New Delhi: Sanjeev Nair will plead, until he goes a disturbing blue in the face, that he has no technological know-how. “I studied at the Delhi School of Economics, boss," Nair, a joint secretary in the department of science and technology, will say with a laugh. “I just need people in my team who understand the science." But in conversations with his scientists, it is inevitably Nair who first remembers the name of that dam on the Tapti river (Ukai Dam) or what a particular metric is for a spot on the river (103m). When this is pointed out to him, he laughs again: “Ah, that’s just stuff you pick up on the job."

The manager of disasters: Nair heads a team of scientists that plans to use satellite imagery for flood relief. Madhu Kapparath / Mint

Nair came to the department in 2004, from a stint in the Pradeshiya Industrial and Investment Corp. of Uttar Pradesh Ltd; two years later, he remembers, the notion of using satellite imaging for disaster analysis came under discussion. “It was just a concept at the time," he says. “How to really translate it into an application was the question."

When, in March 2007, a hailstorm hit the district of Rewari in Haryana, his scientists tested their first model, assessing in real time the texture of damage across the district. “We had no idea whether it would work or not," Nair says. “We went all across Rewari, at the grass roots, gathering data to see if it matched our model’s output." It did.

Last year, Nair’s team began to work with Mike 11, a commercial software package that simulates water flow and levels in rivers and reservoirs around the world.

The government’s earlier flood modelling system, says Pavan Kumar Singh, a senior research officer at the National Disaster Management Authority, had long been outdated. “A model has to take into account the new urbanization along the river, the desilting, the existing drainage systems," he says. Mike 11, by incorporating these factors, “can definitely be adapted to work in India, although I don’t know how far it can help in predictions".

In their pilot project, Nair’s scientists worked on the Sabarmati and Tapti rivers in Gujarat—a soft beginning, Nair admits, since both rivers are far from turbulent. But it was still the start of an effort to seek a practical solution.

“The JS (joint secretary) told me: ‘Get out of your labs,’" recalls Devendra Singh, one of the disaster management cell’s scientists. “‘You need to go into the field.’"

Mike 11’s inbuilt model is a generic one. “We have to calibrate new models based on each river, testing its sensitivity for many variables," Singh says. Knowing how the rivers flooded in the past, Singh and his colleagues patiently figured out how sedimentation, or topography, or releases of dammed water, or snowmelt, or soil quality affected flood water levels. The result was a model that had an error of 10-20%, which Singh claims can be easily refined further. But the team still hungers for better field data. “We’ve had cases where one automated weather station has reported 1mm rainfall, and another station in the same area has reported 3mm," he says wryly.

On a monitor hooked up to a Sun Microsystems mainframe, Singh pulls up an example of the processed satellite imagery that emerges as an output from the model. A sequence of images shows the flooding of the Brahmaputra river last July, created on request for the government of Assam. Large, vivid patches of light blue show where the Brahmaputra has exceeded its banks.

Tracking the flood thus in real time, Singh says, “lets us know where waterlogging has happened. So a state can rapidly deploy forces and resources as needed". The flooding in Assam begins on an image labelled 14 July; only by the time the sequence reaches 28 July does the image revert to its pale green, through which the Brahmaputra flows again as a regular ribbon of deep blue.

In 2010, the flood models will tackle more brawny rivers: the Kosi, Gandak and Baghmati in Bihar, and the Mahanadi, Brahmani and Baitarani in Orissa. Nair also hopes to push the project to the next level. Once the six specifically calibrated models for these rivers have been developed, his team will feed three-day rain forecasts, from Kolkata’s Bose Institute and the Indian Institute of Technology, Kharagpur, and learn, hours in advance, which districts will get flooded.

“We want to try this during the monsoon," Nair says. “That will, obviously, be the real acid test for us."