Internet Thunder
by Rod Gonski
National Weather Service, Raleigh, NC


The study of meteorology has posed significant challenges to teachers at the pre-college level. One of the biggest challenges has been in simply obtaining the data to study weather as it is happening, or as weather forecasters might say, in real time.

Today, with increasing access to the Internet and an ever-faster flow of electronic information, meteorological data is becoming available to schools as it never has before. Meteorology can finally come alive in the classroom! Our growing challenge now is to sort through the overabundance of information and to use real time weather data to achieve a greater understanding of meteorology.

To understand weather, we need upper air data. Specifically we need data from weather balloon instruments (called radiosondes) sent aloft twice a day across the United States. Radiosonde data is the backbone of weather prediction. This data is now available on the Internet.

Imbalances that produce motion in the vertical dimension of the atmosphere provide the energy to drive our "weather engine." On a surface weather map, low pressure, high pressure, cold fronts and warm fronts, and patterns of temperature, humidity, and wind result from the up and down motions in the air above. From the vertical distribution of heat and moisture, as measured by temperature and dew point, we can predict the weather, and especially the most significant of weather occurrences, thunderstorms.

The goals of this session are to show how to obtain radiosonde data off the Internet and to use this data to predict thunderstorms. We will employ a thermodynamic diagram called the Skew-T to display and analyze upper air data. Examples will illustrate the concept of atmospheric stability and instability and we will relate this to the potential for deep convection resulting in thunderstorms.

Though far from simple, we have designed this simplified treatment of thermodynamics in the atmosphere for the non-meteorologist to gain practical knowledge in the interpretation of upper air data. Several figures and passages are modified from a workbook called Basic Convection I by Doswell, Anderson, and Imy (NOAA, 1991) used to train National Weather Service meteorologists. We hope this will give you a jump start in the science of thunderstorm prediction that will help stimulate a more thorough classroom investigation of weather.

Back To Top

Weather Balloons and Radiosondes

Weather balloons and their shoe box-sized packages of instruments called radiosondes have been used to gather weather measurements of the air aloft since the 1930s. Beginning in the 1940s, radiosondes have been the mainstay of the global upper-air observing system with more than 1000 radiosonde stations operated by 92 countries as of the early 1990s. (NOAA, 1992: Strategic Plan for Upper Air Observations. 18pp.) Most of the stations launch weather balloons twice a day, once at midnight UTC (Universal or Greenwich Meridian Time) and again at noon UTC (7AM and 7PM EST). Across the continental United States, weather balloons are launched from about 100 different locations, including National Weather Service (see Figure 1 below), military, and other government agency sites. In North Carolina, National Weather Service sites include Newport (KMHX) and Greensboro (KGSO).

Radiosondes are equipped to measure temperature, dew point (a measure of humidity) and pressure. The instruments convert the measurements to electrical signals sent via an internal radio transmitter to the ground station as the radiosonde rises through the atmosphere. Radiosondes usually rise to an altitude approaching 30 km (18 miles or 98000 feet). As the tracking antenna measures the movement of the balloon rising through the atmosphere, winds are also calculated.

Satellite and ground-based sensors may eventually replace radiosondes as an upper air observing system and allow us to measure the upper air almost continually, not just twice a day. However, no instrument to date has been able to duplicate the high vertical resolution (features resolved with vertical scales of a few tens of meters) needed in the lower atmosphere to make accurate weather predictions.

Figure 1
Figure 1: Weather balloon launch sites in the Continental U.S. - 1997

Back To Top

Vertical Motion in the Atmosphere

Of the many uses of upper air data from radiosondes, one of the most significant uses by forecasters on a daily basis is to determine the potential for vertical motion from the profiles of temperature and humidity. Though we often think of the wind, mainly in the horizontal plane, as important to weather, it is the movement of air up and down in the atmosphere, vertical motion, that leads to the types of weather we experience. Because we cannot measure vertical motion directly, it must be inferred through our understanding of its relationship to temperature, humidity, pressure, and wind.

Very simply, hot air rises and cold air sinks. The sun heats the ground that heats the air. Each day over land in the middle latitudes, the warmest air in the lower atmosphere is usually near the ground or at the bottom of the atmosphere. As air near the ground is heated, it becomes buoyant, similar to the way bubbles of gas become buoyant in a pot of boiling water. This buoyancy force is opposed by gravity and the weight of the air in the atmosphere pressing down from above. Therefore, each day, the lower atmosphere becomes the battle ground between opposing forces that results in warmer, buoyant air moving up and cooler, heavier air moving down.

