Weather Forecasting

The national weather service

Types of weather forecasts

Long-range forecasting

Numerical weather prediction

Accuracy of weather forecasts

Resources

Weather forecasting is the attempt by meteorologists to predict the state of the atmosphere at some future time and the weather conditions that may be expected. Weather forecasting is the single most important practical reason for the existence of meteorology as a science. Knowing the future of the weather can be important for individuals and organizations. Accurate weather forecasts can tell a farmer the best time to plant; an airport control tower what information to send to aircraft that are landing and taking off; and residents of a coastal region when a hurricane might strike.

Humans have been looking for ways to forecast the weather for centuries. The Greek natural philosopher Theophrastus wrote a book, Book of Signs, in about 300 BC, listing more than 200 ways of knowing when to expect rain, wind, fair conditions, and other kinds of weather.

Scientifically based weather forecasting was not possible until meteorologists were able to collect data about current weather conditions from a widespread system of observing stations and organize that data in a timely fashion. By the 1930s these conditions had been met. Vilhelm and Jacob Bjerknes developed a weather station network in the 1920s that allowed for the collection of regional weather data. The weather data collected by the network could be transmitted nearly instantaneously by use of the telegraph, invented in the 1830s by Samuel F. B. Morse. The age of scientific forecasting, also referred to as synoptic forecasting, was under way.

The national weather service

In the United States, weather forecasting is the responsibility of the National Weather Service (NWS), a division of the National Oceanic and Atmospheric Administration (NOAA) of the Department of Commerce. NWS maintains more than 400 field offices and observatories in all 50 states and overseas. The future modernized structure of the NWS will include 116 weather forecast offices (WFO) and 13 river forecast centers, all collocated with WFOs. WFOs also collect data from ships at sea all over the world and from meteorological satellites circling Earth. Each year the Service collects nearly four million pieces of information about atmospheric conditions from these sources.

The information collected by WFOs is used in the weather forecasting work of NWS. The data is processed by nine National Centers for Environmental Prediction (NCEP). Each center has a specific weather-related responsibility: seven of the centers focus on weather prediction—the Aviation Weather Center, the Climate Prediction Center, the Hydrometeorological Prediction Center, the Marine Prediction Center, the Space Environment Center, the Storm Prediction Center, and the Tropical Prediction Center—while the other two centers develop and run complex computer models of the atmosphere and provide support to the other centers—the Environmental Prediction Center and NCEP Central Operations. Severe weather systems such as thunderstorms and hurricanes are monitored at the National Storm Prediction Center in Norman, Oklahoma. Hurricane watches and warnings are issued by the Tropical Prediction Center in Miami, Florida, (serving the Atlantic, Caribbean, Gulf of Mexico, and eastern Pacific Ocean) and by the Forecast Office in and Honolulu, Hawaii, (serving the central Pacific). WFOs, other government agencies, and private meteorological services rely on NCEP’s information, and many of the weather forecasts in the paper, and on radio and television, originate at NCEP.

Global weather data are collected at more than a thousand observation points around the world and then sent to central stations maintained by the World Meteorological Organization, a division of the United Nations. Global data are also sent to NWS’s NCEPs for analysis and publication.

Types of weather forecasts

The less one knows about the way the atmosphere works, the simpler weather forecasting appears to be. For example if clouds appear in the sky and a light rain begins to fall one might predict that rain will continue throughout the day. This type of weather forecast is known as a persistent forecast. A persistent forecast assumes the weather over a particular geographic area simply will continue into the future.

The validity of persistent forecasting lasts for a few hours, but not much longer because weather conditions result from a complex interaction of many factors that still are not well understood and that may change very rapidly.

A more reliable approach to weather forecasting is known as the steady-state or trend method. This method is based on the knowledge that weather conditions are strongly influenced by the movement of air masses that often can be charted quite accurately. A weather map might show that a cold front is moving across the great plains of the United States from west to east with an average speed of 10 mph (16 km/h). It might be reasonable to predict that the front would reach a place 100 mi (1,609 km) to the east in a matter of 10 hours. Since characteristic types of weather often are associated with cold fronts it then might be reasonable to predict the weather at locations east of the front with some degree of confidence.

A similar approach to forecasting is called the analogue method because it uses analogies between existing weather maps and similar maps from the past. For example suppose a weather map for December 10, 1996, is found to be almost identical with a weather map for January 8, 1993. Since the weather for the earlier date is already known it might be reasonable to predict similar weather patterns for the later date.

Yet another form of weather forecasting makes use of statistical probability. In some locations on Earth’s surface one can safely predict the weather because a consistent pattern has already been established. In parts of Peru it rains no more than a few inches per century. A weather forecaster in this region might feel confident that he or she could predict clear skies for tomorrow with a 99.9% chance of being correct.

Long-range forecasting

The complexity of atmosphere conditions is reflected in the fact that none of the forecasting methods outlined above is dependable for more than a few days. This reality does not prevent meteorologists from attempting to make long-term forecasts. These forecasts might predict the weather a few weeks, a few months, or even a year in advance.

The basis for long-range forecasting is a statistical analysis of weather conditions over an area in the past. For example a forecaster might determine that the average snowfall in December in Grand Rapids, Michigan, over the past 30 years had been 15.8 in (40.1 cm). A reasonable way to try estimating next year’s snowfall in Grand Rapids would be to assume that it might be close to 15.8 inches (40.1 cm).

