Wayanad landslides: Why remote sensing alone is not enough to predict floods?

Congress MP Shashi Tharoor has called for more advanced technology than remote sensing to predict landslides.

The horrific impact of the Wayanad landslides have highlighted the gaps in in technological infrastructure when it comes to predicting and mitigating such natural hazards. Congress MP Shashi Tharoor underlined the difficulties around predicting landslides, calling out the inadequacy of remote sensing technologies.  

Tharoor noted that “We rely heavily on remote sensing. But experts are saying this is not enough and that we need to have sensor grids on the ground to be able to anticipate landslides. We don’t have that. We don’t have an on-ground sensor grid. We need more real-time data gathering.”

Remote sensing uses deep learning technology among other methods to understand various aspects of a region including its topological, hydrological, vegetation and other geographic factors. Satellites built specifically for remote sensing observe and gather data of a region by way of photographs. These images capture different features of the area like water, soil, vegetation, rocks, etc in detail. These images are then stitched together to create a map of sorts. Then, deep learning is employed to build predictive models to determine which regions are vulnerable to a certain ecological disaster. However, much like other predictive models built by AI like AlphaFold which makes 3D models of protein structures, the outcomes aren’t assured. 

Akshay Kumar, a geologist at environmental consultancy ERM said, “Remote sensing isn’t predictive. The only thing that is known to work are ground radar surveys which requires engineers to continuously monitor slopes physically. It is a much more accurate technique given how tricky these environmental conditions are. The slopes keep shifting and in case of a sudden cloud burst, we must have real-time predictions to help.” 

(For top technology news of the day, subscribe to our tech newsletter Today’s Cache)

Remote sensing, on the other hand, Kumar says, is just a snapshot of a situation over a period that can be studied over time. 

However, Kumar notes that on-ground radar monitoring is an expensive undertaking and are usually done around mining facilities or while building roads. “Given that local authorities usually do not have the budget for such costly tech, certain slopes should just be considered off-limits. Unfortunately, that doesn’t happen because of human settlements and other interventions in vulnerable parts.”