Weather Models are nothing more than computer programs which try to forecast the weather. A huge amount of data such as temperatures, pressure and winds are “fed” into the model to get it started.
Once the model run begins, it generates forecasts along a time-line. Graphical (maps)and tabular (text) output of these models is usually available within an hour after the model begins for shorter range (0-3 day) models, and within six hours for longer-range (up to 16 day) products.
Since three-fourths of the Earth surface is ocean, satellite measurements give us a wealth of information her. This should allow more frequent model runs and greater accuracy over the next five to ten years. However, there are limits and problems with the models.
Since each one is a different mathematical “guess” at what’s going on, there are errors built in. Many hilly or mountainous areas contain too much terrain data for the model (or computer) to handle. The solution is to craft “average” ground heights for different areas. This will inevitably lead to errors at times. A model may have equations which are used to estimate temperatures inside a complex of thunderstorms, or over snow-covered ground, or any number of situations where assumptions in the original program make that model less reliable in a given situation. This is known as model “bias”, or a tendency to be wrong in a specific way given a particular weather scenario.
Another problem is with quality control. Bad data gums up the works real quick.
Example: A model starts with an incorrect wind speed at the jet stream level, say 80mph instead of 180mph. This single “bad observation” can muck up the forecast for thousands of miles.
These errors are usually flagged and kicked out, but then the model output is no good since the real value will not be known until a new “run” comes out in the next 12 hour cycle. Sometimes an estimated value can be inserted based on surrounding data (interpolation again), with varying results.
So the next time you hear about a weather “model”, remember that it’s just a sophisticated piece of computer software which is guessing at what may happen. A conscientious forecaster tracks several different models, and then combines that with known weather science, the current data, and their own experience to arrive at a good forecast.
One can’t simply look at a few models on the Internet and be qualified as a forecaster anymore than playing an NFL video game makes you ready for the field. It takes years to learn the ropes and even then there will be the occasional “busted” forecast.