Case Study: Hurricane Sandy

This post-event analysis will detail the performance of the Baron Model’s Precipitation and Wind products in forecasting the path of Hurricane Sandy prior and up to landfall. The largest Atlantic hurricane and second most costly storm in U.S. history, Hurricane Sandy has left communities from North Carolina to Massachusetts reeling.

This case study inspects two components of the storm in regards to the Baron model and the actual path of the hurricane. First is the forecast track at 96 hours out from landfall, and the second is the forecast location of landfall at 30 hours out. These two selected examples of the 3km and 15km Baron model at October 25th 18z and October 28th 18z are representative of the overall accuracy of all the model runs from 96 hours beforehand until landfall.

sandy1

October 25 – 96 Hours from Landfall
On Thursday, October 25th, most models had projected the path of Hurricane Sandy to track mostly northward, parallel to the Eastern Seaboard, and remaining well out at sea with only peripheral effects reaching land. The 15km Baron model predicted a significant curvature in the track that would have the eye of the storm making a direct hit on the coast, as seen in the image above. Hurricane Sandy did, in fact, make an abrupt shift in direction toward land.

October 28 – 30 Hours from Landfall

The landfall time and location that the Baron model projected 30 hours out proved to be extremely accurate. When the storm made landfall on the New Jersey southern shore on October 29th (left), the prediction from the 3km Baron model (above), had projected the arrival to occur only a few minutes earlier than actual landfall. In these images it is also possible to see how accurately the 3km Baron Precipitation product predicted the location of snowfall over the Appalachian Mountains.

sandy2sandy3

Looking at a close-up of the landfall location (below), we can see that the BAMS October 28th 18z model run, when compared to the reflectivity data from October 29th, was definitively correct in both the timing and placement of the storm.

sandy4sandy5

Overall, the 15km and 3km Baron models proved their accuracy over the course of the storm, accounting for erratic track changes over the four-day period, and pinpointing the location of landfall 30 hours ahead of time.