Reconstructing Geographical Flood Probability and Analyzed Inundation Flood Mapping on Social Media Implementation

• Published in: Studies in Informatics and Control 2021-03, Vol.30 (1), p.29-38
• Main Authors: RAJESHKANNAN, Chinnaiyan, KOGILAVANI, Shanmuga Vadivel
• Format: Article
• Language: English
• Subjects: Computer science; Data science; Design; Engineering; Floods; Gauges; Geology; Machine learning; Satellites; Social networks; Wavelet transforms
• ISSN: 1220-1766; 1841-429X
• DOI: 10.24846/v30i1y202103

One of the major issues of the planetary ecosystem is natural disaster prediction, which utilizes the flood prediction approach to a larger extent. Inundation prevention considers the emergency response, post-event damage assessment and flood mitigation. The current reconstructed flood-mapping model is divided into two segments namely the near real span and real span procedure based on the timing information. The inherent conditions of each procedure broadly hamper their application for inundation mapping. This paper considers the South Carolina inundation in Columbia as a previous research work. The proposed research work presents flood reconstruction design by improving near real span normalized water index derived images from the collection of USGS datasets. Real span data is collected from river gauge readings and it is implemented in Social Media (Twitter API). The proposed flood mapping technique is divided into three sections namely the local enhancement section, enhanced water height section and global enhancement section. Here the section synthesis with gauge readings and various normalized water index images (Geoscience information) is presented with a view to reconstructing an advanced macro-scale flood probability condition by means of the Improved Markov Chain Monte Carlo method, which is improved by using the analyzed inundation-related social media. The result matches well with the U.S. Geological Survey flood map and its surveyed high-water marks for flood identification. The results suggest that by enhancing near real span imagery with real span data sources, the proposed flood inundation probability reconstruction model renders a more robust, gradually enhanced flood probability index for emergency responders so that they quickly identify areas in need of urgent attention. The methodology used in this paper could seed a wide range of future flood studies that may lead to a rapid and improved flood situational awareness in a city as well as at a regional level.