Authors: Carolynne Hultquist*, Pennsylvania State University
Topics: Geographic Information Science and Systems, Remote Sensing
Keywords: hurricane, crowdsourcing, citizen science, disaster
Session Type: Paper
Start / End Time: 9:55 AM / 11:35 AM
Room: Cleveland 1, Marriott, Mezzanine Level
Presentation File: No File Uploaded
Decision making during disasters involves a multitude of factors and deals with uncertainty from models and observations. There are often significant gaps in spatial and temporal information from satellite imagery during disasters. During a hurricane, satellite remote sensing systems are usually unable to consistently collect imagery of the Earth’s surface due to cloud cover that is typically present when hurricanes make landfall. Like traditional methods for observation, social data is also situationally and locationally dependent. However, the gaps in social data are characterized by different conditions so social data might be able to be used to fill in spatio-temporal gaps in available remote sensing imagery. Citizen science data during hurricanes could meet information needs during the different stages of the disaster management cycle. This study compares citizen science flooding measurements to remote sensing data and traditional observations during Hurricane Florence. Remote sensing data from Landsat, Sentinel, and Planet Labs are collected as well as social data from various citizen science flooding projects and traditional flooding observation systems. The data sources are fused to assess what relevant information from various sources is available for impacted areas during hurricanes. Automated methods are developed for each source to classify flooded regions using machine learning techniques. Citizen science data can play a major role if there are overlaps of datasets which could be validated and places where citizen science collection occurs in spatio-temporal gaps of traditional government-produced data sources.