We use wireless voice-call and text-message volumes to quantify spatiotemporal communication patterns in the New York Metro area before, during, and after the Virginia earthquake and Hurricane Irene in 2011. The earthquake produces an instantaneous and pervasive increase in volume and a ~90-minute temporal disruption to both call and text volume patterns, but call volume anomalies are much larger. The magnitude of call volume anomaly diminishes with distance from earthquake epicenter, with multiple clusters of high response in Manhattan. The hurricane produces a two-day, spatially varying disruption to normal call and text volume patterns. In most coastal areas, call volumes dropped anomalously in the afternoon before the hurricane's arrival, but text volumes showed a much less consistent pattern. These spatial patterns suggest partial, but not full, compliance with evacuation orders for low-lying areas. By helping us understand how people behave in actual emergencies, wireless data patterns may assist network operators and emergency planners who want to provide the best possible services to the community. We have been careful to preserve privacy throughout this work by using only anonymous and aggregate data.