On Sunday, an unusually intense autumn storm swept through the Northeast US, dumping over 80mm of rain in a few hours and generating gusts exceeding 120km/h. Over 30,000 homes were left without power, and one death was reported in New Jersey. According to the European weather service ECMWF, this event ranks among the most violent ever observed in September, highlighting the increasing importance of seasonal anomalies linked to climate change.
Storm Intensity: Precipitation, Winds, and Affected Areas
Satellite data from Copernicus showed that the low-pressure system rapidly intensified over the Atlantic, reaching a central pressure of 980hPa before making landfall. Weather stations in New York, New Jersey, and Connecticut recorded rainfall accumulations exceeding 80mm, with coastal areas seeing peaks of 110mm. Winds measured at over 120km/h on Long Island uprooted trees and caused significant material damage, including torn-off roofs and downed power lines.
Predictive models based on machine learning, such as GraphCast and Pangu-Weather, accurately anticipated the system's rapid intensification with over 85% precision up to 24 hours before impact. These neural networks, trained on decades of atmospheric data, identified a sudden increase in vertical moisture and strengthening of the jet stream over the Gulf of Maine. Despite this accuracy, forecast uncertainty remained high for local precipitation, leading authorities to issue maximum alert warnings in five most-at-risk states.
