Using Social Media and GIS to Estimate Flood Extent and Assess Risk of Critical Infrastructure

As a flood occurs, it often takes weeks and even months after the event to determine the extent of the flooding. For many, it is imperative to receive information about the extent of flooding in a more near real time sense, on the scale of hours. Conventional methods used to determine flood extent, such as high water marks and remote sensing, may not be useful to for estimating the extent of flooding on the time scale necessary for first responders and decision makers. Perhaps a less conventional method, such as Twitter, may be able to provide the timeliness of flooding information while still maintaining accuracy. To test this, geo-located tweets related to flooding were gathered for major flooding events in South Carolina (October 2015), Houston, TX (April 2016), and Monroe, LA (March 2016) and mapped by their location using a Geographic Information System (GIS). Using the filtered and mapped tweets along with a high resolution digital elevation model (DEM), a risk map was developed for critical infrastructure (i.e. hospitals, energy sources, schools, etc.) in each study area. The filtered and mapped tweets are compared against official flood extents/reports from FEMA, USGS, and local government agencies to test the accuracy and feasibility of using Twitter as a means for determining flood extent.