Different Types of Weather Models | WNYT.com

Different Types of Weather Models

Allison Finch
Updated: October 01, 2020 09:59 AM
Created: October 01, 2020 08:25 AM

Upper Air Data Collection Sites | Storm Prediction Center Upper Air Data Collection Sites | Storm Prediction Center
 When forecasting the weather, we rely on many different weather models. Weather models use mathematical models of the atmosphere, oceans, and current data, to predict the weather. While the first numerical weather prediction (NWP) model was attempted in the 1920s, it wasn’t until the 1950s when computer simulations were used, that these models began to output a realistic forecast. One of the fundamental problems with forecasting, is that even the most powerful computers in the world can’t solve the chaotic nature of the partial differential equations that govern the atmosphere. These equations are impossible to solve exactly, which leads to small errors that over time grow into bigger errors. Even the most perfect weather model with the most up-to-date weather data will only be accurate for about 14 days ahead. What makes the equations and problems so hard to solve is the amount of parameterizations that need to be accounted for. Some of these parameterizations include solar radiation, heat exchange, surface water and moist processes, like clouds and precipitation. The input of real-time weather data from observations across the world helps each model predict better.  

Weather observations are most commonly taken by weather balloons that are launched by the National Weather Service (NWS). These balloons collect critical upper-air weather information such as atmospheric pressure, temperature, wind speed and direction, and humidity, with an instrument called a radiosonde. They are launched twice a day at the same time all across the United States. The first one is launched at 12Z (Zulu time). This corresponds to 8AM EST (during daylight saving time) and 7AM EST (not during daylight savings time.) The second balloon is launched at 00Z and this corresponds to 8 PM EST (daylight savings time) and 7AM EST (not daylight savings time).

The reason forecasters use Zulu time is so that we are all under the same period. For example, if you were told to go outside at 6pm, you would go outside when it was 6pm in your time zone. When all locations around the globe are following the same time zone (i.e. Zulu Time), then everyone goes outside at the exact same time.

The information collected from these weather balloons is used in forecast models all across the globe. In the United States, 92 different locations launch weather balloons. When impending weather is approaching, some stations may launch one or two extra weather balloons so that they have a better understanding of what is going on in the atmosphere.

Global Models | WNYT Global Models | WNYT
 You can break all the weather models in the world down into two different categories; global models and mesoscale/regional models. The two most common global models are as follows:

Global Forecast System (GFS)

  • Resolution: 28km
  • Forecast 384 hours (16 days)
  • Updates every 6 hours, with new weather data being input every 12 hours

European Center of Midrange Weather Forecasting (ECMWF)

  • Resolution: 9km
  • Forecast 240 hours (10 days)
  • Updates and inputs new weather data every 12 hours

With broad resolutions, this makes pin pointing the exact location of a storm a bit tricky. Both of these models forecast for a long period of time. The accuracies of the forecast after 7-day starts to decrease, but comparing and contrasting the differences between these two models (after 7 days) allows forecasters to get a good understanding of incoming weather. Most of the time these models over-estimate the amount of rain or snow for a given location. 

Mesoscale/Regional Models | WNYT Mesoscale/Regional Models | WNYT
Beside the domains being different between global and mesoscale models, another big difference is the amount of time mesoscale/regional models forecast for. The mesoscale/regional forecast models use a short forecast time as well as finer resolutions which allow for more precision forecasts and less errors. Some of the most common models that fall under the mesoscales/regional classification are as follows:

Rapid Precision Model (RPM)

  • Resolution: 4km
  • Forecast 48 hours out (2 days)
  • Updates every 3 hours, with new weather data being input every 12 hours

High Resolution Rapid Refresh (HRRR)

  • Resolution: 3km
  • Forecast 18 hours out (Not even a full day)
  • Updates every hour, with new weather data being input every 12 hours

Global High-Resolution Atmospheric Forecasting System by IBM (IBM GRAF)

Note that this is a global forecast, but the resolution and short forecast period make it resemble a mesoscale/regional forecast model.

  • Resolution: 3km
  • Forecast 14 hours out (About a half of a day)
  • Updates every hour, with new weather data being input every 12 hours

In the summertime, these models are used to help pin point the exact location of severe weather or rain showers. In the wintertime, these models help forecasters determine where the freezing line is, and where or how much snow could occur at any given location. 

All of these models receive new weather data every 12 hours, unless there are more than two balloon launches in one day. When this happens, weather models are updated with the most recent weather data to help create the most accurate forecasts. Lastly, it is unlikely that one model will get the entire forecast correct, so forecasters use a combination of different models to help accurately forecast the weather.


Copyright 2020 - WNYT-TV, LLC A Hubbard Broadcasting Company

Comment on Facebook