Microphysics Review

Wintertime Cloud Microphysics Review

Dan Baumgardt
Science Operations Officer
National Weather Service, La Crosse WI

Last Updated: October 1999


This training was designed to give an overview of how ice and liquid particles both are created and interact in the atmosphere and more importantly how this influences operational forecasting. Although many forecasters do not consider cloud microphysics in the operational forecast setting, it has very many applications. This particular review is designed for wintertime forecasting in particular and discusses microphysics' relationship to precipitation type expected at the surface. The module begins with how clouds initiate ice from liquid drops via cloud condensation nuclei and works into critical temperatures for certain precipitation types. It then progresses through hydrometeor growth, operational aids, and finally case examples.

The operational forecaster should come away with a better grasp of precipitation type expected and possible heavy snow microphysics to look for in both observed and model data. We begin with the initial stage of ice growth within a supercooled cloud (T<0C).

Ice Nucleation

* Ice Nuclei - What are they?

First and foremost, cloud condensation nuclei (CCN) are particles suspended in the air which support the growth of cloud droplets or ice on their surface. Of all of the CCN floating in the air, a low percentage act as ice forming nuclei (IN) that have the ability to act as a surface for ice growth to initiate (from water in the vapor or liquid phase). Lets take a simple case of a solo cloud in the sky with no others around it. Also, we assume the cloud has a temperature of <0C and is composed of all supercooled liquid drops and water vapor (NO ICE). The only way to have ice form is by growing it on an IN particle's surface. [Again this assumes no ice comes from anywhere outside the cloud - which can happen (known as the seeder-feeder mechanism)]. Once the ice growth begins on the nuclei, the IN particle is said to be activated.

Since not all CCN particles are IN, or said another way - not all CCNs promote the growth of ice themselves - there must be something special about them. This "something special" is their chemical makeup. Water changing phase to grow as ice in a cloud is very particular as to the chemical composition of the particle on which it would like to grow initially.

It also depends on the relative humidity and temperature of the cloud. CCN particles have a better chance of being an IN as the temperature decreases and the relative humidity increases. In fact, no IN's can be activated (or have ice begin to grow on them) above the temperature of -4C even if the cloud is supersaturated (relative humidity > 100%).

For example, assuming we grow a cloud at temperatures warmer than -4C, liquid droplets may form on certain CCN. However, even if the cloud contains particles which could act as IN, and the air is supersaturated, ice cannot form on the IN particle if the temperature is warmer than -4C. The IN cannot be activated. Note: this is not to say that if an ice crystal is introduced into the cloud, that it will not grow. This case assumes no ice crystals are present in the cloud or dropped into the cloud; they simply did not exist.

* What type of particles act as IN?

Of all particles in the air, 5-10% are considered CCN - or particles available for either water or ice to grow in a cloud. Of that 5-10% of all particles which are CCN, another small percentage of those act as IN. IN are simply a more "selective" group of the larger CCN pool. A look at the larger CCN family shows that CCN are simply particles floating in the air and consist of soils, sand, sea salt, volcanic debris, and also particles emitted from urban factories. The number of CCN vary greatly from place to place dependent on urbanization and continental versus sea locations. CCN also generally decrease rapidly from the surface of the earth and then maximize again in the lower stratosphere.

Work done by various cloud physicists suggests that about 80-90% of all IN over the upper Midwest consist of some type of clay material with vermiculite leading the charge. In these studies, it was found that melted snow crystals typically contained one solid IN particle in the center of the snow crystal. IN activation, or the initiation of ice on a particle in a supercooled liquid cloud, occurs at different temperatures for IN particles of different material. Here are some of the common IN and their activation temperatures:

  • Silver Iodide (used for cloud seeding) -4C
  • Kaolinite (clay family) -9C
  • Volcanic Ash -13C
  • Vermiculite (most prevalent clay) -15C

For example, ice may form on a particle of salt at a temperature of -8C, while on a clay particle it occurs at -15C. This is important because if a majority of the IN are clay over Wisconsin and soundings reveal cloud temperatures of -8C or warmer, no ice will occur and freezing rain/drizzle can be expected (in theory). For our concerns, we will use clay references when looking at temperatures and relative humidities that will produce ice in a cloud.

* Heterogeneous Nucleation - How it works

In the previous section, we discussed IN being any particles in the atmosphere which provide the surface for water to begin ice growth onto from the liquid or vapor form. These are mostly clay particles in the Midwest. Now, we will take the next step to understand exactly how these nuclei initiate an ice crystal. There are basically three modes of ice crystal initiation or activation:

  • Growth by deposition - Water vapor condenses onto an IN particle and freezes.
  • Growth by contact - IN initiates an ice crystal at the moment it contacts a supercooled water droplet.
  • Growth by freezing - IN floating within a supercooled droplet initiates the freezing.

These processes are known as heterogeneous nucleation: any liquid drop or ice forming on a nucleus. For example, the IN (which is a solid particle) and the supercooled droplet (which is a liquid) form a solid phase. It is this process by which ice crystals initiate.

