Hurricanes through the Location Lens: Harvey vs Irma

During Harvey, data showed a 7X increase in population density in residential areas, whereas population dropped to practically zero in Florida Keys during Irma.

Data can save lives. Real-time, accurate data can save even more lives.

In the past, data on the movement and actions of populations during natural disasters used to come long after the events and relied heavily on self-reporting. Today, we have the ability to collect vital data faster in real time based on cell phone location data, and make decisions that could keep more people safe.

In the past few months, two Category 4 hurricanes lashed heavily populated areas along the Gulf Coast during the most active hurricane season in more than a decade. Harvey (Category 4) hit Houston on August 25 and Irma (Category 4) hit the Florida Keys on September 10.

When we compare the data on how people responded to two of this year’s most massive storms, here’s what we found:

Human Behaviour Before and During the Storm —

Fight (Harvey) vs Flight (Irma):

For Harvey, we compared population data on August 16 with August 27 as seen below. We saw a 7X increase in population density in the residential areas on the city outskirts. While people left the city center, most stayed in their homes if they were on higher ground.

Comparison of People Density: August 27 vs August 16

People Density Change During Harvey

For Irma, we saw the population data drop to practically zero in the Keys and noted substantial drops in densely packed neighborhoods in Miami. Mandatory evacuation was ordered in Florida Keys, and partial mandatory evacuation was ordered in Miami.

Florida Keys Before and During Irma — September 4 vs September 11

Miami Before and During Irma — September 4 vs September 11

Human Digital Behaviour Before and During the Storm —

React (Harvey) vs Respond (Irma)

Houston saw heavy drop in app traffic post August 27 due to loss of power, and drained mobile batteries. By August 29, as power was restored, app traffic returned to average levels. Orlando saw an increase in app traffic as the storm hit Florida and it dropped abruptly post September 11, when the storm arrived.


  • Houston saw a 3X increase in weather app users during Harvey.
  • NOAA was one of the top 10 apps used from August 25 to August 29 to track the storm.
  • With people having to brave the storm without power, flashlight usage saw a 2X increase.


  • When evacuation announcements were made, Florida Keys and Miami saw a 2X increase in the number of users using weather apps compared to the start of the month.
  • Orlando, being outside the evacuation zone saw a 3X increase in usage of weather apps by September 11.
  • NOAA was one of the top 10 apps used before the storm made landfall.
  • Flashlight usage was not seen.

In the end, 82 people died as a result of Harvey and 22 people died as a result of Irma. While neither comes close to the 1,833 casualties caused by Katrina, preparations based on better data and timely communication can definitely reduce the danger to life.

Contact Us to use data for superior decision-making.

Also published in Medium.