Hurricane Sandy made landfall in one of the most populated areas of the United States, and affected almost 8 million people. The event provides a unique opportunity to study power outages because of the data available and the large impact to a densely populated area. Satellite nightlight imagery of "before" and "after" the landfall of the hurricane is used to quantify the light dimming caused by power outages. Geolocated tweets filtered by keywords provide valuable information on human activity at a high temporal and spatial resolution during the event. Analysis of brightness change in the satellite data and the density of power related tweets points to a spatial relationship that identifies severely impacted areas with human presence. Classification of tweets through text analysis serves to further narrow the information search to find the most relevant and reliable content. Twitter data fused with satellite imagery identifies power outage information at a street-level resolution that is not achievable with satellite imagery alone.