Find out what a weather model is and how it helps meteorologists make forecasts.
MOLINE, Ill. — Prior to yesterday’s snow event, meteorologists were showing you different patterns when forecasting possible outcomes. Models are tools we use to help us make predictions, but we don’t rely on them alone. In this article, you will learn about weather models and the different types we use to help us with our forecasts.
What is a weather model?
Weather models are computer programs that help predict the weather. These models use a set of physics-based mathematical equations that characterize how air, heat, and moisture move through the atmosphere. The sensors are used to collect weather observations such as temperature, wind and pressure to put these equations to them. This is called data assimilation.
The atmosphere is divided into each 3-D grid and each grid point receives the assimilated data. The first points of the grid are where the assimilated data is placed, called initial conditions. As time progresses, mathematical equations are recalculated and reapplied for the new time. This continues depending on how far the model is programmed to go, which can be anywhere from hours to weeks in advance.
No weather model is perfect. Some problems with weather models are that we cannot know what the weather is like at every point on Earth. As a result, we often have to spread our grid points over hundreds of miles, creating gaps in weather information. The sensors we use to collect meteorological parameters may encounter an error or failure which may then give us a bad result. The model can predict hot or cold.
For these reasons, there are many different types of models. There are long and short term, regional/mesoscale or global models. Some models are used for a specific weather event or parameter. Different mathematical equations are used for different types of models depending on what they are supposed to show. A meteorologist must take all of this into account when making forecasts.
Types of weather patterns
There are global models and regional models. Global models cover the world and do not have such a clear resolution. Mesoscale/regional models have higher resolution. Mesoscales are weather events that occur on a smaller scale, such as thunderstorms, squall lines, or land-sea breezes. Each model has its advantages and disadvantages.
The global model types are the Global Forecasting System (GFS) and European Center for Mid-Range Weather Forecasting (ECMWF) meteorologists usually call it the Euro model.
The GFS is an American model and extinguishes 384 hours. A concern of the GFS is that it can exaggerate or overestimate long term systems, systems that are days or weeks away, as it can predict up to 16 days, to be stronger than they are.
Most meteorologists, when watching the weather ahead, act cautiously when watching the weather so far away so as not to have hope. Overall it’s a good model and it’s widely used by meteorologists.
The Euro (EMWF) goes up to 240 hours, which is the equivalent of 10 days. A concern with this model is that it has cutoff troughs in the southwestern United States, overamplifies upper-level systems, and exhibits biases in the upper troposphere. The last two concerns relate to the upper part of the atmosphere which can then affect the weather at the surface.
World models are used a lot for long term forecasting, trying to spot future trends and upcoming patterns. They can be used for short-term forecasts, but the coarse resolution they would provide can sometimes cause problems.
Regional and mesoscale models include the North American Model (NAM), Rapid Refresh Model (RAP), and Weather Research and Forecasting Model (WRF).
The NAM comes out 84 hours or three and a half days. Some concerns are between 72 and 84 hours, where there is back and forth on what to expect. The video above shows my forecast for Sunday at 10 p.m. before the snow arrives late Tuesday evening and early Wednesday morning. I talk briefly about the change in the NAM from when I did the weather forecast during the 5 p.m. newscast, which falls into the range of flip-flops on what’s to come. Other concerns are that it is not always so accurate with near-surface parameters such as temperatures or precipitation.
The RAP comes out at 6 p.m., but is concerned about its low accuracy in mountainous areas. The WRF comes out 36 hours, but it can predict too low temperatures in the Southeast and predict too much precipitation.
Regional/mesoscale models have higher resolutions and use observations, so they are ideal for short-term forecasts. They can provide more detail about a situation once it gets closer than a long-range model, which helps the meteorologist better determine what will happen as the weather event gets closer.
Models are great tools to help meteorologists make predictions, but we don’t rely on them alone. We use them in combination with what we learn in school about the atmosphere and how it works.
There are times when we decide to go against the patterns because we know what we see is not what we know.