We call this process convection. Meteorologists often use the term convection to mean thunderstorms, although thunderstorms are only one form (albeit one of the most dramatic forms) of convection in our atmosphere. In a broader sense, convection refers to the transport of heat (and moisture) by the movement of a fluid. In atmospheric convection, part of the heat transferred is called sensible heat, since it is the heat one might "sense" with a thermometer. With water vapor in the air, the possibility exists that latent heat will be realized. We call this type of heat "latent" because one cannot sense it until the water vapor condenses. When water vapor condenses into droplets, it releases that heat absorbed when the water was vaporized (heat of vaporization). Therefore, convection can transfer latent heat as well as sensible heat.

For the moment, let's "shut off the sun". Consider the temperature profile of the earth's atmosphere without any heating or cooling from the ground and without water vapor. We call this condition of no heat being added or subtracted from the air, an adiabatic condition. In this case, only compression caused by the weight of the air above warms the air near the ground. Going out toward space, air becomes progressively cooler with height because there is less air above it and less compression (i.e. lower pressure).

We call temperature change with height the lapse rate. With adiabatic conditions and without water vapor, we call this the dry, adiabatic lapse rate. The dry, adiabatic lapse rate is about 9.8C/km (5.5F/kft). In other words, the temperature decreases 9.8C because of decompression for every kilometer in altitude gained. An example of this relationship between temperature and pressure is observed when you release pressure from an aerosol can (i.e. hair spray, spray paint, etc.) and the side of the can becomes very cold.

Dry air moving up or down and undergoing decompression or compression in the atmosphere loses or gains temperature at the dry, adiabatic lapse rate. If air becomes saturated with water vapor and condensation takes place, latent heat is released. If air remains saturated and continues to rise, the rate of cooling will be offset by the release of latent heat. This new lapse rate is the moist, adiabatic lapse rate. The moist, adiabatic lapse rate is about 4.1C/km, less than half the dry, adiabatic lapse rate. This difference between dry and moist adiabatic lapse rates is very important in considering the role of moisture in the development of thunderstorms, as we shall see.

Dry and moist adiabatic lapse rates are represented by lines called adiabats on a thermodynamic diagram. Here we plot the observed temperature and dew point data from weather balloon radiosondes against a backdrop of dry and moist adiabats. In the next section, we will use adiabats on a thermodynamic diagram to find energy in the atmosphere by simulating the up and down movement of air.

Back To Top

The Skew-T, Log P Thermodynamic Diagram

Meteorologists display and analyze upper air data on a chart called the Skew-T, Log P Thermodynamic Diagram (commonly called Skew-T). Upper air data plotted on Skew-Ts are now available daily on the Internet. Figure 2 shows a simplified version of a Skew-T with plotted temperature and dew point data. Here, we plot the radiosonde temperature and humidity (dew point) data against a backdrop of lines representing the dry, adiabatic lapse rate (solid lines rising toward a 10 o'clock direction) and the moist, adiabatic lapse rate (long-dashed curved lines).

The Skew-T is based on temperature and the logarithm of pressure as its coordinates. However, instead of running up and down, the isotherms are skewed toward 2 o'clock from the vertical. (This takes some getting used to.) Isotherms and dry adiabats intersect at 90 degree angles. This does two things: (1) variations in the lapse rate of plotted radiosonde data become more obvious, and (2) equal areas bounded by isotherms and adiabats anywhere on the Skew-T represent equal amounts of energy. These features will be important in assessing the potential for thunderstorms.

The Skew-T, Log P Thermodynamic Diagram

Figure 2
Figure 2: Skew-T diagram showing a plotted temperature (solid bold line)
and dew point (dashed bold line) sounding.

Back To Top

Assessing Vertical Motion of Dry Air Using a Skew-T

Meteorologists use the Skew-T to judge the potential for upward and downward movement of the air given the existing temperature and humidity (dew point) profiles of the airmass. We also use it to determine the characteristics of the airmass once vertical motion has started. We call the theoretical movement of air on a Skew-T "parcel theory" because an imaginary quantity of air being moved on the diagram is called a parcel. (In some literature, parcels may also be simply referred to as "bubbles of air".)

