GRAND RAPIDS, Mich. (WOOD) — If you spend enough time listening to weather reports on the air, at one point or another you’ll likely hear broadcast meteorologists make mention of forecast models.
Our viewer question this week comes from Kevin, who asked, “With all the forecast models you have today, how do you know which one to use for your forecast? And is one considered more accurate than others?’”
Forecast models are one of the top tools that meteorologists use, and there are several different forecast models that are available. On a daily basis, Storm Team 8 meteorologists use the American model (GFS), the European model (ECMWF), the High-Resolution Rapid Refresh Model (HRRR), the North American Model (NAM) and the Rapid Precision Mesoscale model (RPM), just to name a few. Interpreting the differences between forecast models to create an accurate forecast can be a big challenge.
Each model starts with an input, which consists of current weather conditions such as pressure, temperature, humidity, etc. The data is then put into various mathematical equations, which characterize the physics of the atmosphere. The mathematical equations are applied to the input, and then stepped forward in time. The result is an interpretation of how the weather will play out over the next period of time.
There are a few reasons why these forecast models can have different outcomes. First and foremost, each model uses slightly different mathematical equations. Even if a model starts with the same initial conditions, the output will be different due to the different equations used.
In some cases, the initial data is inconsistent between the forecast models. If a weather sensor has an error and sends out faulty data, then the information provided by the forecast model will be skewed. Since forecast models are generated at different times and frequencies, there is a chance of some forecast models receiving faulty data at one point in the day, while other forecast models get good data just a few hours later.
Forecast models also have different resolutions. High-resolution models may only produce a forecast for the next 36-48 hours, but they are better at forecasting small weather features such as thunderstorms or lake-effect snow bands.
Larger resolution models aren’t as precise at forecasting the small weather features, but they forecast much further out than just 36-48 hours. These models give meteorologists a good idea of how large scale, synoptic weather features will develop over the coming days and even weeks.
So, to answer Kevin’s question … it is tricky to know which forecast model to use! We use the high-resolution models to fine tune weather events that will happen over the next day or two and the long-term lower resolution models to see what weather systems will impact us during the next eight days.
There is no one model that has been proven most accurate. The models can have varying accuracy, depending on the type of weather situation and location.
Meteorologists generally look at all possible outcomes from the forecast models, then depend on their previous experiences and knowledge of a specific geographic point to make a decision.