Today this kind of statistical data is augmented by studies of global conditions such as winds in the upper atmosphere and ocean temperatures. If a forecaster knows that the jet stream over Canada has been diverted southward from its normal flow for a period of months, that change might alter precipitation patterns over Grand Rapids over the next few months.

Numerical weather prediction

All forms of forecasting make use of such numerical data as temperature, atmospheric pressure, and humidity. Numerical weather prediction, however, is a term that refers to forecasts obtained by using complicated mathematical calculations carried out with powerful computers.

Numerical weather prediction is based on mathematical models of the atmosphere. A mathematical model is a system of equations that attempt to describe the properties of the atmosphere and changes that may take place within it. These equations can be written because the gases that constitute the atmosphere obey the same physical and chemical laws that gases on Earth’s surface follow. For example, Charles’ law says that when a gas is heated it tends to expand. This law applies to gases in the atmosphere as it does to gases in a laboratory.

The technical problem that meteorologists face is that atmospheric gases are influenced by many different physical and chemical factors at the same time. A gas that expands according to Charles’ law may also be decomposing because of chemical forces acting on it.

In numerical weather predition, meteorologists select a group of equations that describe the conditions of the atmosphere as completely as possible for

KEY TERMS

Analogue method of forecasting —A prediction of future weather conditions based on the assumption that current conditions will produce weather patterns similar to those observed in the past.

Cold front —The leading edge of an advancing mass of cold air.

Hurricane warning —A notice issued when a hurricane has been observed either visually or on a radar screen.

Hurricane watch —A notice to the general public that a hurricane may be expected within a particular area.

Long-term forecast —A prediction of weather conditions over a matter of weeks, months, or a year.

Mathematical model —A system of equations that attempts to describe the properties of the atmosphere and changes that may take place within it.

Numerical forecast —A prediction of future weather patterns obtained by using high speed computers to carry out complex mathematical calculations derived from mathematical models of the atmosphere.

Percent skill —The likelihood that a weather forecast will be better than a pure chance prediction.

Persistent forecast —A prediction of weather conditions based on the assumption that the weather over a particular geographic area will remain constant over the near future.

Short-term forecast —A prediction for weather conditions over a matter of hours or days.

Statistical probability forecast —A prediction of future weather conditions based on an analysis of the likelihood of various conditions having occurred in the past.

Steady-state forecast, Trend method —A prediction of weather conditions based on the movement of air masses over a given geographical area at about the same direction and approximately the same speed as they have been moving.

Synopic forecasting —Scientifically based forecasts derived from the rapid collection and analysis of weather data from as extensive an area as possible.

any one location at any one time. This set of equations can never be complete because even a computer is limited as to the number of calculations it can complete in a reasonable time. Thus, meteorologists pick out the factors they think are most important in influencing the development of atmospheric conditions. These equations are incorporated into a computer program, and the computer will print out the changes that might be expected if atmospheric gases behave according to the scientific laws to which they are subject. From this printout a meteorologist can make a forecast of the weather in an area in the future.

The accuracy of numerical weather predictions depend primarily on two factors. First, the more data that is available to a computer the more accurate its results. Second, the faster the speed of the computer the more calculations it can perform and the more accurate its report will be. The difficulties inherent in numerical weather forecasting are underscored by the fact that, during the early 1960s, meteorologist Edward Lorenz was performing computer simulations of weather when he made discoveries that were important in the development of modern chaos theory. In the period from 1955 (when computers were first used in weather forecasting) to the current time, the so-called percent skill of forecasts has improved from about 30% to more than 60%. The percent skill measure was invented to describe the likelihood that a weather forecast will be better than pure chance.

Accuracy of weather forecasts

Weather forecasters have long been the subject of jokes, probably as much today as they were in Theophrastus’s time. One reason for this is that there is no standard measure of a correct weather forecast. Suppose that a forecaster predicts heavy rain will fall tomorrow. Is the forecast correct if some rain falls, but it is not perceived as heavy rain by some people?

Forecast accuracy also is difficult to judge because the average person’s expectations probably have increased as the percent skill of forecasts also has increased. A hundred years ago, few people would have expected to have much idea as to what the weather would be like 24 hours in the future. Today, a good next-day forecast is a reasonable expectation.

In general, the shorter the time period and the more limited the geographic area involved, the more accurate a forecast is likely to be. For periods of less than a day, a forecast covering an area of 100 sq mi (259 sq km) is likely to be quite dependable. Predictions about weather patterns over large areas weeks or months in the future are likely to be much less reliable.

See also Air masses and fronts; Atmosphere observation; Atmospheric circulation; Atmospheric temperature; Global climate; .

Resources

BOOKS

Ahrens, Donald C. Meteorology Today. Pacific Grove, Calif.: Brooks Cole, 2006.

Palmer, Tim and Renate Hagedorn, ed. Predictability of Weather and Climate. New York: Cambridge University Press, 2006.

David E. Newton

Weather forecasting

Weather forecasting is the attempt by meteorologists to predict the state of the atmosphere at some future time and the weather conditions that may be expected. Weather forecasting is the single most important practical reason for the existence of meteorology as a science. It is obvious that knowing the future of the weather can be important for individuals and organizations. Accurate weather forecasts can tell a farmer when the best time to plant is; an airport control tower what information to send to planes that are landing and taking off; and residents of a coastal region when a hurricane might strike.

