Near Real-Time Single-Image SAR Flood Mapping With Synthetic Pre-Event Reference: On-Demand COSMO-SkyMed Case Study

Near-real-time (NRT) flood mapping from on-demand synthetic aperture radar (SAR) data is often hindered by the lack of a suitable pre-event image in the archive. Under emergency conditions, searching for a recent, radiometrically comparable pre-event acquisition can be unsuccessful, limiting change-...

Full description

Bibliographic Details
Published in:IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Main Authors: Luca Pulvirenti, Giuseppe Squicciarino, Silvia Puca
Format: Article
Language:English
Published: IEEE 2026-01-01
Subjects:
COSMO-SkyMed
flood mapping
near real-time (NRT) monitoring
on-demand SAR data
soil moisture
Online Access:https://ieeexplore.ieee.org/document/11527133/
Description
Summary:Near-real-time (NRT) flood mapping from on-demand synthetic aperture radar (SAR) data is often hindered by the lack of a suitable pre-event image in the archive. Under emergency conditions, searching for a recent, radiometrically comparable pre-event acquisition can be unsuccessful, limiting change-detection approaches and reducing flood mapping accuracy. This article presents a new NRT framework to generate a physically consistent synthetic pre-event baseline, enabling effective bitemporal flood mapping with a single postevent SAR image. First, Copernicus global land service surface soil moisture (SSM, 1 km) is downscaled to sentinel-1 geometry (S1, ∼20 m) using the same S1 data from which SSM is derived. The downscaled SSM is then converted to volumetric soil moisture using a porosity map derived from bulk density data. Second, a pre-event SAR image is simulated via a semi-empirical backscattering model. The simulation accounts for surface roughness uncertainty, which is propagated through three scenarios (smooth, medium, rough). Finally, the synthetic baseline is regridded to the postevent SAR acquisition and integrated into the AUTOWADE2.0 processor, a fully automated flood mapping algorithm described in prior work, to generate flood extent maps through change detection. Experiments on recent Italian flood events demonstrate that the proposed workflow produces pre-event backscatter data that enable fast and reliable SAR-based flood mapping when real pre-event data are unavailable.
ISSN:1939-1404
2151-1535

Similar Items