FloodSimBench: A High-Resolution Physical Flood Inundation Benchmark

FloodSimBench is a standardized benchmark dataset designed for urban flood inundation modeling. It bridges the gap between complex physical hydraulic modeling and deep learning foundation models. The dataset was created by author chrimerss and was last updated on 2026-05-10.

Use Cases

• Train deep learning foundation models for flood prediction based on high-resolution physical modeling data.
• Evaluate AI surrogate models for urban flood risk assessment based on the benchmark's standardized structure.
• Compare the performance of Transformers and CNNs on physical simulation tasks based on the dataset's design.

Strengths

• High-resolution 1-meter spatial detail for urban environments.
• Standardized structure designed specifically for benchmarking AI models.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

huggingface

Freshness

Last updated 2026-05-10 16:08:30.