Researchers from the National Remote Sensing Centre (NRSC) of the Indian Space Research Organisation (ISRO) have developed a novel method, called FlDepth, to measure floodwater depth from space in near-real time. This new geographic information system (GIS) tool uses satellite images of flooded areas combined with detailed 3D terrain maps to calculate exactly how deep the water is. Tested extensively during the severe 2024 monsoon floods in the Indian states of Assam and Bihar, the team aims to improve emergency response, evacuation planning, and disaster mitigation across vulnerable regions by rapidly translating flat, two-dimensional satellite images into three-dimensional depth maps.

FlDepth operates on a fundamental principle of hydraulics: the relationship between water surface elevation and ground elevation. Because measuring flood depth at a large scale during a torrential disaster is practically impossible for humans on the ground, the researchers designed a digital workaround. 

First, they use multi-sensor satellite imagery, which can pierce through thick storm clouds, to draw the exact boundary lines of an active flood. Then, using a digital elevation model, a high-resolution 3D map of the earth's surface, they determine the ground elevation right at those outer flood boundaries. Because the water at the edge of a flood meets the ground, this elevation represents the top level of the floodwater. 

The software automatically draws invisible cross-sections across the flooded area, interpolating the water surface elevation from one side of the flood boundary to the other. Finally, by subtracting the underlying ground elevation from this calculated water surface, the system reveals the actual depth of the floodwater at any given point across the landscape.

Earlier attempts to estimate flood depths from space, such as the widely used Floodwater Depth Estimation Tool, often produced inaccurate maps with abrupt, unrealistic drop-offs, linear stripes, and sharp transitions. On the other hand, traditional hydrodynamic models that can provide accurate depths require enormous amounts of historical rainfall data and river discharge measurements, making them far too slow to run during sudden, real-time emergencies. FlDepth improves on these older systems by treating the flood depth as a continuous, flowing surface. This allows the model to capture the natural, gradual depth transitions across gentle floodplains, resulting in much smoother, more realistic, and highly accurate maps without the need for intense computing power or extensive historical data.

Despite its performance in real-world measurements, the researchers note that because the system relies heavily on edge-based estimates, minor overestimations and underestimations can sometimes occur at river boundaries and in upstream sections. This occasionally creates wavy water-surface patterns in the data rather than flat, realistic profiles. Additionally, the overall accuracy of FlDepth is entirely dependent on the quality of the underlying digital terrain maps. If a river has recently changed its course, or if the terrain is highly complex, it can throw off the software's results. 

Nonetheless, by providing reliable information on how deep the waters are, FlDepth empowers authorities to plan rescues more efficiently. It also allows governments to rapidly estimate potential damage to critical infrastructure and agriculture, and plan safer evacuation routes for displaced communities. For millions of people living in data-scarce regions across the globe, this rapid alternative to complex modelling could be a life-saving tool during the rains.