[Homogeneous nucleation occurs if the precipitation particle is created via no ice nuclei. There is no surface for the drop or ice to initially form onto in this case. Ice homogeneously nucleates near -40C or F - which is much too cold to produce the large amount of ice found in most clouds at warmer temperatures.]

Research done by Koenig (1962) suggests that most of the ice nucleation in clouds occurs mostly by #1 above: deposition. This mode is preferred for a cloud with no ice initially present. Lets focus on temperatures needed for deposition growth and the ice to initially appear in the cloud. We now know clay particles are the most observed IN particles in the Midwest and also that depositional growth is the important initial ice forming growth mechanism. However, we also know that ice will not begin to grow on an any IN at warmer than -4C.

* Temperature - What values are critical for IN to introduce ice initially into the cloud?

"Since water readily supercools, particularly in small quantities, water clouds as well as fogs (which are clouds with ground contact sufficiently extensive to suppress vertical motions) are frequently found in the atmosphere at temperatures below 0C" (Pruppacher and Klett, 1978). On the figure below (Fig. 2-1, from Pruppacher and Klett, 1978), follow the blue line from -8C up to the black lines which represent the ice curves (help with figure) and then follow the green line over to the right side. Notice that this indicates about 40% of all clouds contain some amount of ice. Follow it up further and take the red line left at the water curves and you'll notice that nearly 60% of clouds are composed of ALL supercooled liquid water.

It is also apparent that temperatures warmer than -4C cannot produce cloud ice. The figure actually has dashed lines from 0 to -4C to indicate this principal. Operationally, this translates into clouds which do not have a temperature colder than -4C anywhere will HAVE NO ICE. Freezing precipitation will be observed if the saturated cloud layer extends to the surface with a minimum cloud temperature of 0 to -4C ALWAYS (provided the surface temperature is 0C or below).

Figure 2-1 suggests that the majority of clouds (over 50%) contain all supercooled water, and no ice, when warmer than about -10C. Simply move straight up on the figure from -10C to the water curves and then read the left side margin. At this same temperature, the amount of supercooled liquid water observed in clouds is decreasing rapidly. You can gain this from the slope of the water curves. Now, move up from -14C to the ice curve intersection, then read the right axis. Once -14C is reached, about 75% of all clouds contain some type of ice crystals. By -18C, you can wager the life savings that ice crystals are present in the cloud (Peppler, 1940; Borovikov et. al., 1963; Mossop et.al., 1970; and Morris and Braham, 1968).

Coming back to the earlier discussions of clay being the foremost observed IN in the Midwest, recall

  • Silver Iodide (used for cloud seeding) -4C
  • Kaolinite (clay family) -9C
  • Volcanic Ash -13C
  • Vermiculite (most prevalent clay) -15C

the studies in Figure 2-1 correspond well the activation temperatures above. At -9C, clay INs begin to activate ICE growth. In figure 2-1, about 50% of all clouds contain ice at -9C.

In yet another study by Schichtel (1988), a large RAOB dataset was diagnosed, comparing different environmental variables to the observed precipitation type. One of these variables compared the coldest observed temperature in a saturated layer (below 600 mb) to the precipitation type. Basically, in the figure below (Fig. 4.3.1 from Schichtel, 1988), we are looking at minimum cloud temperatures versus the observation of snow.

NOTE: although cases exist at temperatures warmer than -8C (2 of 55), the majority of observed snow cases had a minimum cloud temperature of -10C or colder.

From the research we have reviewed, here is a summary list of operationally significant temperatures for ice crystal initiation in a cloud:

-4C NO ICE in clouds of this temperature or warmer
-10C 60% chance ice is in the cloud - warm cutoff operationally
-12C 70% chance ICE is in the cloud
-15C 90% chance ICE is in the cloud
-20C ICE is there!

From this, come away with the idea that -12C to -14C should be looked at as a minimum range for a high chance of ice being a cloud. -10C is a very good cutoff point needed for some considerable chance of ice being introduced into the cloud via IN activation. If this temperature is not reached somewhere in the cloud (most likely the cloud top), it is likely that the cloud contains only supercooled water droplets. By -15C, you can bet you have ice introduced. Also, as the temperature decreases, IN become more easily activated and allow ice to grow on themselves...and thus a better chance of ice and more of it. If the cloud top extends very high into very cold temperatures, about -20C, expect a high amount of ice production within the cloud and effective seeding of supercooled liquid drops at lower levels of the cloud. In other words, a great ice producing cloud!

Sidelight: In research done by Schnell and Vali (1976) and Vali et. al. (1976), it was found that a particular bacteria produced by decaying leaf material could nucleate ice at temperatures warmer than -4C. In fact, it was found that at -1.3C ice was nucleated by the bacteria. However, it was found that its ability to nucleate ice was very volatile. It made me think, does this mean we should have less cases of freezing rain in fall given the same environment? Cloud ice would be produced at lower temperatures than in spring! Hmmm....

* Relative Humidity - What relative humidities are important?

