Geometric Flood Depth Estimation: Fusing Transformer-Based Segmentation with Digital Elevation Models

2026-05-08

AI Summary

Existing two-dimensional flood segmentation methods struggle to provide water depth information essential for assessing passability and structural risk. This work proposes a novel approach that integrates the Mask2Former segmentation model with a digital elevation model (DEM) to estimate per-pixel flood depth from monocular aerial imagery. By leveraging the principle of local hydrostatic equilibrium, the method identifies water–land boundaries and infers a global water surface elevation, enabling geometric derivation of flood depth without requiring hydrodynamic simulations. It represents the first integration of a high-performance Transformer-based segmentation architecture with DEM data to efficiently reconstruct three-dimensional flood depth while balancing accuracy and computational efficiency. Experiments on the FloodNet and CRASAR-U-DROIDS datasets demonstrate the method’s effectiveness in extracting practically valuable 3D flood volume information from 2D imagery.

Abstract

Post-disaster situational awareness relies heavily on understanding both the extent and the volume of floodwaters. While 2D semantic segmentation provides accurate flood masking, it lacks the vertical dimension required to assess navigability and structural risk. This paper presents a geometric "Water Surface Elevation" approach for estimating flood depth from monocular aerial imagery. Our pipeline utilizes Mask2Former, a state-of-the-art transformer-based segmentation model, to generate precise 2D flood masks. These masks are fused with Digital Elevation Models (DEMs) to identify the water-land boundary, calculate a global water surface elevation ($Z_{water}$), and compute per-pixel depth based on the principle of local hydrostatic equilibrium. We evaluate this workflow using the FloodNet and CRASAR-U-DROIDS datasets, demonstrating how high-performance segmentation can be leveraged to extract 3D volumetric data from 2D imagery without the latency of hydrodynamic simulations.

Problem

• flood depth estimation
• situational awareness
• monocular aerial imagery
• water surface elevation
• digital elevation models

Innovation

• flood depth estimation
• transformer-based segmentation
• digital elevation model
• water surface elevation
• monocular aerial imagery