Floods have been increasing in frequency and severity due to climate change and urban expansion. While the physical causes have been widely studied, public interest, which influences community and government response, remains poorly understood. News coverage serves as a proxy for public interest and can reveal changes in public interest in the ongoing flood over time.

This study analyzes 1,342 district-level flood events in South Korea from 2012 to 2021. Using flood-related news data, we gauge public attention and compare it with actual damage. We develop a Feel-Like Flood Risk Index (FLFRI) to assess media attention relative to flood damage before, during, and after flood events. An AI model (XGBoost) with strong predictive performance identifies key drivers of media attention, including rainfall, flood duration, population, and economic loss.

Results show that major cities like Seoul and Busan have received more media attention despite less damage, while rural districts in central and southern regions with severe damage have received less attention.

This urban-rural disparity in coverage was most pronounced during flood events. Uneven attention may threaten community preparedness and resilience, highlighting the need for targeted risk communication in rural areas facing severe flood impacts but limited media coverage.