Although the relative humidities of clouds and fogs usually remains close to 100%, research indicates that many clouds depart from this value. Measured values have been as high as 107% on the interior portion of clouds while some fogs have been measured near 81% near the exteriors. In any case, the amount of IN activated in clouds increases as the relative humidity increases. So simply lowering the temperature (which raises the relative humidity) will make the environment more preferred to form ice. That means in a cloud, the cooler regions (typically toward the tops) are a more preferred area for IN and supercooled liquid water to get together to form ice.

Growth of Ice Crystals


To this point, we have discussed cloud physics specifically focused on a saturated, supercooled liquid cloud and how ice is initially created through IN activation. Since mama Nature is more complex, we need to consider the introduction of ice crystals from outside that supercooled liquid cloud layer as well. Also, once the ice is made in the cloud, how does the ice grow and multiply?

* How Ice-Crystals Grow In a Supercooled Liquid Cloud

Growth by deposition - a big one! [also called diffusion deposition]

Growth by deposition, physically, is the change from water in the vapor form to water in the solid form (or water vapor to ice). This process is governed by the Bergeron-Findeisen Process which states that ice crystals will grow at the expense of liquid droplets in an environment where the relative humidity is 100% (or the environment is saturated with respect to (wrt) water). The saturation vapor pressure over ice is less than that of water and therefore the vapor will want to move toward the ice or ice nuclei versus a liquid drop in the same environment. Deposition occurs by water vapor depositing on the ice in a liquid form and immediately freezing, or directly depositing as a solid. Once this water vapor changes to a liquid/solid, the relative humidity of the surrounding air falls slightly below 100%, and more water drops can evaporate. This is the common process of the ice crystal growing at the expense of the water droplets.

Since the amount of supersaturation is important to this growth process, we need to look at it a bit more. In an environment that is saturated with respect to water (relative humidity=100%), by simply lowering the temperature, the environment with respect to the ice crystal becomes increasingly supersaturated. Figure 9.3 (from Rodgers and Yau) below shows this property: at -5C the environment is about 5% supersaturated wrt/ice, and by -20C it 's more than 20% supersaturated wrt/ice. Thus, the growth by diffusion deposition is greater at colder temperatures.

However, this ice growth mechanism also depends on pressure. In fact, changing the pressure caused the growth of snowflakes to be the maximized near -15C.

Operationally, key in on maximum vertical motion areas at the -15C temperature area for heavy snowfall (Auer and White, 1982). If the preferred snowflake growth is in an area of maximum lift, it goes bonkers!

Growth by Accretion

Commonly, this is the growth of an ice particle accomplished when it overtakes or captures supercooled liquid droplets. It follows that it should occur more readily after the ice-phase particle has grown to a sufficient size to begin to fall and collect the supercooled droplets. Thus, initially the ice grows via the diffusion method at the top or mid level of the cloud and later by this process.

The collection of liquid supercooled droplets is best for ice particles which fall the most rapid. These are graupel (which are really a collection of frozen drops), needles of snow, and finely dendritic or powder snows, in order of decreasing fall speeds.

While deposition dominates ice formation and growth in the upper to middle portion of the cloud, accretion dominates the lower portion of the cloud. This is the main growth mechanism for ice crystals - see riming below.

Growth by Aggregation

Aggregation is the coming together of multiple ice particles to form one main snowflake. Although not a great deal is known about this behavior due to many "unknown variables", the process is maximized when temperatures are warmer than -10C. This allows for effective sticking and refreezing of ice crystals. In figure 14-19 below (from Pruppacher and Klett, 1978), you can see that the largest snowflakes occur when temperatures are near 0C (red box). Thus, cloud layers which have extended regions where the wet-bulb temperature is near 0C near the surface will produce larger flakes. Remember this for accumulation amounts.

* Other Types of Ice Growth and Multiplication

Seeder-Feeder Mechanism

Up to this point, the discussion has been mostly about all liquid clouds initiating ice and subsequent ice growth. This happens when the cloud is deep or cold. However, during transition seasons, it is not uncommon to have various temperature regimes and cloud layers which are not conducive to this internal ice production.

One complexity is the growth of ice crystals via the seeder-feeder mechanism. A cloud with ICE crystals (seeder) at a given temperature (e.g., -14C) moves over the top of a lower, supercooled liquid cloud (e.g., -6C) and precipitates ice down into that cloud (feeder). If the supercooled cloud is supersaturated wrt/ice (which we will assume all non-dissipating liquid clouds are), then the ice crystals will grow in the lower, warmer cloud. Thus, we have "seeded" the supercooled liquid cloud layer with ice and it now will produce snowflakes and ice!

The limit to the seeder-feeder mechanism seems to be about 5000ft or 1500m. That is to say that the seeder (ice) cloud must be not more than 5000ft above the supercooled liquid cloud (feeder). Figure 13-29 from Pruppacher and Klett illustrates a sampled environment where ice crystals were present in varying relative humidity environments.

The figure shows a few important features of ice crystals falling through and below the cloud:

  • The larger the crystal, the further it will survive - especially in an RH environment < 70% (Compare 1A versus 4A).
  • The crystals fall over 2 km in saturated conditions, but after reaching RH<70% around 500 mb, the crystals rapidly disappear within about 1 km taking at most 20 minutes.
  • The smaller crystals actually died within a more saturated environment.

