Authors: Edwin Chow*, Texas State University - San Marcos, Abdullatif Alyaqout, Texas State University
Topics: Geographic Information Science and Systems, Hazards, Risks, and Disasters, Quantitative Methods
Keywords: spatial accuracy, error, twitter, water depth, flood
Session Type: Paper
Start / End Time: 8:00 AM / 9:40 AM
Room: Forum Room, Omni, West
Presentation File: No File Uploaded
During a disaster, people share critical information on social media. GIScientists harvest the locational-based social media (LBSM) for spatial analysis. Thus, location accuracy is of vital importance for effective emergency response and disaster management. Nevertheless, social media users can manage the setting of locational sharing to be determined by Global Positioning System (GPS) or mentioning of a place either from the user profile or in the meme. In the context of multimedia, whether it’s a picture or video, the area of interest (AOI) can be captured either on-scene (i.e. where the meme was post) or off-scene (i.e. where the AOI could be a certain distance away from the post location). This paper proposes a taxonomy of locational accuracy for social media, which can be either space-based or place-based, and the AOI can be on-scene or off-scene. Using this taxonomy as a guide, this paper investigates the locational accuracy of social media captured during Hurricane Harvey in 2017. To explore the practical use of this taxonomy, this paper limits the sample of LBSM to those contain useful information about water depth portraying the flooded landscape. Based on Google StreetView imagery and other auxiliary data, the researchers estimated and compared the planimetric location of AOI against post location to evaluate locational accuracy for these memes. The findings provide useful insights to strategies appropriate to understand, quantify and improve the locational accuracy of LBSM.