Quick question - when you analyze a complex storm like this and have to nail down an exact forecast, what's the mechanism? Do you just blend model output or factor in some forecasting experience that no model can reproduce? If the latter, what's the relative weighting? Just curious how it works in your world. Thanks!