More Information on the Forecast Models

The mathematical models used at NHC are of three basic types:Statistical, Dynamical or Combination (Statistical and Dynamical together). Statistical models forecast the future by using current information about the hurricane and comparing it to historical knowledge about the behavior of similar tropical cyclones. The historical record for storms over the north Atlantic begins in 1871, while the record for storms for the east Pacific extends back to 1945. Dynamical models work differently. They are designed to use the results of global atmospheric model forecasts in different ways to forecast tropical cyclone motion and intensity. Global models take current wind, temperature, pressure and humidity observations and make forecasts of the actual atmosphere in which the cyclone exists.

If you read the hurricane discussions (described in the products subsection), you will see forecasters referring to different models and giving their reasons why they think particular models are doing a better job of representing current and future conditions. Acronyms for models commonly discussed are GFDL, CLIPER, AVN, LBAR, BAM, NOGAPS (the U.S. Navy's model), and UKMET (the model run by the United Kingdom's Meteorological Service). These are all track models. Intensity model acronyms include SHIFOR, SHIPS, and GFDL.

It is important to know that these models are only run a few times a day and cannot, therefore, take into account all of the short-term changes the atmosphere is constantly undergoing. Models cannot produce forecasts more frequently because they require huge amounts of data and long computational times. This is one source of forecast error (discussed more fully in the Uncertainty section).

Hurricane forecasters must look at all of the models' results, which frequently give widely different pictures of the future. When the models do disagree, hurricane forecasters must use their experience and judgment to decide which model is performing the best under the current conditions. Unfortunately, we are not at a point in this science where one model can reliably be used for forecasts in all the different situations that can occur given the complexity of our atmosphere.

A good forecaster has an extensive education in the science of meteorology and considerable experience in tropical forecasting. Nonetheless, many times the different data sources are too conflicting for forecasters to have a high degree of confidence in their predictions. Even when they are more sure, forecasters still recognize that conditions can change quickly. This is why forecasts talk about "probabilities" and "margin of error". This is also why emergency managers consider planning for a hurricane one category higher than is currently forecast and why they prepare in advance to take action in case the track shifts suddenly or the storm speeds up as was the case with Hurricane Opal, 1995. For More Information on the Forecast Models