07-24-2016, 03:07 AM
As I pointed out at time, it is difficult to predict turns in slow moving hurricanes. The GFS model turned out to be quite accurate for the 72 and 48 hour forecasts however the forecasters went with a composite prediction.
All hurricanes are steered by the upper atmosphere. However when they are slow moving or there are very tall mountains (10,000 feet or higher) there are no models that have a high success rate at predictions.
Weather forecasters who do this professionally like at NOAA are not just "second guessing and getting lucky". They know there are statistical weaknesses in the models. They assess the computer prediction and try to see if that makes sense and adjust the forecast accordingly -- including the error boundaries.
In reality the prediction error was well within the cone of their error boundaries. This means they both understand their models and have correctly estimated how much they models may fail to predict. This is way different from someone saying I think it will miss us because it feels like it.
The approximate error for landfall was about 45 miles, which is quite good considering how slow the storm moved, how disorganized it was and the interaction with the mountains.
All hurricanes are steered by the upper atmosphere. However when they are slow moving or there are very tall mountains (10,000 feet or higher) there are no models that have a high success rate at predictions.
Weather forecasters who do this professionally like at NOAA are not just "second guessing and getting lucky". They know there are statistical weaknesses in the models. They assess the computer prediction and try to see if that makes sense and adjust the forecast accordingly -- including the error boundaries.
In reality the prediction error was well within the cone of their error boundaries. This means they both understand their models and have correctly estimated how much they models may fail to predict. This is way different from someone saying I think it will miss us because it feels like it.
The approximate error for landfall was about 45 miles, which is quite good considering how slow the storm moved, how disorganized it was and the interaction with the mountains.