Humans have been looking for ways to forecast the weather for centuries. The Greek natural philosopher Theophrastus wrote a book, Book of Signs, in about 300

b.c. listing more than 200 ways of knowing when to expect rain, wind , fair conditions, and other kinds of weather.

Scientifically based weather forecasting was not possible until meteorologists were able to collect data about current weather conditions from a relatively widespread system of observing stations and organize that data in a timely fashion. By the 1930s these conditions had been met. Vilhelm and Jacob Bjerknes developed a weather station network in the 1920s that allowed for the collection of regional weather data. The weather data collected by the network could be transmitted nearly instantaneously by use of the telegraph , invented in the 1830s by Samuel F. B. Morse. The age of scientific forecasting, also referred to as synoptic forecasting, was under way.

The National Weather Service

In the United States weather forecasting is the responsibility of the National Weather Service (NWS), a division of the National Oceanic and Atmospheric Administration (NOAA) of the Department of Commerce. NWS maintains more than 400 field offices and observatories in all 50 states and overseas. The future modernized structure of the NWS will include 116 weather forecast offices (WFO) and 13 river forecast centers, all collocated with WFOs. WFOs also collect data from ships at sea all over the world and from meteorological satellites circling Earth . Each year the Service collects nearly four million pieces of information about atmospheric conditions from these sources.

The information collected by WFOs is used in the weather forecasting work of NWS. The data is processed by nine National Centers for Environmental Prediction (NCEP). Each center has a specific weather-related responsibility: seven of the centers focus on weather pre diction—the Aviation Weather Center, the Climate Prediction Center, the Hydrometeorological Prediction Center, the Marine Prediction Center, the Space Environment Center, the Storm Prediction Center, and the Tropical Prediction Center—while the other two centers develop and run complex computer models of the atmosphere and provide support to the other centers—the Environmental Prediction Center and NCEP Central Operations. Severe weather systems such as thunderstorms and hurricanes are monitored at the National Storm Prediction Center in Norman, Oklahoma. Hurricane watches and warnings are issued by the Tropical Prediction Center in Miami, Florida, (serving the Atlantic, Caribbean, Gulf of Mexico, and eastern Pacific Ocean) and by the Forecast Office in and Honolulu, Hawaii, (serving the central Pacific). WFOs, other government agencies, and private meteorological services rely on NCEP's information, and many of the weather forecasts in the paper , and on radio and television , originate at NCEP.

Global weather data are collected at more than 1,000 observation points around the world and then sent to central stations maintained by the World Meteorological Organization, a division of the United Nations. Global data also is sent to NWS's NCEPs for analysis and publication.

Types of weather forecasts

The less one knows about the way the atmosphere works the simpler weather forecasting appears to be. For example if clouds appear in the sky and a light rain begins to fall one might predict that rain will continue throughout the day. This type of weather forecast is known as a persistent forecast. A persistent forecast assumes the weather over a particular geographic area simply will continue into the future.

The validity of persistent forecasting lasts for a few hours, but not much longer because weather conditions result from a complex interaction of many factors that still are not well understood and that may change very rapidly.

A somewhat more reliable approach to weather forecasting is known as the steady-state or trend method. This method is based on the knowledge that weather conditions are strongly influenced by the movement of air masses which often can be charted quite accurately. A weather map might show that a cold front is moving across the great plains of the United States from west to east with an average speed of 10 mph (16 kph). It might be reasonable to predict that the front would reach a place 100 mi (1,609 km) to the east in a matter of 10 hours. Since characteristic types of weather often are associated with cold fronts it then might be reasonable to predict the weather at locations east of the front with some degree of confidence.

A similar approach to forecasting is called the analogue method because it uses analogies between existing weather maps and similar maps from the past. For example suppose a weather map for December 10, 1996, is found to be almost identical with a weather map for January 8, 1993. Since the weather for the earlier date is already known it might be reasonable to predict similar weather patterns for the later date.

Yet another form of weather forecasting makes use of statistical probability. In some locations on Earth's surface one can safely predict the weather because a consistent pattern has already been established. In parts of Peru it rains no more than a few inches per century. A weather forecaster in this region might feel confident that he or she could predict clear skies for tomorrow with a 99.9% chance of being correct.

Long-range forecasting

The complexity of atmosphere conditions is reflected in the fact that none of the forecasting methods outlined above is dependable for more than a few days at best. This reality does not prevent meteorologists from attempting to make long-term forecasts. These forecasts might predict the weather a few weeks, a few months, or even a year in advance. One of the best known (although not necessarily the most accurate) of long-term forecasts is found in the annual edition of the Farmer's Almanac.

The basis for long-range forecasting is a statistical analysis of weather conditions over an area in the past. For example a forecaster might determine that the average snow fall in December in Grand Rapids, Michigan, over the past 30 years had been 15.8 in (40.1 cm). A reasonable way to try estimating next year's snowfall in Grand Rapids would be to assume that it might be close to 15.8 inches (40.1 cm).