The authors state: "...ice crystals falling from cirrus clouds can survive fall distances of up to 2 km (about 6500 feet) when the relative humidity is below 70% in a typical midlatitude atmosphere."

At greater distances than about 5000 ft between two cloud layers, ice falling from the seeder will tend to evaporate/sublime before reaching the feeder cloud.

Distances between the layers can be tough to assess in realtime away from sounding times and locations. Operational forecasters may want to use IR2 GOES imagery to assess water versus ice cloud tops and the WSR-88D VWP profile to assess the distance between cloud layers. On the VWP, wind barbs should be evident in both the seeder and feeder layer with the dry layer between them showing no barbs. This gap is measurable.

Operationally, this means your great forecast of freezing rain can depart to snow for a period of time while a higher level cloud deck moves over the top of the feeder layer. Check the satellite imagery to help this diagnosis and try to address it in the nowcast. Remember GARP gives cloud top temperatures by placing the cursor at a given location on IR imagery as well.

Tough, yes. However, knowing WHY a particular weather phenomena is occurring is very valuable to the forecast process.

Multiplication by Rime Splintering

This is a form of secondary growth of ice crystals and snow which is very similar to accretion growth. The ice particles collect the supercooled drop (via accretion), and the liquid begins to freeze. If the particle is of a certain size, temperatures near -5C cause the supercooled drop to freeze from the outside edge of the liquid surface inward, it expands, and the frozen shell then shatters. This splintering produces many more ice nuclei for the collection of supercooled droplets! Many orders of magnitude of growth in snowflakes and ice occur near -5C. The specific size of the ice particles needed for this to occur are normally in stratus clouds. This process can completely glaciate a cloud in 45-90 minutes.


{paraphrased from Rodgers and Yau, 1988} Precipitation from mid-latitude stratiform clouds is thought to develop chiefly by the ice-crystal process. These clouds have relatively low liquid water content. If clouds persist at altitudes where the temperature is at least -10C but more toward -15C, the ice-crystal process can lead to precipitation. Each level in a stratiform cloud has a certain job. The cold, upper levels (T of -15 to -20C) supply ice (via depositional growth) that serve for precipitation development at lower levels. The mid-level cloud area (T<-15C) provides the right environment for rapid depositional growth of ice as well (maximized near -15C). Aggregation, rime splintering, and accretion proceed most rapidly at the lower levels of the cloud at temperatures between 0C and -10C. Recall aggregation is maximized near 0C. The lowest levels dominate the particle growth through the splintering process and accretion because fall velocities increase to nearly 8 m/s.

Operational Application

A good amount of the previous information contained theory, observational studies, as well as operational tools to improve forecasts. This section will try to summarize and suggest techniques in forecasting precipitation type from a microphysical approach. The first section will discuss a few more needed microphysical tools, the second will be the actual "top-down" precipitation type approach, followed by forecast challenges, and finally a section on data sources.

* Hydrometeor Altering Environments

Wet-bulb Temperature and Unsaturated Layers

Wet-bulb temperature is VERY useful in helping to assess precipitation type because as a hydrometeor falls, it will possibly melt, refreeze, or evaporate/sublime. When a hydrometeor comes into an unsaturated environment, it will begin to evaporate if it is liquid water or sublime (solid to vapor) if it is frozen. It can also melt if T>0C. In all of these cases, energy is going FROM the environment TO the hydrometeor to change the phase. This LOWERS the temperature of the surrounding environment. Water in the vapor form is also being added to the surrounding environment and thus the dewpoint (frost point if below 0C) RISES. Thus, the temperature cools and the dewpoint rises to approach the WET-BULB temperature.

In areas of cloud separated from the surface, some precipitates may fall through surface-based dry layers to reach the ground. It is not uncommon for these "dry layers" to exist aloft as well - between two saturated cloud layers. When precipitation falls through the dry, unsaturated layers, the ice/liquid begins to sublime/evaporate. The environmental temperature will begin to cool through evaporation and the dewpoint will rise...both approaching the wet-bulb temperature (Tw).

Application:If frozen particles fall into a layer with i.e., Tw=-1C and T=5C Td=-5C, the particles will begin to melt/evaporate/sublime thus the temperature will begin to decrease. One study (Penn, 1957) suggests this cooling of the environment via evaporation and melting (diabatic processes) can be on the order of 5 to 10C per HOUR. Once the layer is saturated with vapor and the T=Tw=Td are at -1C, the frozen particles will remain frozen through the layer. Thus, a partially melted precipitation type (FZ first then PE) could occur for maybe an hour and then snow will continue after the layer is saturated and cooler (Tw=T=Td=-1C).

For surface based dry layers, examine Tw and their profiles to at least 850 mb. If the Tw is below zero and the precipitation is entering the dry layer as ice, expect a possible initial melting and a mix to be observed. Once saturated, no melting will occur and the precipitation will be in the form of snow. Rain is more likely when the Tw>0C and the maximum surface Tw for snow is about 34F or 1.5C (Keeter, 1992 and Schichtel 198x).