Figure 3 shows a small portion of a Skew-T with a bold line labeled "Environmental Sounding" illustrating the observed temperature profile of the airmass. If a parcel in the atmosphere at any height along the environmental sounding is forced upward (i.e. by a cold front), it undergoes decompression as it encounters lower pressure with altitude. The parcel expands, and it cools at a rate prescribed by the dry, adiabatic lapse rate. Its projected temperature off the sounding is traced along or parallel to a dry adiabat.

Notice now in Figure 3 that the rising parcel is colder than its environment (as measured by the temperature scale and corresponding isotherms). In the given example, the rising parcel has a temperature close to -13C whereas its surrounding environment at the same level is about -11C. Because the rising parcel is colder than its surroundings, its density is greater. It resists further upward movement and it tends to sink to a lower level where its temperature (and density) is similar to its surroundings. On the Skew-T, this downward motion is again traced along the dry adiabat toward the environmental sounding.

If the parcel's downward motion overshoots the sounding and sinks to a lower level, as in Figure 3, its temperature becomes warmer than its surroundings. In the given example, the descending parcel has a temperature close to +4C whereas its surrounding environment at the same level is about 5 degrees colder at -1C. Because the sinking parcel is warmer now than its surroundings, its density is less (i.e. it becomes buoyant) and it resists further downward movement. In this situation, the parcel seeks temperature/density equilibrium and, on the Skew-T, moves back toward the environmental sounding. This is called a stable condition because parcels resist change and the sounding is an example of a stable environment.

Figure 3
Figure 3: Diagram showing the temperature of a parcel moving up
and down along a dry adiabat in a stable environment.

Figure 4 shows a different environmental sounding on a Skew-T. Again, if a parcel in the atmosphere at any height along the environmental sounding is forced upward, it cools at a rate prescribed by the dry, adiabatic lapse rate. Therefore, its movement off the sounding is traced along or parallel to a dry adiabat.

Figure 4
Figure 4: Diagram showing the temperature of a parcel moving up
and down along a dry adiabat in an unstable environment.

Here, the rising parcel is warmer than its environment (as measured by the temperature scale and corresponding isotherms). In this case, the rising parcel has a temperature close to -13C, but its surrounding environment (assuming the sounding temperature can be extrapolated along the bold line) is much colder. The parcel is buoyant because it is less dense, and the parcel accelerates upward. Similarly, if a parcel is forced downward, its temperature becomes colder than its surroundings. Here, the descending parcel has a temperature close to +9C, but its surrounding environment at the same level is about 5 degrees warmer at +14C. Because the sinking parcel is colder, its density is greater. It is heavier than its surroundings, and it accelerates toward the ground. We call this condition unstable because parcels are prone to move up or down. This sounding is an example of an unstable environment.

In Figure 5, the environmental sounding shows a temperature profile that is equivalent to the dry adiabatic lapse rate. A parcel forced up or down In this case will have a temperature and density similar to its surroundings. It will tend to stay at the level to which it is moved. In this case, the airmass is said to be neutral.

Figure 5
Figure 5: Diagram showing the temperature of a parcel moving up
and down along a dry adiabat in an neutral (dry adiabatic) environment.

We have examined the stability of an atmospheric layer by comparing the layer's lapse rate to the dry adiabatic lapse rate. By heating a fluid (air, of course, being a fluid) from below, we increase the lapse rate. This means temperature changes more per unit altitude because the ability of the fluid to conduct heat without motion is exceeded. Since heat conduction is a slow process, heating a fluid from below normally creates a large temperature difference across the depth of the fluid. Once the dry, adiabatic lapse rate is exceeded, the slightest push up or down on the parcel causes that parcel to rise or sink. Thus, a fluid with a lapse rate exceeding the dry adiabatic rate overturns very easily (Doswell, C.A.III, et al., 1991: Basic Convection I - A Review of Atmospheric Thermodynamics, NOAA-NWS, Norman, OK).

Rarely do we see an environmental lapse rate as great as the one in Figure 4 except near the ground on days of strong surface heating by the sun. In a shallow layer near the ground, it is possible for extremely high lapse rates to develop on a sunny day. Here, the rapid overturning of the air creates the shimmering effect you see near the ground on such days. Elsewhere in the atmosphere, sufficient random motion exists to quickly relieve the imbalance by mixing the layer of unstable air before it becomes established.

