MalaysiaGIS.com

Source: New Straits Times

For smugglers, illegal immigrants and loggers with theft on their minds, the forests’ thick green canopy is no longer the perfect camouflage for their dirty deeds.

Thanks to a little green box with a lens, that is one fast fading dream.

This sensor that looks like a harmless biscuit tin, has chartered secret smuggler trails, discovered illegal settlements and captured loggers felling trees outside their legal limits.

The sensor is hooked up to a powerful computer and global positioning system device and together form the airborne hyperspectral imaging kit.

From a thousand feet above ground, mounted on a modified fixed-wing aircraft, the kit has mapped smuggling trails less than a metre wide in some jungles surveyed over the past few years.

Better known as lorong tikus (rat lanes) for its narrow, winding nature, the lanes would be easy to miss if someone searched on foot, explains Professor Dr Kamaruzaman Jusoff.

If flown lower than the usual 1,000 metres above ground level, the sensor would be able to map trails — legal or illegal — as narrow as half a metre wide.

The forestry-engineering professor from Universiti Putra Malaysia, who put together the system, is reluctant to divulge more saying only that the technology will be useful in surveillance of our forested border areas.

Flying on a routine research survey over a state forest reserve some years ago, the sensor was able to pick up logging outside licensed concession area.

These were in hard-to-reach mountainous areas that enforcement officers couldn’t get to.

More recently an illegal settlement in the heart of the Air Itam Forest Reserve, near the university, turned up bright and clear on the sensor’s reading.

"No one even guessed an entire colony of illegal immigrants was living in the forest, but there it was, on the images on the screen," said Kamaruzaman, whose report led authorities to the precise site.

The system was born from a challenge thrown to the scientist who had been researching satellite and radar systems for some years.

When he told Forestry Department officers he could create a sensor that could count individual trees and tell them apart by their species, they flatly refused to believe him. A sensor in the clouds taking an inventory of a dense, multi-layered forest sounded like complete nonsense.

Listing tree species is painstaking work, done on foot, by experts with years of learning, they told the lecturer who has also served as deputy director at the Malaysian Centre for Remote Sensing.

But the soft-spoken Kelantanese politely disagreed and went on to create the kit specifically for timber inventory and species mapping.

In three years, with RM3 million in government grants, he proved them wrong.

In 2004, UPM set up a joint-venture company — UPM-Aeroscan Precision Sdn Bhd — to commercialise the system and the sensor had its first flight in February, 2004.

Fussing over the equipment before an aerial survey of Kuala Lumpur’s few remaining green lungs, Kamaruzaman explains that the sensor is like a colour photocopier in the sky.

Only this flying Xerox machine sees things in much greater detail.

If we were flying over an oil palm estate for instance, it can show where young, matured and stressed trees are, says Kamaruzaman over the drone of the Cessna’s engines.

Up in the air, the 15kg of equipment, much of it strapped to the floor of the Cessna for the bumpy flight, are busy scanning and processing what it sees. The aircraft is flown in a series of straight parallel lines over the area surveyed because the sensor’s sees a swath of land about 30 metres wide.

It takes hours to survey a small area and the flight isn’t for the weak stomach.

When the Kelantan Forestry Department wanted to know what was in the Gunung Stong Forest Reserve which it planned to declare a state park, it turned to Kamaruzaman.

With the technology, they had the answer to that question within three months, and the area has since been made a state park.

"If we’d done it the old way, going from tree to tree in the forest, we’d still be in the jungle now," says the department’s former director Datuk Dahlan Taha.

The forest reserve is over 21,000 hectares and it is a tough place to walk about in, says Dahlan, now the deputy director-general for planning and development at Peninsular Malaysia’s forestry headquarters.

It would have taken the department almost two years to complete a survey on foot, he says.

The department now plans to expand the use of this sensor to other areas in the country.

The system had also been helpful for something Kamaruzaman had not anticipated — search and rescue efforts, but it has contributed in two cases so far.

The technology helped narrow the area of search in mid 2004 when a 206 Bell Long Ranger helicopter carrying Sarawak Assistant Minister Dr Judson Tagal and six others crashed in the dense jungles between Bario and Ba’Kelalan.

Earlier this year it helped again by showing the embedded wreckage of the Hawk 208 fighter jet that crashed in Rompin’s coastal waters.

"It was stressful work and we slept at 2am, processing all the images so we could advise search and rescue teams."

He was not happy about being unable to instantly tell people on the ground what the sensor saw in those two cases. He is working out the kinks in the system.

Kamaruzaman is working on other limitations too, like the fact that the sensor depends on light, limiting to use in the day when the sun shines strongest.

To work at night, he’ll need an infrared sensor but one that works despite our high humidity.

The sensor also cannot detect temperature differences and this is the other aspect the intrepid professor is working on.

"But there are lots more it can do. The system can easily pick up a factory that is illegally discharging toxic waste into a waterway.

"It could locate illegal toxic chemical dumps in estates and forests."

How it works
DARK green, light green and emerald green is about as specific as we get when it comes to telling shades of colour apart.

But that’s not the case with airborne hyperspectral imaging technology.

Every object reflects, emits or absorbs a different intensity of light.

The leaves of a durian tree would have a different reflectance from those of a teak tree.

An open space, a river and different parts of an aircraft, would all produce their own unique reflectance.

And the sensor in the hyperspectral system can be taught to recognise these subtle differences.

To survey a forest, Dr Kamaruzaman Jusoff begins work by recording the reflectance each species of tree gives off. This is called a spectral signature.

All signatures are compiled into a library and the sensor draws from it, to identify different species.

This is where it differs from a satellite that will detect forests simply as a big patch of green — it can see visible light and near infra-red light.

His first library of trees was compiled when he surveyed Kalabakan highlands. It took his team a month to compile spectral signatures of about 1,000 species, and his library has grown to 6,000 species now.

So when the sensor is flown over a forested area, it captures a reflectance from all the trees and produces an image that looks like a map with hundreds of different coloured dots.

That image goes through several rounds of processing, where it is enhanced and sharpened.

Then the map is checked against what is on the ground to provide the client with an accuracy reading. A report is then prepared and submitted.

If a bunch of coloured dots on the image doesn’t match anything in the library, the team checks to see what it is on the ground. It could turn out to be a new species.

Most of the components of the system come from Australia, Finland and Canada.

It’s creation was inspired by work Kamaruzaman had done at Nasa’s Jet Propulsion Lab and with Australia’s CSIRO (Commonwealth Scientific and Industrial Research Organisation) airborne scientists about a decade ago.

The system was first used in a flight over Bukit Lanjan, looking for cracks in boulders along the highway, then in a survey of the Air Itam forest reserve and when mapping sedimentation in Sungai Langat and Sungai Selangor in 1994.