If the precipitation enters the surface based dry layer as a liquid initially, and T>0C it will likely be rain. Then after some time passes the environment will be saturated with Tw<0C and it may be freezing precipitation depending on surface temperature. If the Tw is -10C or lower in the surface layer, remember ice nucleation can occur again to create snow/mixed precipitation.

Confused yet? Dont worry - we will organize all these thoughts.

Warm Layers or Inversions

Typically, precipitation type forecast problems occur in environments with very cold air near the surface with a sharp frontal inversion aloft where a tropical air mass is being transported into the area. This environment is significant to precipitation type if the warm layer becomes warmer than 0C given a surface temperature below or near 0C.

In this situation, two basic items must be assessed by the operational forecaster: 1.) If ice is introduced into the top of the warm layer and 2.) What the warm layer temperature regime will be at a given time (tough!).

  1. Is ice introduced into the warm layer? Using the previously discussed techniques, the forecaster should look for the minimum temperature of the saturated cloud above the warm layer. If the temperature reaches at least -10C, the chance for ice exists. If it is colder than -15C, ice will probably be introduced, and if -20C or lower exists, ice will be introduced.
  2. Warm layer temperature? The table below illustrates the behavior of the warm layer temperatures and the expected surface precipitation from Stewart (1985).

Assuming the surface temperature is less than or equal to 0C:

Warm Layer Max Temperature

Precipitation Type
with ice introduced

Precipitation Type
without ice introduced


Snow Freezing Rain/Drizzle

1C to 3C

Mix(1C) to Sleet(3C) Freezing Rain/Drizzle

> 3C

Freezing Rain/Drizzle Freezing Rain/Drizzle

Basically, if an ice crystal enters:
  • a warm layer with a temperature below 1C, the snow/ice will not melt enough to give a precipitation type change.
  • warm layer with temperatures between 1C to 3C, the snow/ice will melt but not completely. Thus, a liquid drop with an ice particle will be descending. Due to the presence of the ice particle, the liquid will immediately refreeze in the cold surface layer resulting in ice pellets (near 3C) or a mix of ice pellets, snow, freezing rain, or rain (1 or 2C).
  • a warm layer with a temperature of greater than 3C, the snowflakes will completely melt to liquid and result in rain or freezing rain if surface temperatures are at or below 32F.

If no ice crystals were present, the process is dominated by the warm rain process unless somehow seeded from above. This is collision-coalescence (coll-coal) of cloud droplets. Thus, when no ice is present, drizzle, rain or freezing rain (surface temperature at or below 0C) occurs.

The depth of the warm layer does play a role in determining how much melting will occur if snow falls into the warm layer. So does the depth of the cold air near the surface! However, the research is geared toward the temperature min or max in the layer more than the depth. Later, we will discuss a recent paper that uses depth.

* Top-Down Precipitation Type Forecast Method

Top-Down Operational Flow Chart - jpg format

Let's bring all this full circle and organize your thoughts. As we have seen through the previous discussion, the introduction of ice into the environment is a key forecasting issue. Cloud temperatures are also very important to the outcome of precipitation type observed at the surface. Remembering a few key ideas and numbers will help you assess RAOBs and model forecast soundings for your area of interest to hopefully gain a better forecast to the public.

The top-down approach starts at the top of the environmental sounding and works toward the surface tracing an actual hydrometeor trajectory through either an observed or forecasted environmental temperature and moisture profile.

Typically in a precipitation type event, three basic critical regions exist in the environment (from top-down):

  1. Cooler mid level air mass (ICE PRODUCING LAYER)
  2. Elevated warm tropical air mass (WARM LAYER:MELTING!)
  3. Surface-based arctic air mass (COLD SURFACE LAYER:REFREEZING/CONTACT)

In the diagram below, these layers are critical to assess from the top-down and each have a job in determining precipitation type.


Will ice be introduced?

Beginning at mid-levels and working down, determine where the first cloud is present and what is the coldest temperature of the cloud. Recall the following criteria:

  • -4C NO ICE in clouds of this temperature or warmer
  • -10C 60% chance ice is in the cloud ( warm cutoff for ice operationally)
  • -12C 70% chance ICE is in the cloud
  • -15C 90% chance ICE is in the cloud
  • -20C ICE is there!

IF ICE: Ice now descending.

IF NO ICE: Supercooled liquid drops if at or below 0C.

Consider unsaturated layers during descent:

  • Do they evaporate totally?
  • Do they cool layer to Tw?
  • Consider Seeder-Feeder clouds.
  • Consider Convection to develop?

CLOUD MID IN LEVEL (about 700-800 mb)

Is there a warm layer? What will occur in that layer?


  • Warm Layer Max Temp <1C:Remains ICE/SNOW.
  • Warm Layer Max Temp 1 to 3C:Partial Melt - Now ICE and WATER present.
  • Warm Layer Max Temp >3C:Total Melt: Now LIQUID WATER


  • Supercooled drops and above freezing drops of liquid exist.
  • Consider unsaturated layers during descent - Tw=0C?


Descending through surface cold layer.


