Hurricane Florence made landfall at Wrightsville Beach, North Carolina, on the morning of September 14, 2018. Although it had been downgraded to a Category 1 hurricane by that time, Florence’s extremely slow progress and broad circulation resulted in historic flooding, as well as wind damage and power outages across the Carolinas and Virginia. Though less severe, flooding and damage was seen as far north as New England.
To enhance situational awareness for first responders and emergency managers, National Alliance for Public Safety GIS Foundation (NAPSG) Programs Manager (and GISCorps volunteer) Paul Doherty developed a web-based application to facilitate collection and mapping of on-the-ground photos of hurricane-affected areas. Based on Esri’s Story Map Crowdsource template, the NAPSG 2018 Hurricanes Crowdsourced Photos app allows volunteers and the public to post photos of affected areas, geolocate those photos to the town level or better, and provide brief descriptions and source information.
NAPSG requested support from GISCorps to accelerate the crowdsourcing effort. Thirty-six GISCorps volunteers answered the call and volunteered a total of 335 hours scouring social media and news outlets for photos depicting conditions across the affected area. For each photo, volunteers pinpointed the location of the image using visual clues and accompanying descriptions, uploaded the photo to the Crowdsourced Photos app, and collaborated round the clock in a Slack channel.
In further support of this effort, Rob Neppell of Crowd Emergency Disaster Response (CEDR) Digital Corps configured an experimental Slack channel populated by Zapier with auto-gathered tweets based on geographic place names and relevant keywords. GISCorps member Erin Arkison developed a custom dashboard to motivate and track volunteer efforts.
A small team of GISCorps volunteers (Erin Arkison, Dacey Zelman-Fahm, Miguel Leon, and Holly Torpey) took shifts vetting each photo contribution for relevance, locational accuracy, and attribute completeness using the Esri Crowdsource Manager application.
By the end of the mission, volunteers had logged more than 900 pictures and the map had been viewed over 12,000 times. According to Paul Doherty, many emergency response agencies used the map to inform their response strategies, including federal emergency managers, the U.S. Coast Guard, State Emergency Operation Centers in Virginia, North Carolina, and South Carolina, and federal, state, and local search and rescue teams.
In early October, NAPSG and the Pacific Northwest National Laboratory (PNNL) invited a handful of participating GISCorps volunteers to return for a second phase of the mission. In this phase, volunteers revisited every photo that had been collected in the first phase, tagged those that showed evidence of flooding, and estimated floodwater depth using PNNL floodwater modeling categories. PNNL scientists then used those observations to help validate their flood models.
GISCorps volunteers Paul Doherty, Erin Arkison, Dacey Zelman-Fahm, Jason San Souci, Miguel Leon, Holly Torpey, and German Whitley contributed to the PNNL flood model validation effort, blazing through all 900 photos in under 6 hours.