Evaluating Cross-Country Generalization of Deep Learning-Based Flood Mapping Using Sentinel-1 SAR and Sentinel-2 Imagery
Abstract
Benchmark performance in flood mapping does not reliably predict how a model will behave in a new country. This paper tests that directly. Using the Sen1Floods11 dataset, we evaluate twelve model configurations—two architectures (Vanilla U-Net and ResNet34-encoder U-Net) across three sensing modalities (Sentinel-1 SAR, Sentinel-2 augmented with NDWI and MNDWI spectral indices, and multimodal Fusion)—under a standard random train-test split and a geographically disjoint country-level split where test countries are held out entirely from training. The gap between these two conditions is large and consistent. The worst case is Vanilla U-Net with Sentinel-2: IoU drops from 0.727 under random splitting to 0.464 under country splitting, a 36.2% relative loss. SAR degrades least. ResNet34+Fusion is the strongest country-level configuration, reaching IoU 0.555 and average precision 0.850. NDWI and MNDWI provide the clearest class separation of all input features (mean flood-to-background difference of 2.21 and 2.27 standard deviations respectively), and their inclusion improves country level IoU consistently across architectures. Somalia is harder than Paraguay in every configuration, with SAR-only models showing a 2.4× country ratio that narrows to 1.3× for ResNet34+Fusion. The key finding: the evaluation protocol matters as much as the model design.
Related
- Threshold Sensitivity and Reference Uncertainty in Sentinel-1 Inundation Mapping: The July 2021 Mahad Flood
- Mapping Land Use Typologies During Fallow Periods in France Using Sentinel-1 and -2 Time-Series
- Earthquake damage mapping from Sentinel-1 modeling against coherence damage proxy
- Super-resolved canopy height mapping from Sentinel-2 time series using airborne LiDAR HD reference data across metropolitan France
- Evaluating Cross-region Generalization for Wavelet-Diffusion Precipitation Downscaling
BibTeX
@misc{dhagat2026evaluating,
title = {{Evaluating Cross-Country Generalization of Deep Learning-Based Flood Mapping Using Sentinel-1 SAR and Sentinel-2 Imagery}},
author = {Nitya D Dhagat},
howpublished = {EarthArXiv},
year = {2026},
month = sep,
doi = {10.31223/x59z2m},
url = {https://eartharxiv.org/repository/view/15258/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Evaluating Cross-Country Generalization of Deep Learning-Based Flood Mapping Using Sentinel-1 SAR and Sentinel-2 Imagery
AU - Nitya D Dhagat
PY - 2026
DA - 2026/09/28
PB - EarthArXiv
DO - 10.31223/x59z2m
UR - https://eartharxiv.org/repository/view/15258/
AB - Benchmark performance in flood mapping does not reliably predict how a model will behave in a new country. This paper tests that directly. Using the Sen1Floods11 dataset, we evaluate twelve model configurations—two architectures (Vanilla U-Net and ResNet34-encoder U-Net) across three sensing modalities (Sentinel-1 SAR, Sentinel-2 augmented with NDWI and MNDWI spectral indices, and multimodal Fusion)—under a standard random train-test split and a geographically disjoint country-level split where test countries are held out entirely from training. The gap between these two conditions is large and consistent. The worst case is Vanilla U-Net with Sentinel-2: IoU drops from 0.727 under random splitting to 0.464 under country splitting, a 36.2% relative loss. SAR degrades least. ResNet34+Fusion is the strongest country-level configuration, reaching IoU 0.555 and average precision 0.850. NDWI and MNDWI provide the clearest class separation of all input features (mean flood-to-background difference of 2.21 and 2.27 standard deviations respectively), and their inclusion improves country level IoU consistently across architectures. Somalia is harder than Paraguay in every configuration, with SAR-only models showing a 2.4× country ratio that narrows to 1.3× for ResNet34+Fusion. The key finding: the evaluation protocol matters as much as the model design.
ER -