  • Tw near to 0C or less to surface:SNOW.
  • Tw above zero for any time and surface Tw>33F: RAIN POSSIBLE.


  • If Tw or T is 0C or less, the particle will immediately refreeze to form ICE PELLETS at the surface.
  • Tw near OC or above and surface Tw>33F: RAIN POSSIBLE.


  • Minimum cold layer temperature warmer than -10C and surface T=0C or less:FREEZING RAIN/DRIZZLE.
  • Minimum cold layer temperature warmer than -10C and surface T warmer than 0C:RAIN/DRIZZLE.
  • Minimum cold layer temperature < -10C and surface Tw<34F:SNOW AND SLEET(why?)POSSIBLE.
  • Consider unsaturated layers during descent - Tw=0C ?

The Tau Technique - Czys et. al. 1996

In the December 1996 Weather and Forecasting quarterly, a paper discussed a technique to identify areas of freezing rain versus ice pellets in the local area. It consists of two basic ideas: 1. The mean warm layer temperature and 2. The depth of the warm layer. The authors used microphysics and fall speeds to calculate the time needed for an ice crystal to melt in the warm layer given a mean temperature of the layer. They came up with the following diagram:

Operationally, this can be pretty easy to calculate for the upper midwest in a matter of about 20 minutes by looking at local RAOBS or model forecast soundings. For a RAOB (2 minutes!), simply do the following:

  1. Identify the depth of the warm layer (depth above 0C).
  2. Identify the mean temperature of the warm layer (equal area technique on a SKEWT).
  3. Find the coordinate on the ISONOMOGRAM above. Then, if the coordinate is to the right (left) of the blue line you will have freezing rain (ice pellets).

Operational Forecast Problems

This section discusses a few of the "daggers" which can throw your microphysical analysis into jeopardy. Remember that knowing WHY something is occurring is just as important as the forecast. The top-down approach is good, however, these points need to be kept in mind when forecasting or diagnosing precipitation type.

  • Terrestrial Temperature

    Although supercooled drizzle or rain may be predicted, these will not freeze on contact with objects at the surface ALWAYS when the temperature is 0C or below. For significant icing: the problem that needs to be considered is the temperature of the terrestrial or surface objects. Ask yourself: Have we been significantly below or above freezing for an extended period? What has been the mean temperature over the past 3 days? The first cold storm of the fall season is typically a tough beast to produce significant icing as the terrestrial object temperatures will more than likely be above freezing. How much above depends on recent conditions. The opposite (easier icing) is true in spring or for any unusually cold period followed by a warm storm.

    One other effect in this area is the urban heat island. Regions of significant urbanization (La Crosse or larger) will emit heat to locally warm the air a few degrees. Significant icing may be predicted from the top-down approach, but heat island effects can keep icing to a minimum in the local city area.

  • Convection

    Convection, usually elevated convection with inflow at and above the top of the warm layer inversion, can cause significant problems to a precipitation type forecast. For example, soundings with saturated conditions where the minimum cloud temperature is not -10 or cooler should produce supercooled liquid. No warm layer exists. If the surface T is 0C or less, FREEZING DRIZZLE would be likely. Now, let us place a elevated thunderstorm into the environment. The storm produces ice crystals, no warm layer exists, so it SEEDS the entire supercooled liquid layer and convectively induced SNOWS occur. Then, the convection moves north of the area, and FREEZING DRIZZLE occurs once again. Thus, convection can SEED a supercooled layer.

    In ANY case, treat convection as an ice producer in the top-down approach. IF a warm layer exists, assume ice will be put into that warm layer. The convection may also REMOVE the warm layer through cooling via massive evaporation and melting. This is where warm advection should be assessed.

  • Advections

    One very important point to the warm layer's existence is the strength of the warm advection. After all, it is the warm advection which creates (or advects) the warm layer in the first place. When precipitation is falling through the layer, we have already mentioned that the melting/evaporation can cool the layer 5-10C per hour. The amount of cooling in the layer depends on the wet-bulb temperature in the layer, the amount of warm advection to counteract the cooling, and vertical motions to act with the cooling.

    Strong vertical motion induced through warm advection will cause the warm layer to cool. The warm air is not actually warming the environment but rather forcing vertical motion and the response is adiabatic cooling.

    If no vertical motion is occurring, then the warm advection must continue to be significant for a prolonged change of phase (melting) to occur. Remember ice storms need extended periods of the freezing rain environment. Warm advection must counteract the diabatic effects of cooling.

    The warm layer wet-bulb temperature can be diagnosed. If Tw in the warm layer is >1C, the forecaster should be especially aware that melting in the layer may be prolonged. The temperature will cool down to the Tw with precipitation onset (into that layer) and it will remain there as long as hydrometeors exist. If the layer becomes free of hydrometeors, and strong warm advection continues, the temperature will rise again!

