The 1938 hurricane is well known to not only locals of New England, but also to the (re)insurance industry. Very often, many ask "what would happen if the storm were to occur again?", in light of not only the dramatic coastal and financial growth in the region, but also the dramatic improvement in technology. Further, we still do not know exactly what the structure was of the storm as it made landfall: pure tropical cyclone, transitioning tropical cyclone, post-tropical cyclone, or the most deadly (but rare) warm-seclusion cyclone?

The presentation begins by analyzing trajectories from 1957-2002 using ERA40 reanalysis to quantify the rarity of a parcel of air originating along the hurricane's track eventually reaching southern New England. This analyzes in a generic way the frequency with which the overall large-scale pattern supports a flow into New England from the source region. The results show that at most 1-2 days a year is such a path permitted. Combination with the rarity of having a tropical cyclone in that location (at the same time), the rarity of the New England landfall and track from the SSE is illustrated.

New global reanalysis grids from NOAA/CIRES (20th Century Reanalysis Project) uses surface-only data to produce global 3D atmospheric grids back to the 19th century. This permits for the first time numerical simulations of the 1938 hurricane. Further, the reanalysis project produces 56 ensemble members that permits quantification of the uncertainty of the numerical simulations, in terms of track, intensity, and structure, before and after landfall.

The presentation concludes with an examination of the variability of this 56 member of track, intensity, and structure ensemble. The structural analysis in particular is intriguing, since it argues for a predominance of the (most dangerous) warm-seclusion lifecycle into and beyond New England, which is more typically seen over the maritime far north Atlantic. However, the structural uncertainty overall of the cyclone is far greater than that seen for ensemble simulations of other New England hurricanes (e.g., Donna, Gloria, Bob, Floyd). Landfall timing is approximately nine hours late (the actual storm landfalls 9 hours earlier than the model forecast). This timing bias is not only consistent with operational forecasts of the prior mentioned storms, but also with a nightmare scenario where emergency preparedness and evacuations are not completed before the storm makes landfall at an intensity and speed likely in doubt based upon long-term climatology.