Assessing Geo-Foundational Models for Flood Inundation Mapping: Benchmarking Models for Sentinel-1, Sentinel-2, and Planetscope

Geo-Foundational Models (GFMs) enable fast and reliable extraction of spatiotemporal information from satellite imagery, improving flood inundation mapping by leveraging location and time embeddings. Despite their potential, it remains unclear whether GFMs outperform traditional models like U-Net. A systematic comparison across sensors and data availability scenarios is still lacking, which is essential to guide end-users in model selection. To address this, we evaluate three GFMs—Prithvi 2.0, Clay V1.5, DOFA—and UViT (a Prithvi variant), against TransNorm, U-Net, DeepLabv3+, and Attention U-Net using PlanetScope, Sentinel-1, and Sentinel-2. We observe competitive performance among all GFMs, with 2–5% variation between the best and worst models across sensors. Clay outperforms others on PlanetScope (0.79 mIoU) and Sentinel-2 (0.72), while Prithvi leads on Sentinel-1 (0.57). In leave-one-region-out cross-validation across five regions, Clay shows slightly better performance across all sensors, with mIoU scores of 0.72, 0.66, and 0.51 for PanetScope, Sentinel-2, and Sentinel-1 respectively, compared to Prithvi (0.70, 0.64, 0.49) and DOFA (0.67, 0.64, 0.49).Across all 19 sites, cross-validation reveals a 4% improvement by Clay over U-Net. Visual inspection highlights Clay's superior ability to retain fine details. Few-shot experiments show Clay achieves 0.64 mIoU on PlanetScope with just five training images, outperforming Prithvi (0.24) and DOFA (0.35). In terms of computational time, Clay is a better choice due to its smaller model size (26M parameters), making it approximately 3× faster than Prithvi (650M) and 2× faster than DOFA (410M). Our results suggest GFMs offer small to moderate improvements in flood mapping accuracy at lower computational cost and labeling effort.