Creating a national urban flood dataset for China from news texts (2000–2022) at the county level

Fu, Shengnan; Schultz, David M.; Lyu, Heng; Zheng, Zhonghua; Zhang, Chi

doi:https://doi.org/10.5194/hess-29-767-2025

Articles | Volume 29, issue 3

https://doi.org/10.5194/hess-29-767-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/hess-29-767-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 29, issue 3

Research article

| Highlight paper

|

13 Feb 2025

Research article | Highlight paper |

| 13 Feb 2025

Creating a national urban flood dataset for China from news texts (2000–2022) at the county level

Shengnan Fu, David M. Schultz, Heng Lyu, Zhonghua Zheng, and Chi Zhang

Data sets

Urban Flood Inventory in China Shengnan Fu https://doi.org/10.5281/zenodo.14000094

LandScan Global 2022 K. Sims et al. https://doi.org/10.48690/1529167

Video abstract

Creating a National Urban Flood Dataset of China from News Texts at the County Level Shengnan Fu https://doi.org/10.5446/69669

Executive editor

This paper uses information from news sites with natural language processing tools to infer data on a hydrological process at the regional scale (flooding). The paper demonstrates the technique's applicability and opens new avenues to use advanced computing techniques and web resources to improve the understanding of hydrological processes.

Short summary

We create China’s first open county-level urban flood dataset (2000–2022) using news media data with the help of deep learning. The dataset reflects both natural and societal influences and includes 7595 urban flood events across 2051 counties, covering 46 % of China’s land area. It reveals the predominance of summer floods, an upward trend since 2000, and a decline from southeast to northwest. Notably, some highly developed regions show a decrease, likely due to improved flood management.