Trending Now

Exponential Changes in Forecasting

By Erik Pindrock
September 3, 2013; 10:33 AM ET
Share |

Some of the first weather analysis and forecasting methods came from Aristotle around 2,400 years ago in 340 B.C. He organized his thoughts in a formal document titled "Meteorologica" - where he observed and distinguished fire, air, water, and earth as four elements that caused "events in this world".

In today's day, the weather prediction has come a long way from simply observing the sky and earth. Meteorologists have computer models which model the atmosphere using parameters such as barometric pressure, temperature, humidity, and a plethora more of complicated mathematical equations.

The first computer model ever run for meteorological purposes was in 1950. As this computer model was processing information, several breakdowns and an abundance of data meant that it took over 24 hours for the model to process the data and output a forecast. Despite the length of time, the developer of the first computer in our history, John Van Neumann, recognized that computers could have an invaluable purpose in weather forecasting.

By the late 1950s, computer models were being run twice a day for forecasters. The users of these early model forecasts were the U.S. Weather Bureau (today's National Weather Service), the U.S. Navy, and the U.S. Air Force. Initially, forecasts were not that great; however, with human forecaster intervention, the forecasts from these computer models steadily improved.

Advancement in weather models was steady during the 1960s-1990s as satellite data and other technology helped mathematicians and scientists better model the atmosphere. When new computer resources became available, higher resolution models that required more data and space were substituted in for older models.

Here in the 21st century, meteorologists have several dozen computer models to look at. Within some of these individual computer models are other model forecasts with a slightly different output. These other model forecasts are referred to as a model ensemble. In other words, you start out with a single model and you perturb or alter the initial atmospheric conditions of that model. The end result is a five, ten, or maybe twenty or more different outputs of one particular computer model. So in reality, meteorologists have hundreds of computer weather models to choose from!

In terms of the different types computer models; tropical weather, severe weather, aviation weather, and air quality are just some fields in or related to meteorology in which modeling is essential. In addition, computer models can replicate the entire earth (a global scale model) or just a state or two in the U.S. (a Mesoscale or smaller scale model). In the United States alone, over 210 million observations per day (primarily from satellites) go into the computer models that forecasters use. The computation speed of all of this data is about 14 trillion calculations per second. As a result of this seemingly supersonic speed, nearly 15 million model fields such as temperature and precipitation are generated every day. Just imagine the computing power!

Looking ahead, weather prediction will continue to advance as more and more understanding of our atmosphere occurs. The suite for meteorological models is expected to expand well beyond weather to include other fields of study like the transportation, health, and energy communities. Many in the scientific community regard the implementation and advancement of weather computer models as one of the greatest achievements of the last century.

Meteorologists have gone from forecasting weather three days ahead to forecasting out 25 days. To project weather into the future, computer models are the tools that meteorologists use in the creation of a weather forecast. Newer computers with faster processors have been better able to tackle the immense number of variables that go into creating a good forecast.

Weather forecasting is difficult because the number of variables that must be factored into a good forecast are almost infinite. In the past, computers couldn't process the wind speeds, wind directions, temperatures and more, for multiple levels of the atmosphere. New technology makes that possible.

Paul Knight is a senior lecturer in meteorology at Penn State and also the Pennsylvania State Climatologist. He has been in the field of meteorology for about 38 years.

"Well, the computer models, of course, have revolutionized this field," Knight said. "What was just a field of simple extrapolation that was just moving things along back in the 1950s, when the computer models were just beginning, has become an entirely new field since 1966 when we first went with our first operational and numerical models."

Meteorologists have since developed computer models that take in initial conditions of the weather and factor in the dynamic behavior of weather. The weather does not operate in a linear fashion, and so complicated mathematics are used by computers to project weather into the future.

Today, a plethora of computer models, including Global Forecast System (GFS), North American Mesoscale Model (NAM), the European and the Canadian, project the weather for up to 16 days into the future. Current computer models give a look at many different levels of the atmosphere, so meteorologists can gain a full understanding of the weather patterns evolving.

Content contributed by AccuWeather Meteorologists Meghan Evans and Erik Pindrock.

Comments

Comments left here should adhere to the AccuWeather.com Community Guidelines. Profanity, personal attacks, and spam will not be tolerated.