Abstract
On August 21, 2024, severe rainfall caused backwater from manholes and flooding of
subway stations in Tokyo, many videos and images of the damage were posted on SNS. There
are few studies estimating inundation depths using image record for localized flooding in urban
areas with complex topography and drainage systems. In this study, we obtained estimated
inundation depths from SNS videos and images for the rainfall on August 21, 2024, and
compared them with the predicted results of S-uiPS, an urban inundation prediction method.
The images and video data collected from SNSs showed a good correspondence with the
calculated values. It was also found that flooding on roads targeted in this study occurs locally
in depressions due to the crossfall and curbs, and that the inundation depth can vary within the
same road section.