Today this kind of statistical data is augmented by studies of global conditions such as winds in the upper atmosphere and ocean temperatures. If a forecaster knows that the jet stream over Canada has been diverted southward from its normal flow for a period of months, that change might alter precipitation patterns over Grand Rapids over the next few months.

Numerical weather prediction

The term numerical weather prediction is something of a misnomer since all forms of forecasting make use of numerical data like temperature , atmospheric pressure , and humidity . More precisely numerical weather prediction refers to forecasts that are obtained by using complex mathematical calculations carried out with high-speed computers.

Numerical weather prediction is based on mathematical models of the atmosphere. A mathematical model is a system of equations that attempt to describe the properties of the atmosphere and changes that may take place within it. These equations can be written because the gases which comprise the atmosphere obey the same physical and chemical laws that gases on Earth's surface follow. For example, Charles' law says that when a gas is heated it tends to expand. This law applies to gases in the atmosphere as it does to gases in a laboratory.

The technical problem that meteorologists face is that atmospheric gases are influenced by many different physical and chemical factors at the same time. A gas that expands according to Charles' law may also be decomposing because of chemical forces acting on it. How can anyone make use of all the different chemical and physical laws operating in the atmosphere to come up with a forecast of future atmospheric conditions?

The role of computers in weather forecasting

The answer is that no human can solve such a problem. The mathematics involved are too complex. The task is not too much for computer, however. Computers can perform a series of calculations in a few hours that would take a meteorologist his or her whole lifetime to finish.

In numerical weather predicting meteorologists select a group of equations that describe the conditions of the atmosphere as completely as possible for any one location at any one time. This set of equations can never be complete because even a computer is limited as to the number of calculations it can complete in a reasonable time. Thus, meteorologists pick out the factors they think are most important in influencing the development of atmospheric conditions. These equations are fed into the computer. After a certain period of time, the computer will print out the changes that might be expected if atmospheric gases behave according to the scientific laws to which they are subject. From this printout a meteorologist can make a forecast of the weather in an area in the future.

The accuracy of numerical weather predictions depend primarily on two factors. First, the more data that is available to a computer the more accurate its results. Second, the faster the speed of the computer the more calculations it can perform and the more accurate its report will be. In the period from 1955 (when computers were first used in weather forecasting) to the current time, the percent skill of forecasts has improved from about 30% to more than 60%. The percent skill measure was invented to describe the likelihood that a weather forecast will be better than pure chance.

Accuracy of weather forecasts

Weather forecasters have long been the subject of jokes, probably as much today as they were in Theophrastus's time. One reason for this is that there is no standard measure of a "correct" weather forecast. Suppose that a forecaster predicts heavy rain for your area tomorrow. Does a rainfall of 1 in (2.5 cm) prove that prediction correct? Or a rainfall of 1.5 in (1 cm)? Or a rainfall of 5 in (13 cm)?

Forecast accuracy also is difficult to judge since the average person's expectations probably have increased as the percent skill of forecasts also has increased. A hundred years ago, few people would have expected to have much idea as to what the weather would be like 24 hours in the future. Today, a good next-day forecast often is possible.

In general it is probably safe to say that the shorter the time period and the more limited the geographic area involved, the more accurate a forecast is likely to be. For periods of less than a day, a forecast covering an area of 100 sq mi (259 sq km) is likely to be quite dependable. Predictions about weather patterns six months from now for the state of California are likely to be much less reliable.

See also Air masses and fronts; Atmosphere observation; Atmospheric circulation; Atmospheric temperature; Global climate; Weather mapping.

Resources

books

Danielson, Eric W., James Levin, and Elliot Abrams. Meteorology. 2nd ed. with CD-ROM. Columbus: McGraw-Hill Science/Engineering/Math, 2002.

Hodgson, Michael, and Devin Wick. Basic Essentials: Weather Forecasting. 2nd ed. Guilford, CT: Globe Pequot Press, 1999.

Lutgens, Frederick K., and Edward J. Tarbuck. The Atmosphere: An Introduction to Meteorology. 8th ed. New York: Prentice-Hall, 2000.

Lynott, Robert E. How Weather Works and Why. Gadfly Press, 1994.

periodicals

"Boundary-Layer Meteorology." Boundary-Layer Meteorology 105, no. 3-3 (2002): 515-520.

Lee, Thomas. "Eleventh AMS Conference on Satellite Meteorology and Oceanography." Bulletin of the American Meteorological Society 83, no. 11 (2002): 1645-1648.

Spellman, Greg. "Experiences Teaching Meteorology To Adults." Journal Of Meteorology 27 no. 268 (2002): 133-137.

other

The National Weather Service [cited 2003]. <http://www.nws.noaa.gov>.

David E. Newton

KEY TERMS

Analogue method of forecasting

—A prediction of future weather conditions based on the assumption that current conditions will produce weather patterns similar to those observed in the past.

Cold front

—The leading edge of an advancing mass of cold air.

Hurricane warning

—A notice issued when a hurricane has been observed either visually or on a radar screen.

Hurricane watch

—A notice to the general public that a hurricane may be expected within a particular area.

Long-term forecast

—A prediction of weather conditions over a matter of weeks, months, or a year.

Mathematical model

—A system of equations that attempts to describe the properties of the atmosphere and changes that may take place within it.

Numerical forecast

—A prediction of future weather patterns obtained by using high speed computers to carry out complex mathematical calculations derived from mathematical models of the atmosphere.

