A comprehensive dataset of riverine levee overtopping events for advancing risk assessment.

Earthen levees are the primary flood protection infrastructure worldwide, with overtopping recognized as the most prevalent mode of their failure. Despite its significance, detailed field data on overtopping events are often limited and fragmented. Here we present the Levee Loading and Incident Dataset - Overtopping (LLID-OT) version 1.0, a curated dataset of 487 overtopping events in U.S. riverine levees compiled over the past 15 years. Data were collected from diverse sources, including U.S. Army Corps of Engineers reports, aerial imagery, hydrologic data repositories, and eyewitness accounts, and include detailed information on hydraulic loading, soil and foundation classifications, construction/maintenance characteristics, geometry, and breach dimensions. Quality control included cross-referencing sources and verifying key parameters to ensure dataset consistency and reliability. Analysis reveals consistent relationships between breach characteristics and variables such as soil type, overtopping depth and duration, and levee geometry. The LLID-OT supports the development and calibration of data-driven and physics-based models, advancing levee risk assessment, design, and flood management.