The first thing to ALWAYS look at with ANY weather model is the initiation time and valid date and time. The initiation date/time tells us on which day the model was run and at what time. The valid date tells us what specific day/time you are looking at within a specific model run. These times are important because a lot of the time the most recent model run will be the most accurate. This is because weather models get less accurate with time because they are dynamic models.
The official forecast track of a hurricane is released by the National Hurricane Center in Miami Florida. These are meteorologist who specialize in tropical meteorology and hurricanes. The last opening they had in the office I read somewhere that they had over 900 applicants for the job. The NHC releases forecast tracks like the one below. The black line is the forecast rack with the white being the cone of error.
The track of the center of the storm can be anywhere within the white cone, which means even the areas outside the cone can see hazardous conditions. I added orange arrows to show the potential tracks of Irma. As I mentioned before the further out the model run goes the less accurate/more error that there will be. We can see this in the forecast track of Irma, the tighter the white cone the smaller the error. Which explains why the further out the track goes the wider the potential error.
Another image likely to be seen during a hurricane will be the spaghetti models. These can be confusing and misguiding if you aren't familiar with how models work. Below is the latest spaghetti model run at the time I am writing this blog post. Each line is a model run for the track of the storm. The closer the lines are together the higher the confidence in a track for the storm.
It is important to NEVER create a forecast track for any type of storm from a single model one. Noise can get into a single model run and make it invalid. Persistence forecasting using multiple model runs is the best way to forecast the track of a storm. Persistence forecasting is using multiple model runs to look for trends. If the last 3 model runs have Irma tracking East and the latest model takes it West into the gulf it is best to look at other models initialized at the same time period and later model runs before going against what has been the persistent Eastward movement in the latter model runs.
The above pictures are model runs starting at 9/5 12z and ending at 9/6 06z. Going forward in time we can begin to see an Eastward shift in the track of Irma. This is an example of how to use persistence modeling.
