DeepSARFlood

A fully automated SAR-based flood inundation mapping system, delivering high-accuracy results even under clouds, vegetation, and nighttime conditions.
About DeepSARFlood

DeepSARFlood is a fully automated flood inundation mapping system developed to support real-time flood monitoring and response using Synthetic Aperture Radar (SAR) satellite data.

The platform addresses the limitations of optical imagery during flood events, such as cloud cover and low-light conditions, by leveraging SAR’s all-weather, day-night imaging capability.

Using deep learning models trained on weakly supervised flood labels, the system produces accurate flood extent maps directly from SAR observations.

DeepSARFlood is optimized for large-area, rapid processing, enabling flood mapping over thousands of square kilometers within seconds while also providing uncertainty estimates. Designed for operational deployment, the system delivers decision-grade flood intelligence to support emergency response, disaster mitigation, and long-term flood risk planning.

Explore Our Delivered Projects

Amrit Sarovar Observatory

A centralized digital observatory enabling real-time tracking of India’s largest water rejuvenation mission covering 69,000+ ponds with AI-driven detection and monitoring.

Read More

Indian Landslide Susceptibility Map (ILSM)

India’s first machine learning–based landslide susceptibility map, delivering probabilistic risk estimates at 100 m resolution across the entire country.

Read More