Back To Top

Assessing Vertical Motion of Moist Air Using a Skew-T

In order to achieve overturning on a larger scale, or deep convection, to produce thunderstorms several kilometers tall, there must be another source of heat energy to destabilize the atmosphere. That source is contained in moisture, water vapor.

If condensation of water vapor occurs, the release of latent heat changes things considerably. When a parcel of saturated air is lifted, condensation takes place and releases latent heat. Now instead of cooling at a dry, adiabatic lapse rate, the addition of latent heat offsets the cooling caused by decompression as the parcel rises. The new rate of cooling is called the moist, adiabatic lapse rate and can be seen as the long dashed curved lines on the Skew-T (see Figure 2).

The amount of moisture in the air aloft is shown by the bold dashed line representing the dew point at each level as measured by the radiosonde. Parts of the sounding where temperature and dew point come close together represent layers with higher humidity and more nearly saturated air.

The dew point is the temperature of the air if cooled to saturation at constant pressure. On the ground, dew forms when the air is cooled to its dew point temperature, usually at night. Though we know pressure changes at ground level with the weather, these changes are very small, and pressure can be assumed constant, compared to changes over a vertical distance in the atmosphere. We can attest to these pressure changes when we travel in airplanes or up and down steep terrain and feel our ears pop.

With an air parcel that is rising, the parcel's pressure changes greatly. So changes in dew point must be taken into account during ascent. The rate at which dew point changes is guided by a factor called the mixing ratio. The mixing ratio is the ratio of the mass of water vapor per unit mass of dry air. Since mass is not added or subtracted from the unsaturated parcel, this ratio remains the same as the parcel moves up or down. On the Skew-T in Figure 2, lines of constant mixing ratio are represented by the straight short-dashed lines running generally from 7 o'clock to 1 o'clock on the diagram.

On the Skew-T, recall for rising parcels of air, the temperature changes from its environment at a rate parallel to the lines representing the dry, adiabatic lapse rate. Starting from the same level, the corresponding dew point changes from its environment at a rate parallel to the lines representing constant mixing ratio. Figure 6 shows the process of a parcel moving up from near the surface, 1000 millibars in this case. The point of intersection between the dry adiabat and mixing ratio line is, theoretically, the point at which the parcel becomes saturated and condensation begins. The height of this point is known as the Lifted Condensation Level, or LCL. Meteorologists use this to determine the height of the cloud bases above ground. If the parcel continues to rise, it will cool more slowly because latent heat caused by condensation is added to the parcel. The parcel will cool at the moist, adiabatic rate as long as it remains saturated.

Figure 6
Figure 6: Location of the Lifted Condensation Level (LCL)
of a parcel lifted from ground level.

Back To Top

Determining Potential for Deep Convection - Thunderstorms

Keeping parcel theory in mind, we will examine a sample temperature and dew point sounding (remember, sounding refers to the environmental data obtained from the radiosonde) to determine its potential for deep convection. Here, deep means from near the ground to near the top of the troposphere, usually 8 to 10 kilometers.

Bold solid and dashed lines representing sounding temperature (T) and dew point (Td) respectively are plotted on a simplified Skew-T in Figure 7. If a parcel of air is lifted from near the ground, its temperature decreases along a specific dry adiabat drawn from the surface. Here the surface is assumed to be at the 1000 mb level. Its dew point decreases along a specific mixing ratio line drawn from the surface. Moisture in the parcel condenses at the height of intersection between the dry adiabat and mixing ratio line for the ground-based parcel. This is the LCL, and, theoretically, the base of a developing cloud. (Note, by convention, we call the mixing ratio line running through the LCL the saturation mixing ratio.)

If the parcel continues to rise, driven upward by a front or forced upward over a mountain or lifted by a channel of warm air called a thermal, its temperature decreases from the LCL along a moist adiabat. At the lower levels, the parcel's temperature is colder than its environment. Its normal tendency is to sink back toward the ground. We can observe this process on a sunny day by carefully watching the up and down growth and dissipation of "fair-weather" cumulus clouds. Here, the upward surge of heated air is sufficient to lift the air to its LCL to produce cumulus clouds, but it is soon overcome by cooling from decompression.