  • Model Forecast Soundings

    Although it is many times the best guidance we have to forecast precipitation type, the models can be in error. Remember in the top-down approach that simply injecting a cloud or saturated layer at temperatures cooler than -10C can initiate your analysis with ice. So, pay close attention to the initial 00hr analysis to see how well the model began with moisture profiles. Model resolution is mostly the problem here as some mixed precipitation type regions can be very narrow (50 miles/90 kms) yet very long (200 miles). Especially in the short term models (RUC), check the initial analysis versus satellite imagery or current observations.

  • Drizzle versus Rain

    In cases where ice is not introduced into the cloud and the temperature is between 0 and -10C through the surface, freezing drizzle is dominant. It is drizzle because the warm-rain process is occurring - growth of droplets by collision-coalescence. Also, these events are usually dominated by a strong inversion with the moisture and clouds located below the inversion top. What would turn this to rain?

    Although hard to elicit a good amount of vertical motion below/in an inversion when saturated, geostrophic warm advection (winds veering with height) through the layer would cause some rising motion (via QG Omega equation). This may cause a response of increased collision-coalescence and bigger drops...giving a rain. Operationally, the key to rain versus drizzle is just vertical motion...the more you have, the more likely that the process will be rain. One key may be to look at the sounding - if the saturation quickly stops at the inversion and dries out, drizzle is likely. If saturated above the inversion, isentropic lifting/warm advection and elevated convection are all possible lifting mechanisms that could occur to enhance the warm cloud coll-coal process. Convection can also introduce ice crystals into the warm layer!

  • Unsaturated Layers

    Recall previous discussion on these layers.

  • Seeder-Feeder Mechanism

    Recall previous discussion.

Datasets to Help Assess Precipitation Type
    • Surrounding Sites Observed RAOB data - SHARP/NSHARP
    • Use the Tau Technique also
    • Hourly bufr sounding data from ETA/NGM/MESOETA
    • Includes Model Ptype, Inversions, RH, Tw, Raw Data for 2m temp
      wind profile, and max temp of warm layer
    • Assess cloud top temperatures with cursor in IR imagery
    • Fog/reflectivity and IR2 product for ice/water clouds
  • FWC and MAV Guidance
    • MAV CPOS (Conditional prob. of snow)
  • WSR-88D
    • Assess the bright band/freezing level.
    • Use VWP to assess inversion height and seeder/feeder cloud layer separation distance.
    • Use to assess convective bands which may move into the freezing drizze environment and SEED it with ice.

    AWIPS Era

    • Forecast soundings via RUC, NGM, MesoEta, Eta, AVN, MRF
    • Analysis soundings via LAPS (Local Domain) and Hourly RUC (CONUS)
  • Plan View - Volume Brower
    • Snow Accumulator Forecasts - ALL Models - no idea on the background science here.
    • LAPS Precipitation Type Analysis - "Other" in Volume Browser
      • Precip Type - Surface
      • Cloud Ice, Cloud Liquid, Hydrometeor Concentration: Surface-Tropopause
AWIPS Satellite Color Curve To Assist Ice Assessment

The table below gives an example of one IR channel color scheme to higlight the liquid to ice temperature range.  Based on cloud top temperature, and the color regimes below, you could operational assess via IR if ice is in the cloud via cloud-top sampling. This is best for freezing drizzle environments with low stratus which may have areas of more vertical development.  Also, it can assist in seeder-feeder assessment.  This is a tool to assess why a precipitation type may be occuring at a given point, which is half of the forecast battle.

Enhancement Color Temperature Deg C Ice or Liquid? AWIPS RGB numbers
Blue -8C Likely Liquid 0     0     200
Light Blue -10C 60% Chance Ice is there 0   100   150
White -12C 70% Chance Ice is there 255  255  255
Pink -15C 90% Chance Ice is there 250  0 160 (-15C) to 250  80  170 (-20C) : Interpolated
Black to White -20C or less ICE is there! 0  0  0 (-20C) to 255  255  255 (coldest temp): Interpolated


    Practice CASES


    A. Cloud layer relative humidity near 100%, minimum temperature of -7C in cloud, entire layer temperature below 0C including the surface. What would you expect?

    In this case, no IN is activated as temperatures are not cold enough and therefore the cloud is supercooled liquid drops. Thus, this would produce a freezing drizzle case or if the vertical motion was strong, freezing rain. Here is an actual case...freezing rain was reported example.

    B. Cloud layer relative humidity 100%, minimum temperature of -17C at cloud top, entire layer below -4C including surface. What would you expect?

    This case has IN activated near the cloud top because the temperature is quite cold (colder than the -12C to -15C range). Recall since IN are activated easier with decreasing temperature ice is likely. Thus, ice will be created near cloud top and begin to fall toward the surface. Since temperatures are below 0C through the surface and ice is already made aloft, snow is observed at the surface. The lower temperature of -4C does not matter so much (still OC or less) except that aggregation may not be a big factor - usually favored more toward 0C - thus smaller snowflakes. If vertical velocities are strong near -15C and/or -5C, expect heavier snowfall rates. This case did not include any low-mid level warm layer inversion.

