On Friday, a very complicated meteorological situation unfolds, as Sandy interacts with a trough of low pressure approaching the U.S. East Coast and trough of low pressure over the Central Atlantic. The Central Atlantic trough may be strong enough to pull Sandy northeastwards, out to sea, as predicted by the official NHC forecast, and the 06Z GFS, 00Z UKMET, 00Z Canadian, and 06Z HWRF models (00Z is 8 pm EDT, and 06Z is 2 am EDT.)
However, an alternative solution, shown by the 00Z ECMWF, 06Z GFDL, and 06Z NOGAPS models, is for Sandy to get caught up by the trough approaching the Eastern U.S., which will inject a large amount of energy into Sandy, converting it to a powerful subtropical storm that hits the mid-Atlantic or New England early next week with a central pressure below 960 mb and sustained winds of 60 - 70 mph.
Such a storm would likely cause massive power outages and over a billion dollars in damage, as trees still in leaf take out power grids, and heavy rains and coastal storm surges create damaging flooding. The full moon is on Monday, which means astronomical tides will be at their peak for the month, increasing potential storm surge flooding. A similar meteorological situation occurred in October 1991, when Hurricane Grace became absorbed by a Nor'easter, becoming the so-called "Perfect Storm" that killed 13 people and did over $200 million in damage in the Northeast U.S.
Figure 3. The Wednesday morning 06Z (2 am EDT) run of the GFS model was done 20 times at lower resolution with slightly varying initial conditions of temperature, pressure, and moisture to generate an ensemble of forecast tracks for Sandy (pink lines). These forecasts show substantial uncertainty in Sandy's path after Friday, with a minority of the forecasts taking Sandy to the northeast, out to sea, and the majority now predicting a landfall in the Northeast or mid-Atlantic states of the U.S. The white line shows the official GFS forecast, run at higher resolution.