Spatiotemporal damaged area estimation with location information tweets

DOI Web Site

Bibliographic Information

Other Title
  • 位置情報付きtweetによる被害状況の逐次把握可能性の検討
  • イチ ジョウホウ ツキ tweet ニ ヨル ヒガイ ジョウキョウ ノ チクジ ハアク カノウセイ ノ ケントウ

Search this article

Abstract

This study aims to understand damaged area spatiotemporally by tweet data with location information as a case study of the Great East Japan Earthquake. Firstly, we make a list of tweet data with location information in Tohoku area from March 11 to March 18, 2011, from raw data of tweets in Japanese. In this process, bot accounts including seismic intensity information and "google person under" are excluded because they don't represent actual local conditions. Secondly, considering each tweet on the list as a expression of human activity of an area where tweet is made, we spatially extrapolate the tweets using Kernel density estimation approach and draw a "tweet-density distribution map." Then we try to understand the place and condition of damaged area is by this map, as well as comparison with other information such as DMSP night time map which is regarded as an indicator of ordinary human activities. As a result, we show the possibility of tweet data with location data to estimate heavily damaged area spatially and temporally. In addition, future works for further utilization of this map e.g. a review of useful comparison data for interpretation of proposed map are listed.

Journal

  • SEISAN KENKYU

    SEISAN KENKYU 65 (4), 529-532, 2013

    Institute of Industrial Science The University of Tokyo

Keywords

Details 詳細情報について

Report a problem

Back to top