Percent skill

—The likelihood that a weather forecast will be better than a pure chance prediction.

Persistent forecast

—A prediction of weather conditions based on the assumption that the weather over a particular geographic area will remain constant over the near future.

Short-term forecast

—A prediction for weather conditions over a matter of hours or days.

Statistical probability forecast

—A prediction of future weather conditions based on an analysis of the likelihood of various conditions having occurred in the past.

Steady-state forecast, Trend method

—A prediction of weather conditions based on the movement of air masses over a given geographical area at about the same direction and approximately the same speed as they have been moving.

Synopic forecasting

—Scientifically based forecasts derived from the rapid collection and analysis of weather data from as extensive an area as possible.

Weather forecasting

Weather forecasting is the attempt by meteorologists to predict the state of the atmosphere at some future time and the weather conditions that may be expected. Weather forecasting is the single most important practical reason for the existence of meteorology as a science. It is obvious that knowing the future of the weather can be important for individuals and organizations. Accurate weather forecasts can tell a farmer the best time to plant, an airport control tower what information to send to planes that are landing and taking off, and residents of a coastal region when a hurricane might strike.

Humans have been looking for ways to forecast the weather for centuries. The Greek natural philosopher Theophrastus wrote a Book of Signs, in about 300 b.c. listing more than 200 ways of knowing when to expect rain, wind , fair conditions, and other kinds of weather.

Scientifically-based weather forecasting was not possible until meteorologists were able to collect data about current weather conditions from a relatively widespread system of observing stations and organize that data in a timely fashion. By the 1930s, these conditions had been met. Vilhelm and Jacob Bjerknes developed a weather station network in the 1920s that allowed for the collection of regional weather data. The weather data collected by the network could be transmitted nearly instantaneously by use of the telegraph, invented in the 1830s by Samuel F. B. Morse. The age of scientific forecasting, also referred to as synoptic forecasting, was under way.

In the United States, weather forecasting is the responsibility of the National Weather Service (NWS), a division of the National Oceanic and Atmospheric Administration (NOAA) of the Department of Commerce. NWS maintains more than 400 field offices and observatories in all 50 states and overseas. The future modernized structure of the NWS will include 116 weather forecast offices (WFO) and 13 river forecast centers, all collocated with WFOs. WFOs also collect data from ships at sea all over the world and from meteorological satellites circling Earth. Each year the Service collects nearly four million pieces of information about atmospheric conditions from these sources.

The information collected by WFOs is used in the weather forecasting work of NWS. The data is processed by nine National Centers for Environmental Prediction (NCEP). Each center has a specific weather-related responsibility: seven of the centers focus on weather prediction—the Aviation Weather Center, the Climate Prediction Center, the Hydrometeorological Prediction Center, the Marine Prediction Center, the Space Environment Center, the Storm Prediction Center, and the Tropical Prediction Center—while the other two centers develop and run complex computer models of the atmosphere and provide support to the other centers—the Environmental Prediction Center and NCEP Central Operations. Severe weather systems such as thunderstorms, tornadoes, and hurricanes are monitored at the National Storm Prediction Center in Norman, Oklahoma, and the National Hurricane Center in Miami, Florida. Hurricane watches and warnings are issued by the National Hurricane Center's Tropical Prediction Center in Miami, Florida, (serving the Atlantic, Caribbean, Gulf of Mexico , and eastern Pacific Ocean) and by the Forecast Office in Honolulu, Hawaii, (serving the central Pacific). WFOs, other government agencies, and private meteorological services rely on NCEP's information, and many of the weather forecasts in the paper, and on radio and television, originate at NCEP.

Global weather data are collected at more than 1,000 observation points around the world and then sent to central stations maintained by the World Meteorological Organization, a division of the United Nations. Global data also are sent to NWS's NCEPs for analysis and publication.

The less one knows about the way the atmosphere works the simpler weather forecasting appears to be. For example, if clouds appear in the sky and a light rain begins to fall, one might predict that rain will continue throughout the day. This type of weather forecast is known as a persistent forecast. A persistent forecast assumes the weather over a particular geographic area simply will continue into the future. The validity of persistent forecasting lasts for a few hours, but not much longer because weather conditions result from a complex interaction of many factors that still are not well understood and that may change rapidly.

A somewhat more reliable approach to weather forecasting is known as the steady-state or trend method. This method is based on the knowledge that weather conditions are strongly influenced by the movement of air masses that often can be charted quite accurately. A weather map might show that a cold front is moving across the Great Plains of the United States from west to east with an average speed of 10 mph (16 kph). It might be reasonable to predict that the front would reach a place 100 mi (160 km) to the east in a matter of 10 hours. Since characteristic types of weather often are associated with cold fronts it then might be reasonable to predict the weather at locations east of the front with some degree of confidence.

A similar approach to forecasting is called the analogue method because it uses analogies between existing weather maps and similar maps from the past. For example, suppose a weather map for December 10, 2002, is found to be almost identical with a weather map for January 8, 1993. Because the weather for the earlier date is already known it might be reasonable to predict similar weather patterns for the later date.

Another form of weather forecasting makes use of statistical probability. In some locations on Earth's surface, one can safely predict the weather because a consistent pattern has already been established. In parts of Peru, it rains no more than a few inches per century. A weather forecaster in this region might feel confident that he or she could predict clear skies for tomorrow with a 99.9% chance of being correct.