    C. Cloud layer A and B separated by 4000 ft, A is the lowest. What would you expect?

    The top cloud layer produces ice, and then seeds the lower layer cloud top to initiate ice in that cloud layer. If further apart, the ice crystals maybe evaporate before reaching the lower cloud top. In any case, the environment would produce snow here also. Thinking operationally, if freezing rain is observed with just case A, and then another cloud deck moves over it with ice involved, a precipitation type change to snow may occur.

    D. Layer:

    • 400-700 mb: relative humidity near 100%, minimum temperature of -13C in cloud.
    • 700-850 mb: warm layer, RH=50%, T=7C Tw=2C Td=-6C, 4500ft thick.
    • 850 mb - Surface: cool layer, RH=100%, Tw=-4C.
    • Weak warm advection, strong vertical motion, non-city area, elevated convection possible.

    What precipitation would you expect? Does elevated convection matter?

    The top cloud layer produces ice at -13C. Thus, ice enters the warm layer with precipitation onset and begins to melt. It may evaporate completely! For a short time it will melt to completely liquid (warm layer max temp >3C). During this period, it will fall through the surface cold layer and NOT refreeze because no ice is present and the lowest temp is only -4C. FREEZING RAIN. However, after about one hour with little warm advection, the T=Tw (2C) which results in only partial melting. The particles will then refreeze to sleet in the -4C layer near the surface. Thus, SLEET.

    Convection would only act to enhance precipitation rates as ice is already introduced into the warm layer by the pre-convective cloud.

Sounding Problems

Sounding Number




? ?


? ?


? ?


? ?


? ?


? ?


? ?
Seminar Soundings
Observed Precipitation
Observed Precipitation
Observed Precipitation
Observed Precipitation
Observed Precipitation
Observed Precipitation

Cross-section problems

Cross-section Solution
Cross-section w/Convection at "X"




Auer, A.H. Jr. and J.M. White, 1982: The Combined Role of Kinematics, Thermodynamics, and Cloud Physics Associated with Heavy Snowfall Episodes. J. Meteor. Soc. Japan, 60, pp 500-507.

Borovikov, A.M., 1963: Cloud Physics, p65. Transl. by Isreal Program f. Scientific Translation, U.S. Dept of Commerce, Wash. D.C.

Czys, R.R., R.W. Scott, K.C. Tang, R.W. Przybylinski, and M.E. Sabones: A Physically Based, Nondimensional Parameters for Discriminating between Locations of Freezing Rain and Ice Pellets, Weather and Forecasting, Volume 11 1996, pp 591-597.

Lussky, G., 1995: Physical Rational Associated with Various Minneapolis, Minnesota Heavy Snow Checklist Parameters. NOAA/NWS 4th National Winter Weather Workshop, Kansas City, MO. pp. 33_1 to 33_7.

McNulty, R.P.: Winter Precipitation Type. NOAA Technical Attachment 88-4: NWS Central Region Scientific Services Division, 1988.

Morris, T.R., and Braham, 1968: Proc. Weather Modif. Conference, Albany, NY, April 1968, p306, Am. Meteor. Soc., Boston, MA.

Mossop, S.C., 1970: Bull. Am. Meteor. Soc. 51, 474.

NOAA Technical Attachment 73-25: Forecasting Ice Storms, NWS Central Region Scientific Services Division, 1973.

NOAA Technical Attachment 71-21: Forecasting Freezing Rain, NWS Central Region Scientific Services Division, 1971.

Penn, S., 1957: The prediction of snow versus rain. Forecasting Guide No. 2, U.S. Weather Bureau, 29 pp.

Peppler, W., 1940:Forschg. u. Erfahrung. Reichsamf f. Wetterdienst., B., No. 1.

Pruppacher, H.R., and J.D. Klett: Microphysics of Clouds and Precipitation. D. Reidel Publishing Company, Boston, 714 pp.

Rauber, R.M., M.K. Ramamurthy, and A. Tokay, 1994: Synoptic and Mesoscale Structure of a Severe Freezing Rain Event: The St. Valentines Day Ice Storm. Weather and Forecasting, Volume 9, 1994, pp 183-208.

Rodgers, R.R., and M.K. Yau: A Short Course in Cloud Physics. Pergammon Press, New York, 293 pp.

Schichtel, M., 1988: Specification of Precipitation Type in Oklahoma Winter Storms. University of Oklahoma. Masters Thesis. 131 pp.

Schnell, R.C., and G. Vali.:Biogenic ice nuclei: Part I. Terrestrial and Marine Sources. J. Atmo. Sci., Volume 33, 1976, 1554-1564.

Stewart, R.E. 1985: Precipitation Types in Winter Storms. Pure Appl. Geophys., 123, 597-609.

Stewart, R.E. 1987: Nowcasting Rain/snow Transitions and Freezing Rain. Proc. Symp. Mesoscale Anal. and Fcstg., Vancouver, pp. 155-161.

Stewart, R.E., and P. King, 1987: Freezing Precipitation in Winter Storms. Mon. Wea. Rev., 115, pp. 1270-1279.

Vali, G.: The Origin and Concentration of Ice Crystals in Clouds. Bull. Amer. Meteor Soc., 66, 1985, pp. 264-273.

USA.gov is the U.S. government's official web portal to all federal, state and local government web resources and services.