I won't make this too long, I promise, but it's worth doing a small post-mortem on Erika.
All week I had been telling you how disorganized she was and how dry air and wind shear were impacting the storm. Erika was resilient due to her larger size, but in the end the mountains of Hispaniola finally disrupted the storm's circulation to the point of dissipation.
You also heard me talk a lot about the difference between computer forecast models and reality. I think this is an important take-away from this experience. Every one of those squiggly spaghetti models on the map were *projections* for a possible future scenario. Yes, there is powerful science in those models, and thousands of man-hours invested in refining them. But at the end of the day each one of those lines bringing a hurricane to the Bahamas or Florida was imaginary.
In the virtual world of the models, Erika wasn't as disorganized as she was in real life. I don't think the models ever resolved the low-level swirl of a circulation that was frequently detached from the robust thunderstorm activity (and at times mid-level circulation) off to the southeast. Some global models like the GFS, ECMWF and Japanese did perform better than others because a lot of times they correctly forecasted dissipation of Erika over the Greater Antilles. But they weren't consistent in this forecast. Overall, the models erred in the short-term trajectory forecast (a LOT), and consequently they erred in the longer term trajectory forecast (a LOT), and they erred in the intensity forecast (a LOT). As a consequence, the National Hurricane Center erred in their forecast (a lot more than usual) too.
It's time every generation of operational meteorologists reassess and revisit old-school methods of forecasting. Models are ONE tool in the shed, not a crutch. Observing present trends via our many remote-sensing and in-situ platforms CAN help improve short and long term forecasts. Yes, tropical cyclones are notoriously finicky -- I also have made mistakes. But the human element must remain part of the equation so as to better serve our audience. Otherwise, we might as well leave the drawing of a "cone of possibilities" to an algorithm based solely on the consensus of the models. And, in my opinion, such "cone" should never be extended to 7-days.
Meanwhile Florida readies for significant rain from Erika's remnants while she crosses over the Gulf of Mexico. Several inches of precipitation could cause flooding. In addition there is a slight risk of tornadoes, and minor coastal flooding could occur due to the onshore wind flow.
