downtoearth-subscribe

Ground truths

Ground truths VILLAGLRS reading satellite images like the back of their hands: it's an absurd oxymoron. But this virtual impossibility is precisely what the Bangalore-based Actionaid team discovered in Nepal recently.

The team had started with an unlikely hypothesis: that the local people, with their intimate knowledge of the vicinity's topography, would probably read the images better.

In Nepal, satellite imagery is important because much of the country's forest area is inaccessible. Satellite imagery can help as an information -gathering tool for identifying resources and using local knowledge to manage them. Says Madhab, a team member, "The extent of the havoc caused by deforestation becomes visible on the image. Since village people can comprehend degradation better, there is a possibility that they will assume the role of guards."

In remote sensing parlance, the process of identifying ground spots from satellites is called "ground truthing". In all, Actionaid met about a dozen groups of people over a 3-day hill trek, carrying with them satellite images of the area on a 1:50,000 scale. Most villagers they met "read the imagery better than us," says Sanjay Rana, a team member.

The initial attempt bounced. On the banks of the partially dry Sindhuli Khare river, Tarang, an old man was asked to identify on the image the spot they were standing on. He failed. Disappointed, the team, however, decided to push on, realising that Tarang, 70, was too old and had bad eyesight.

Finally, at the conjunction of the Sindhuli and the Marine rivers, they met 3 men who were told that the images had been taken from 900 kin above the earth. They were told also that on the satellite image vegetation looks red, water appears either blue or black, and sand and gravel, white.

Studying the images for about 20 minutes, the men accurately identified not only Sindhuli and Marine, but also the parallel steams of Dandi Khare, Jalewa, Hathibandha and Ghagar.

Now motivated, the team set out to see if others could repeat this feat. Sure enough, another group in a different village identified some more rivers. And on images dating back to March 5, 1992, they located several landslides dating back to the pre- 1992 floods. They showed up as white patches.

But could the villagers identify really minute details? Across the wide Ghagar river, Chandra, Bojha Majhi and their friends helped the team identify 2 pineapple orchards on either side of a stream, which appeared as red specks on the image. They even identified a fish pond.

The currently available resolution on Indian Remote Sensing (IRS) data is about 36 m. Resolution is the minimum distance required for 2 objects to be distinctly identified on an optical instrument. Actionaid feels that with greater data resolution from the forthcoming'IRS 1c, images can be better used for micro-level planning, implementation and monitoring. Aided, of course, by something that has been around far longer than satellites: village wisdom.

---M N Srinivas is a member of the Actionaid team.

Related Content