The complexity of atmospheric conditions is reflected in the fact that none of the forecasting methods outlined above is dependable for more than a few days, at best. This reality does not prevent meteorologists from attempting to make long-term forecasts. These forecasts might predict the weather a few weeks, a few months, or even a year in advance. One of the best known (although not necessarily the most accurate) of long-term forecasts is found in the annual edition of the Farmer's Almanac.

The basis for long-range forecasting is a statistical analysis of weather conditions over an area in the past. For example, a forecaster might determine that the average snow fall in December in Grand Rapids, Michigan, over the past 30 years had been 15.8 in (40.1 cm). A reasonable way to try estimating next year's snowfall in Grand Rapids would be to assume that it might be close to 15.8 inches (40.1 cm).

Today this kind of statistical data is augmented by studies of global conditions such as winds in the upper atmosphere and ocean temperatures. If a forecaster knows that the jet stream over Canada has been diverted southward from its normal flow for a period of months, that change might alter precipitation patterns over Grand Rapids over the next few months.

The term "numerical" weather prediction is something of a misnomer because all forms of forecasting make use of numerical data such as temperature, atmospheric pressure , and humidity . More precisely, numerical weather prediction refers to forecasts that are obtained by using complex mathematical calculations carried out with high-speed computers.

Numerical weather prediction is based on mathematical models of the atmosphere. A mathematical model is a system of equations that attempt to describe the properties of the atmosphere and changes that may take place within it. These equations can be written because the gases that comprise the atmosphere obey the same physical and chemical laws that gases on Earth's surface follow. For example, Charles'Law says that when a gas is heated, it tends to expand. This law applies to gases in the atmosphere as it does to gases in a laboratory.

The technical problem that meteorologists face is that atmospheric gases are influenced by many different physical and chemical factors at the same time. A gas that expands according to Charles' Law may also be decomposing because of chemical forces acting on it. How can anyone make use of all the different chemical and physical laws operating in the atmosphere to come up with a forecast of future atmospheric conditions? The answer is mathematically complex. The task is not too much for computers, however. Computers can perform a series of calculations in a few hours that would take a meteorologist his or her whole lifetime to finish.

In numerical weather predicting, meteorologists select a group of equations that describe the conditions of the atmosphere as completely as possible for any one location at any one time. This set of equations can never be complete because even a computer is limited as to the number of calculations it can complete in a reasonable time. Thus, meteorologists pick out the factors they think are most important in influencing the development of atmospheric conditions. These equations are fed into the computer. After a certain period of time, the computer will print out the changes that might be expected if atmospheric gases behave according to the scientific laws to which they are subject. From this printout a meteorologist can make a forecast of the weather in an area in the future.

The accuracy of numerical weather predictions depend primarily on two factors. First, the more data that is available to a computer, the more accurate its results. Second, the faster the speed of the computer, the more calculations it can perform, and the more accurate its report will be. In the period from 1955 (when computers were first used in weather forecasting) to the current time, the percent skill of forecasts has improved from about 30% to more than 60%. The percent skill measure was invented to describe the likelihood that a weather forecast will be more accurate than pure chance.

Forecast accuracy also is difficult to judge because the average person's expectations probably have increased as the percent skill of forecasts also has increased. A hundred years ago, few people would have expected to have much idea as to what the weather would be like 24 hours in the future. Today,

an accurate next-day forecast often is possible. For periods of less than a day, a forecast covering an area of 100 mi² (259 km²) is likely to be quite dependable.

See also Air masses and fronts; Atmospheric chemistry; Atmospheric circulation; Atmospheric composition and structure; Atmospheric inversion layers; Drought; El Niño and La Nina phenomena; Hydrologic cycle; Isobars; Land and sea breeze; Lightning; Ocean circulation and currents; Thunder; Tornado; Tropical cyclone; Weather forecasting methods; Weather radar; Weather satellite; Wind chill

Weather Forecasting Models

The weather has an astonishing impact on our lives, ranging from the frustration of being caught in a sudden downpour to the trillions of dollars spent in weather-sensitive businesses. Consequently, a great deal of time, effort, money, and technology is used to predict the weather. In the attempt to improve weather prediction, meteorologists rely on increasingly sophisticated computers and computer models.

When researchers created the first computer models of Earth's atmosphere in the mid-twentieth century, they worked with computers that were extremely limited compared to the supercomputers that exist today. As a result, the first weather models were oversimplified, although they still provided valuable insights. As computer technology advanced, more of the factors that influenced the atmosphere, such as physical variables, could be taken into account, and the complexity and accuracy of weather forecast models increased.

Today's computer-enhanced weather forecasts can be successful only if the data observations are accurate and complete. Ground-based weather stations are equipped with many instruments, including barometers (which measure atmospheric pressure, or the weight of the air); wind vanes (which measure wind direction), anemometers (which measure wind velocity, or speed); hygrometers (which measure the moisture, or humidity, in the air); thermometers; and rain gauges. Data obtained through these instruments are combined with data from aircraft, ships, satellites, and radar networks. These data are then put into complex mathematical equations that model the physical laws governing atmospheric conditions. Solving these equations with the aid of computers yields a description—a forecast—of the atmosphere derived from its current state (that is, the initial values). This can then be interpreted in terms of weather—rain, temperature, sunshine, and wind.

