken apart.

To summarize, small clouds are more likely to generate rainfall if they contain a lower concentration of larger droplets rather than a higher concentration of smaller droplets, and the difference between such clouds is due to the number of available CCN. Very large droplets can form in the tropics and midlatitudes, in clean and polluted air, but a prerequisite seems to be an exceptionally large liquid water content in the cloud, which enables larger droplets to grow quickly by collisions and coalescence. Hail grows as a falling ice particle collects droplets at subfreezing temperatures. If the hail melts just before reaching the ground, large (and cold) raindrops are the result.

I have always wondered why there are so many shapes and sizes of snowflakes. I thought maybe it had to do with the amount of water incorporated in the flakes. Also, can you explain why some snow falls in the form of a ball? This is snow, not hail, right?

Annie Scoles
Orofino, Idaho

I've been waiting for more than 25 years for someone to ask these questions, because the structure and form of falling snow is a fascinating subject. The answer to your questions is in four parts: 1) background material, 2) a discussion of how the temperature and water vapor supply in the cloud determines the crystal shape, 3) an examination of new-fallen snow, and 4) a look at how riming changes the appearance of snow crystals.


Consider two different clouds. The first consists only of tiny droplets. If the cloud droplets are neither growing by condensation nor evaporating (equilibrium prevails), the partial pressure exerted by the water vapor surrounding these droplets (e) has a very specific value: the saturation vapor pressure with respect to liquid, esi it depends only upon the temperature, as shown by the solid curve in Figure 1. Note that the saturation vapor pressure rises rapidly with temperature. The second cloud consists only of ice crystals. If the ice crystals are neither growing nor sublimating (losing mass in the conversion from solid to vapor), the pressure exerted by the vapor surrounding the crystals has a very specific value: the saturation vapor pressure with respect to ice esi. It also varies only with temperature (dashed curve in Figure 1), but it is always less than es. Figure 2 shows the difference between saturation vapor pressures for liquid and solid H2O (esesi). The difference peaks at about −12°C (10°F).

Caption: Figure 1. Saturation vapor pressure (millibars) and its dependence upon temperature (°C). The solid curve is for liquid water; the dashed curve is for ice. Figures 1 and 2 are from Atmospheric Thermodynamics by Craig F. Bohren and Bruce A. Albrecht.

Caption: Figure 1. Saturation vapor pressure (millibars) and its dependence upon temperature (°C). The solid curve is for liquid water; the dashed curve is for ice. Figures 1 and 2 are from Atmospheric Thermodynamics by Craig F. Bohren and Bruce A. Albrecht.

Caption: Figure 2. The difference between the saturation vapor pressures for water and ice (es – esi) in millibars for a range of temperatures below freezing.

Caption: Figure 2. The difference between the saturation vapor pressures for water and ice (es – esi) in millibars for a range of temperatures below freezing.

Unless the temperature is very low [−40°C (−40°F) or lower], there is a good chance that liquid water droplets are present in subfreezing clouds. In fact, at temperatures between 0° and −5°C (32°F and 23°F), these supercooled droplets are often the dominant species in clouds. But what happens when supercooled droplets and ice crystals mix in the same cloud? If the actual water vapor pressure e in the cloud lies between esi and es, there will be a vapor pressure difference between the droplets and the ice crystals. Water molecules will migrate toward lower pressure, leaving the surface of the droplet and depositing themselves on the surface of the ice crystal. In other words, the crystals will grow at the expense of the droplets. If the vapor supply is great enough (e > es), then both species can grow, but the crystals will grow faster. As the crystals grow, their fall speed exceeds that of cloud droplets. If a crystal collides with a supercooled droplet, it freezes to the crystal in a process called riming—another way to deplete cloud liquid.

