Extended Abstract (820K)1.5
Storm of the Century? Insights from a massive ensemble forecast experiment
Peter S. Dailey, AIR-Worldwide, Boston, MA; and R. G. Fovell
In March, 1993, a "Nor'easter" subsequently dubbed "Superstorm 1993" and the "Storm of the Century" (SOC) brought paralyzing snows and damaging winds to a large portion of the Eastern United States. True to its moniker, the SOC event ranked first on Kocin and Uccellini's list of 70 storms over a roughly 100 year period, as measured by their North East Snow Impact Scale (NESIS). But was the SOC just a one in a century storm or a more or less rare event? Many parties, particularly property insurers, are interested in this question, but the answer is not clear from the all-too-short historical record.
There is little doubt that the SOC was born to be a very significant event. It occurred during a period in which the Available Potential Energy (APE) in the Northern Hemisphere was far in excess of climatological values. But was the SOC as strong as it could have been? Given substantially similar initial conditions, what is the probability that an even more severe Nor'easter could have been produced? In a nutshell, we inquire regarding the recurrence interval (so-called return period) of the SOC event.
We are investigating this issue with a massive forecast ensemble, consisting of 325 members at this writing. Ensemble members are created by introducing random perturbations into the three-dimensional initial conditions. Thus far, simulations are made using MM5 employing data from the ECMWF reanalyses; and the authors are in the process of expanding the ensemble to include other models and data sources. To enhance ensemble diversity, we are employing a variety of lead times. Selection of the ensemble member representing "truth" is an issue considered very carefully. We find that relatively few of the ensemble members are able to exceed the truth case with respect to event integrated snow production, implying a relatively long recurrence interval. The analysis can be conducted from the perspective of maximum wind intensity as well. Lessons learned from this massive and expanding ensemble will be discussed, and ideas for future ensembling efforts will be outlined.
Session 1, Mesoscale Predictability
Monday, 6 August 2007, 8:40 AM-10:15 AM, Waterville Room
Previous paper
Recorded presentation