Early Attempts to Forecast Weather

Several decades prior to the advent of the modern computer, Vilhelm Bjerknes, a Norwegian physicist-turned-meteorologist, advocated a mathematical approach to weather forecasting. This approach was based on his belief that it was possible to bring together the full range of observation and theory to predict weather changes. British scientist Lewis Fry Richardson was the first person to attempt to work out Bjerknes's program. During and

shortly after World War I, he went on to devise an algorithmic scheme of weather prediction. Unfortunately, Richardson's method required 6 weeks to calculate a 6-hour advance in the weather. Moreover, the results from his method were inaccurate. Consequently, Richardson's work, which was widely noticed, convinced contemporary meteorologists that a computational approach to weather prediction was completely impractical. Yet it laid the foundation for numerical weather prediction.

Following World War II, a Hungarian-American mathematician named John von Neumann began making plans to build a powerful and versatile electromechanical computer devoted to the advancement of the mathematical sciences. Von Neumann's objective was to demonstrate, through a particular scientific problem, the revolutionary potential of the computer. He chose weather prediction for that problem, and in 1946 established the Meteorology Project at the Institute for Advanced Study in Princeton, New Jersey, to work on its resolution.

The Chaos Connection. By the 1960s the work of American atmospheric sciences researcher Edward Lorenz showed that no matter how good the data or forecasting methods, accurate day-by-day forecasts of the weather may not be possible for future periods beyond about two weeks. Because Earth's atmosphere follows the complex, nonlinear rules of fluid dynamics , the smallest error in the determination of initial conditions (such as wind, temperature, and pressure) will amplify because of the nonlinear dynamics of atmospheric processes, so that a forecast at some point becomes nearly useless. This dependence of prediction on the precision of input values is a basic tenet of mathematical chaos theory .

The fundamental ideas of chaos theory are captured in what is known as the "butterfly effect," which, roughly stated, holds that a butterfly flapping its wings can set in motion a storm on the other side of the world. Yet with the next flap, nothing of meteorological significance may happen. Hence, chaos places a limit on the accuracy of computer-enhanced weather forecasts as the forecast period and geographic scale increases.

Relationship between Forecasting and Computers

The desire to make better weather predictions coincided with the evolution of today's modern electronic computers. In fact, computers and meteorology have helped shape one another. Moreover, weather forecasting has helped drive the development of better multimedia software, both for forecasting purposes and for communicating forecasts to specific audiences.

Forecasting the weather by computer is called numerical weather prediction. To make a skillful forecast, weather forecasters must understand the strengths and limitations of computer models and recognize that models can still be mistaken. Forecasters study the output from models over time and compare the forecast output to the weather that actually occurs. This is how they determine the biases, strengths, and weaknesses of the models. They also may modify what they learn from computer models based on their own experience forecasting the weather through physical and dynamical processes. Hence, good forecasters must not only have knowledge of meteorological variables, but they must also be able to use their intuition.

Improvements in Forecasts. Continued improvements in computer technology allows more of the dynamical and physical factors influencing the atmosphere to be added to models, thereby improving weather forecasts, including those for specific conditions. For example, some forecasting models help predict tropical weather features such as hurricanes, typhoons, and monsoons. Other models help forecast smaller-scale features such as thunderstorms and lightning. As these forecasting models improve, forecasters should be able to issue more accurate and timely storm warnings and advisories.

Improvements in Communication. Even if weather data were extraordinarily precise, weather forecasts would have little meaning if consumers of the data did not receive them in an effective and timely manner. Local and national weather reports demonstrate the strong relationship between weather forecasting and computer technology. Today's state-of-the-art visualization techniques for weather forecasts, including three-dimensional data, time-series, and virtual scenery, have made the weather segment a highly watchable part of television newscasts. A three-dimensional representation of weather patterns is typically superimposed on local and national geographic maps. With a click of a computer mouse, cloud formations and fronts appear, and predictions about temperature, precipitation, and other weather-related data march across the screen.

The twentieth century saw the inventions of radio broadcasting, television, and the Internet. The twenty-first century undoubtedly will bring other innovations in technology and media that will play an equally significant role in the dissemination of weather information.

see also Chaos; Computer Simulations; Predictions; Temperature, Measurement of; Weather, Violent.

Marilyn K. Simon

Bibliography

Hodgson, Michael. Weather Forecasting. Guildford, CT: The Globe Pequot Press, 1999.

Lockhart, Gary. The Weather Companion. New York: John Wiley & Sons, 1988.

Lorenz, Edward. The Essence of Chaos. Seattle: University of Washington Press, 1996.

Internet Resources

American Meteorological Society. <http://www.locusweather.com/media.html>.

"Forecasting by Computers." User Guide to ECMWF. ECMWF—European Centre for Medium-Range Weather Forecasts. <http://www.ecmwf.int/research/fcbycomputer.html>.

Ostro, Steve. "Weather Forecasting in the Next Century." <http://www.weather.com/weather_center/special_report/sotc/topten/next.html>.