If the mechanism causing the air to be lifted is sufficiently strong, the parcel's temperature continues to follow the moist adiabat until it intersects the sounding temperature. This level is called the Level of Free Convection, or LFC. Here, the parcel's temperature is equal to that of its surroundings. Assuming the parcel remains saturated with water vapor, any additional upward push makes the parcel warmer than its environment. The parcel now becomes buoyant and accelerates upward. When this happens, we can see cumulus clouds suddenly tower, sometimes appearing to explode into the upper troposphere. Towering cumulus will continue to grow unless drier air is drawn in (entrained) from the sides. With the entrainment of drier air, the parcel is no longer saturated. Thus, it can no longer ascend with a temperature changing at the moist adiabatic rate. It cools faster at a dry adiabatic rate to reach equilibrium with its surroundings at a lower level.

Figure 7
Figure 7: Showing the calculation of the LFC and EL and the resultant
energy areas on the Skew-T diagram for a parcel lifted from
ground level.

If the channel of rising air is sufficiently large, more of the air inside the channel is unaffected by dry air entrainment. Here we can assume parcels remain saturated. So in this case, our parcel continues to accelerate upward moist adiabatically until it reaches the sounding temperature once again. Sometimes this does not happen until the parcel rises several kilometers up and hits the bottom of the stratosphere. At this height, the parcel loses its buoyancy and its temperature is again equal to that of its surroundings. This is called the Equilibrium Level, or EL. Theoretically, the EL is the highest extent of precipitation within thunderstorms for this environment and we can use this to forecast the height of thunderstorms observed on radar.

On the diagram in Figure 7, two areas are defined, bounded on one side by the environmental temperature sounding and, on the other side, by the projected path (temperature) of the lifted air parcel. These areas are shaded and labeled "Negative Area" and "Positive Area." The size of each area is equivalent to the amount of buoyant energy in the atmosphere characterized by the sample sounding.

The shaded area labeled "Negative Area" is the amount of "negative buoyant energy" that a rising parcel must overcome if it is to break through to the Level of Free Convection. Meteorologists often refer to this negative area as "the cap," because, in effect, it puts a cap on deep convection. You can also think of this as environmental resistance to upward vertical motion in the lower atmosphere.

The shaded area labeled "Positive Area" is the amount of "positive buoyant energy" that a rising parcel can use to accelerate to the equilibrium level and sustain thunderstorms. Meteorologists refer to this positive area as CAPE, or Convective Available Potential Energy, because this energy is available to sustain deep convection. You can think of this as the fuel that powers thunderstorms.

The smaller the Negative Area and the larger the Positive Area, the easier it is for thunderstorms to form and become strong with a given environment. Some use the analogy of the pressure cooker on a hot day to compare the sudden explosive development of thunderstorms to a cooker popping its lid. Potential energy is quickly converted to kinetic energy to accelerate air upward at speeds sometimes approaching 50 meters per second or 100 mph. Radiosonde data displayed on Skew-Ts showing Negative and Positive Areas are now available in real time on the Internet for assessing thunderstorm potential.

In a qualitative manner, we can see that the size of the Negative Area in Figure 7 grows smaller by adding heat and/or humidity at ground level, thus, increasing temperatures and/or dew points at 1000 mb. We can also see that the size of the positive area grows larger if colder air arrives aloft (as often happens near jet streams) and the environmental temperature cools at a faster rate above the LFC (i.e. the solid bold line bends further to the left above the LFC). The process of heating from below or cooling from above stimulates and enhances the growth of thunderstorms.

To show what happens during daytime heating with increasing ground level temperatures, the environmental sounding in Figure 7 is modified and shown in Figure 8. The ground level temperature has been increased to 20C (68F) while leaving the dew point unchanged. We can assume that conditions near the ground will be dry adiabatic (neutral) because of vigorous convective mixing of the air near the ground. Graphically, we draw a line from 20 C at the 1000 mb level upward and parallel to a dry adiabat until it intersects with the sounding temperature. This line represents the new environmental conditions modified by the heating of air near the ground.

Figure 8
Figure 8: Showing the calculation of the LFC and EL and the resultant
energy areas on the Skew-T diagram as in Figure 7, but for a parcel
heated to 20C then lifted from ground level.

