Skip to main content

Differentially Private Evacuation Rates Dashboard

Discover mobility trends observed during catastrophic events

During natural disasters, such as hurricanes and wildfires, it is critical to understand the rate at which residents are evacuating impacted areas and where they are going. Spectus’ Evacuation Rate dashboards report the percentage of residents who have evacuated their homes by county, in addition to the average distance traveled by evacuees, and the destination counties traveled to by evacuees who left their home county.

Furthermore, Spectus analyzes evacuation rates by income group in order to provide insights into the impact of natural disasters on various income segments. By providing these insights, Spectus’ aim is to support emergency managers and public planners to better prepare for and respond to natural disasters.

Hurricane Ian Evacuation Analysis

Hurricane Ian made landfall in Fort Myers, Florida on Wednesday, September 28th at 3:10pm as a Category 4 hurricane. With wind speeds up to 150mph (241km/h), Ian caused destructive storm surges and left millions of people without power.

Evacuation Routes

Hurricane Irma Evacuation Analysis

Hurricane Irma made landfall in Cudjoe Key, Florida on Sunday, September 10th at 9:10am as a Category 4 hurricane. With wind speeds up to 180mph (285km/h), Irma caused widespread destruction and inflicted $77 billion worth of damage.

Differential Privacy

In order to improve the utility of evacuation analyses without sacrificing user privacy, Spectus has applied Differential Privacy to the computation of all evacuation metrics. Differential Privacy is the application of statistical noise to processes, in order to obfuscate user-level data points while maintaining the overall utility of the aggregate insights.

Our datasets are generated using the SmartNoise differential privacy toolkit, which is developed by Microsoft and Harvard University and is part of the OpenDP open-source software project. You can learn more about SmartNoise and OpenDP at the OpenDP Smartnoise repository.