Weather Forecasting

It was not long ago that television weather reports were presented by local personalities standing in front of a marker or chalkboard with a map of the United States drawn on it. To add some spice, small vinyl symbols of the sun and clouds were applied. The local personality was—and for the most part still is—the star. But during the 1980s, a transformation took place. Computers entered the modern television forecast office. This was the beginning of a new era in the forecasting and presentation of TV weather.

In big cities across America, TV weather has become big business. Broadcast meteorologists have always known that "the weather" is one of the most important parts of a local newscast. Weather affects everybody, and reports help people answer important questions such as "Should I wear a coat today?" or "Should I carry an umbrella?" Television station managers believed that the more "toys" a TV forecaster had, the more people would watch their station and their forecaster. They were correct for the most part. Weather segments and the associated broadcaster are huge audience draws.

What is the role of a local broadcast weather person? Forecast models and guidance are produced by National Weather Service (NWS) supercomputers in various places around the United States. A local TV forecaster has no need or ability to crunch all of those numbers. So, the local weather person receives data in raw form and also obtains prepared forecasts to aid in creating the forecast for the newscasts of the day.

Up to this point, the NWS has done a lot of the work. Measurements, or soundings, as they are called, are taken at various cities around the country. These data are sent to the NWS for analysis and forecast model production. The final "products" are then sent to various outlets for distribution. The National Weather Service makes these data available to end users and to private data vendors like Weather Services International (WSI).

From Data to Televised Weather Forecasts

Data supplied by the government (NWS) are sent to a vendor. That private company takes the data and sends the information, either untouched or reconfigured, to television stations via a satellite delivery system. A small dish is placed on the TV station roof to receive the data. The information is then "squeezed" into a Windows-based PC for further distribution at the television station. Many people think that this information and raw data are then analyzed by the local PC to produce an instant five-day forecast, but this is not the case. The data PC only receives and distributes data; it does not "automatically prepare" anything. The role of the local forecaster is to take these data, analyze the information, do a bit of math, and rely on one more element … instinct. Only then is a forecast prepared and ready to go to the public. To make this happen, more computers are necessary. The main PC sends data to one or more printers that create printouts of forecasts and raw data. Local weather bulletins and statements are also printed. The main PC also sends information to other important office-based computers, as well, where it is processed in different ways for several purposes.

Weather Warnings.

One system receiving data is a Windows-based PC that takes weather bulletins, like watches and warnings, and automatically creates informational maps and crawls to alert viewers to dangerous or potentially dangerous weather conditions. This First Warning System not only creates the maps and crawls but signals the forecast staff and others in the TV station that severe weather is upon the community and action must be taken. At this point, other TV station staffers, such as engineers in Master Control, can interface with the First Warning System remotely and get the prepared maps or crawls on the air, even if no trained forecaster is on duty at the time.

Weather Maps.

Data are also sent to another computer that handles the production of colorful graphics that can be displayed behind the on-camera weather person. This color graphics computer must be able to prepare and display satellite pictures and other visual aids. It must stand ready to be used as an artistic tool by the forecaster or weather producer. Such a system is rarely a Windows-based PC. Most are manufactured by Silicon Graphics to run various "supercharged" operating systems that can handle the graphic requirements of sophisticated programs quickly and with great detail.

Forecasters use rather dull informational maps behind the scenes. Making these maps look understandable and interesting to the viewer is the challenge of the color graphics system. This system must also monitor or create incoming satellite, radar, and temperature maps. It must present thumbnail shots of each individual map to allow the forecast staff to select easily the ones they want to use. This powerhouse of a graphics system not only acts as a producer but must then become a director as well. Various color graphics ready for broadcast can be displayed in time lapse mode or be custom-produced as an animated movie. These animations might show various fronts and features moving across the TV screen behind the on-camera forecaster. The staff also chooses special effects to highlight the display of various graphics. Different "wipes" between frames or movies can add interest to a TV weather segment and make it stand out from the competition.

Radar, Internet Access, and Communication Computers

A TV weather office often includes a PC-based control system for Doppler Radar. The radar system itself might be located miles away from the TV station. But a Windows-based PC acts as the command and control center to determine how many miles out the radar should scan. And the computer can determine which radar maps should be displayed.

Various other PC-based systems exist in most TV weather offices. They might include PCs used for news gathering and scriptwriting. Usually these systems serve many functions. Along with news gathering, they can usually access the Internet and intranet systems for various administrative tasks. These systems can provide backup support to the primary data computer, by acquiring data from the Internet. If the main data PC crashes or experiences a problem receiving signals from the satellite or dish, this backup system can be invaluable. This PC may also be used for sending and receiving e-mail and interoffice communications.

In summary, there is no question that computers now rule the local TV weather office. Surge protection and backup generators are used regularly at stations all across the country to safeguard the reliance on these systems. Although the "old days" of markers and chalk bring back fond memories, the computer-based systems in a modern TV weather office offer tremendous data capabilities and provide the on-air weather people with a host of sources to display the very latest weather from down the block or around the world.

see also Agriculture; Aircraft Traffic Management; Display Devices; Supercomputers.

Chuck Gaidica

Bibliography

Cox, John D. Weather for Dummies. Foster City, CA: IDG Books Worldwide, 2000.

Grazulis, Thomas P. The Tornado: Nature's Ultimate Windstorm. Norman: University of Oklahoma Press, 2001.

Williams, Jack. The Weather Book. New York: Vintage Books, 1997.