Extracting disaster information based on Sina Weibo in China: A case study of the 2019 Typhoon Lekima

Abstract

As an emerging big data source, social media data has been attracted more and more attention in the field of disaster emergency management. This study took the case of the Super Typhoon Lekima, which landed in China in 2019, to explore how the public's disaster risk perception changed during disaster response based on microblog data from Sina Weibo, commonly regarded as “Chinese Twitter”. We first analyzed characteristics of microblogs and found that microblog activities were closely related to Super Typhoon Lekima's landing process, the public could sense typhoon landfall about 72 h in advance. Second, we found that there exists a significant linear correlation between microblog counts and precipitation on a daily and provincial scale. Finally, we constructed Chinese disaster-relevant keyword sets and used Naïve Bayesian classification to calculate population-adjusted disaster score of affected-people, collapsed-house and affected-agriculture, and we found that the extracted disaster information from microblogs was corelated to real loss at provincial scale, and Weibo provide further detailed spatial distribution characteristics of Lekima's destruction. This research demonstrates that disaster information extracted form Sina Weibo could reflect the public's disaster risk perception well and have potential to serve as a data source for disaster management in China.