Snow Crystal Type Depends upon Temperature and the Vapor Supply

Beginning in the 1930s, the Japanese physicist Ukichiro Nakaya learned how to grow snow crystals in his laboratory under controlled conditions of temperature and vapor pressure. A modern representation of his findings appears in Figure 3, taken from a little handbook I highly recommend, Ken Libbrecht's Field Guide to Snowflakes (Voyager Press, 2006). The horizontal axis represents the temperature within the cloud where snow crystals form. The vertical axis represents the supersaturation, namely, the amount of water vapor in grams per cubic meter (g/m3) in excess of that required for saturation with respect to ice. In other words, when the supersaturation is zero, the vapor pressure is esi. Along the curve labeled “water saturation,” the vapor pressure is es. As noted before, the greatest difference between es and esi occurs near −12°C. In Figure 3, water saturation is defined in terms of density, not pressure, which has the effect of shifting the peak of the water saturation curve to a slightly lower temperature, about −14°C. At low supersaturation, small, solid, plate-like crystals can form at all temperatures. Solid prisms also form near −7°C. Crystal shapes remain fairly simple until supersaturation extends above the water saturation curve. Then ice needles form, and all kinds of intricately branched crystals. The largest filigreed crystals (dendrites) form at high supersaturation and at temperatures near −15°C. Why crystal types follow these “rules” is still mostly a mystery and the subject of research.

The trajectory of a snowflake through a cloud determines how it will grow. As Figure 3 aptly demonstrates, small changes in temperature and vapor supply can lead to huge changes in growth habit. In a deep cloud, a growing crystal may wander through many sectors of the morphology diagram. Perhaps it is not surprising that no two snowflakes are exactly alike, because no two have identical trajectories.

Caption: Figure 3. Snow crystal habit as a function of temperature and the supply of water vapor in clouds where the crystals grow. Supersaturation (grams of water vapor per cubic meter) is the amount of water vapor in the air in excess of that required to sustain a snow crystal in equilibrium.

Caption: Figure 3. Snow crystal habit as a function of temperature and the supply of water vapor in clouds where the crystals grow. Supersaturation (grams of water vapor per cubic meter) is the amount of water vapor in the air in excess of that required to sustain a snow crystal in equilibrium.

The six-sidedness of snow crystals is a result of the structure of water molecules and how they bind together in a solid. The six sectors or arms of a single crystal are very similar because they all experience the same conditions of growth.

Examining New-Fallen Snow

The next time it snows, take a good hand-magnifying glass and look closely at crystals collected on some black felt. Few of them are perfectly formed. Defects in crystal growth can be caused by the incorporation of microscopic particles in the air, by the capture of tiny ice fragments as the crystal grows, by fragmentation of crystals that collide, by sublimation, or by partial melting. Crystals that stick together to form flakes often show irregular growth because of the competition for vapor among so many entangled arms and edges. Large flakes can easily include over 100 individual crystals.


As mentioned above, when ice crystals fall relative to supercooled cloud droplets, the crystals collect the droplets, which freeze to the crystals upon contact. Supercooled drops will slowly evaporate in a cloud unless the vapor pressure is at least es. On the other hand, if the vapor pressure is much greater than es and the temperature is not too low, then droplets are both plentiful and growing, and the frequency of collisions between crystals and droplets is high. When riming is light, the collected droplets appear as tiny dots on the crystal surfaces. As the frozen droplets pile up, the crystal surfaces are covered up. With still more riming, the crystals are no longer recognizable; they become little white balls of low-density ice, perhaps a few millimeters across, called graupel or snow pellets.

Graupel is considered a type of snow. It is always present in thunderclouds and can provide embryos for the formation of hail. Graupel can collect so much supercooled water in thunderclouds that its entire surface becomes coated with liquid, which may not freeze immediately. Hail is usually larger than graupel and has higher density. Large hail often has concentric shells, some of them consisting of clear ice. Graupel contains a lot of air and reminds one more of Styrofoam than a mini ice cube.

You can find out much more about snow crystals and snowflakes at

Weatherwise Contributing Editor THOMAS SCHLATTER is a retired meteorologist and volunteer at NOAA's Earth System Research Laboratory in Boulder, Colorado. Submit queries to the author Attn: Weatherwise; Taylor & Francis LLC; 325 Chestnut Street, Suite 800; Philadelphia, PA 19106; or by e-mail to

In this Issue

On this Topic

Privacy Policy

© 2018 Taylor & Francis Group · 530 Walnut Street, Suite 850, Philadelphia, PA · 19106