If a parcel continues to rise from the surface, it follows the same line of the dry adiabat to the mixing ratio line drawn from the surface dew point. Where the two lines meet is the new LCL which is higher than the previous. What this means is that the bases of cumulus clouds tend to rise with heating during the course of the day if there is little change in dew point near the ground. See if you notice this, especially during the warm season. Cloud bases tend to be closer to the ground in the morning and further aloft during the afternoon.

Continue lifting the parcel above the new LCL parallel to a moist adiabat to a new LFC at the intersection with the temperature sounding. The new Negative Area is smaller than before meaning the cap has been weakened by surface heating and a smaller amount of additional energy is needed for the air to become buoyant at the Level of Free Convection. If the parcel continues to rise and its temperature changes along the moist adiabat above the LFC to the EL, notice that the size of the Positive Area is larger than before. So surface heating not only reduces the cap and makes it easier for deep convection to get started, but it also increases the CAPE and makes deep convection stronger.

The Skew-T analysis helps explain in terms of temperature, dew point (humidity), and energy, why thunderstorms tend to form and are often strongest late in the day when the air near the ground is hottest.

Back To Top

Basic Ingredients for Thunderstorms

For thunderstorms to develop, three ingredients must be present in the atmosphere: moisture, instability, and upward vertical motion. On a Skew-T, moisture is assessed by the dew point and mixing ratio. Instability is determined by the lapse rate of temperature and the size of the positive area (as in figures 7 and 8).

Parcels must reach their LFCs in order to have deep convection for thunderstorms. A parcel needs to be lifted by some external process to reach its LFC. Here, we must examine other weather information besides the Skew-T to determine if an external source will be available to provide lift. Common sources of lift include cold fronts, sea breeze fronts, orographic lift (i.e., winds blowing up a mountain side), the outflow gust front of other thunderstorms, and strong surface heating that produces channels of rising air called thermals. An external source or combination of sources must supply energy to the parcel to overcome the negative area on the Skew-T. An accurate forecast will include an correct assessment of the amount of lift needed and amount of lift available to produce thunderstorms.

Back To Top

Modifying the Morning Sounding to Forecast Thunderstorms

Usually, the morning radiosonde data will show a temperature inversion (temperature increasing with height) just above the ground (Figure 9). This results from cooling near the ground at night. Assuming normal daytime heating will eliminate the morning inversion, we need to estimate the temperature profile for later in the day in order to accurately assess negative and positive areas on the sounding.

The most common way to modify the temperature profile is to plot the expected high temperature for the day at the ground and to follow a dry adiabat from that point until it intersects the environmental sounding. A similar modification was performed in Figure 8. In Figure 9, we've drawn the dotted lines along adiabats from expected temperature for each hour, 07 to 13 (7 AM to 1 PM) to represent the temperature profile that hour. With regard to moisture, on calm, cool mornings, the dew point is often very close to the low temperature reading. Water vapor from transpiration and evaporation sources at the ground gets trapped by the inversion near the ground overnight. As the air gets heated during the day, this water vapor will mix through the depth of the heated layer as defined by intersection of the high temperature adiabat with the environmental sounding (D in Figure 9).

Figure 9
Figure 9: Modifying the morning temperature profile for surface heating during the day.

It is more difficult to modify the dew point profile to reflect conditions later in the day than it is to modify the temperature profile because moisture may be lost or gained in various ways. But most times, meteorologists use an averaging process to obtain a representative value for moisture through the heated layer. A typical modification is to draw a mixing ratio line that averages the lowest 100 millibar layer of the dew point sounding (see Figure 10). We draw a line parallel to the mixing ratio that straddles the dew point sounding such that areas either side of the line are roughly equal. When a parcel is lifted off the surface using the high temperature dry adiabat and the average mixing ratio line, by convention, the point of intersection is called the Convective Condensation Level, or CCL, instead of the LCL as in earlier discussions. But the principle is the same. The only difference is that the CCL represents a layer average, and the LCL is obtained using a discrete level in the sounding. Depending on the situation, a meteorologist may use one or the other to predict the level of clouds bases.

Figure 10
Figure 10: Modifying the morning sounding using the forecast high temperature
and the average dew point in the lowest 100 mb to determine the CCL.

Back To Top


Doswell, C.A.,III, L.C. Anderson, D.A. Imy, 1991: BASIC CONVECTION I: A Review of Atmospheric Thermodynamics, NOAA-NWS, Norman, OK.

Back To Top