Modeling and simulation is dependent on setting conditions, standards, and limitations at the start of the modeling process. Historically, modeling and simulations do NOT deal well with emergent "pop-up" events i.e. "reality" which can skew the validity and/or fidelity of the model results.
For example, I would love to see the sensitivity analysis of their model results to a Category 3 storm tracking across the Caribbean, up the Gulf Stream, and going ashore between Wilmington and Cape Hatteras (not an uncommon storm track). Now look at a Category 4 tracking across the Keys and show me the deltas in the model results. Are they significant? If so, then the current predicted results may not be worth a warm bucket of spit.
(For a quick&dirty study of M&S limitations, consider the recent controversy over grounding air traffic in Europe based on atmospheric modeling of the Icelandic volcano.)
As I've reminded my M&S people: "Modeling and simulation is nice, but reality bites."
I call BS on this article. Common sense says dilution is the solution. You will never notice it outside the Gulf.