Geospatial assessment of wetness dynamics in the October 2015 SC flood with remote sensing and social media

Real-time data on flood extents and dynamics are important for risk assessment and emergency response during the event. While real-time imagery is often unavailable due to heavy cloud cover during a flood, remote sensing platforms can be used to monitor its development through a synoptic view. Record rainfall occurring October 1-5, 2015 in coastal South Carolina caused the October 2015 South Carolina Flood. The Congaree River Watershed downstream of Columbia, SC experienced historic flooding. This study utilizes two satellite images acquired on October 8^th (EO-1 ALI) and 18^th (Landsat8 OLI) to examine flood dynamics. Using a normalized difference wetness index (NDWI), the flooded and highly wet areas were extracted. Social media, such as Twitter, from public users allows quick awareness of floods in an area, but geolocation is not necessarily accurate. Assisted with real-time Twitter data, satellite images after a flood helps to assess water retreat and potential risks for emergency responders. Since social media data sets are big, highly unstructured and noisy in nature, sophisticated data mining algorithms are needed for the verification process from millions of tweets in a region. When automatic tweets verification approaches are available, integrating social media into geospatial science could become important data sources for disaster assessment and management.