Many people talk to their plants. At MITRE, some researchers believe that the plants are talking back—and telling us a lot.

Botanists and other scientists have long known that visibly damaged plant life indicates long-term negative effects on the environment, such as pollution. Yellowed leaves, wilted stems, and bare branches are hard to miss. But what about the less-obvious effects that occur during the early stages of damage? And what would this indicate? What if plants could provide hints to illegal or destructive activities even before the effects are visible to the eye?

MITRE researchers are employing a technique called "vegetation forensics" to explore the possibilities of detecting illicit (or negligent) human actions by using cutting-edge sensors and monitors to ferret out plant damage before it becomes readily apparent. The idea is to discover whether certain types of illegal actions—such as chemical dumping—cause early, detectable impact on plants in the surrounding environment. The work also might point the way to identifying natural scenarios, such as the effects of mineral deposits in soil.

Seeing the Forest for the Trees

Sherry Olson, principal investigator on MITRE's vegetation forensics project, got the idea that this new area of research could benefit MITRE's sponsors. She was inspired after hearing a talk at MITRE by Professor Barry Rock of the University of New Hampshire. Rock talked specifically about the devastation that acid rain from sulfur emissions of coal-burning factories had caused to the environment in Czechoslovakia and Poland. He raised the question: While we know that visible stress affects plants, can we use hyperspectral data to detect early (pre-visible) stress?

Illegal activities, especially deep in forests, can be hard to see directly. And sometimes these activities are transient or temporary. But if they leave behind vegetation stress indicators, Olson hypothesized that it might be possible to detect those activities sooner rather than later.

These visible signs of plant damage enable the researchers to witness what's called "ground truth"—direct evidence. The problem with ground truth is that sometimes it's too difficult or dangerous to collect evidence of criminal or negligent activity on site, particularly in inaccessible or unstable locations. This is where hyperspectral data collection using remote sensing becomes critical.

Following Rock's talk, Olson—who has an extensive background in remote sensing techniques—began to pursue research for the MITRE Technology Program to make the link between hyperspectral data and the secrets hidden inside plant life.

Fingerprints of Light

"Hyperspectral data is a series of measurements taken at hundreds of distinct wavelengths—many of them beyond the visible light spectrum—simultaneously by specialized sensors," Olson explains. "Taken together, using mathematical algorithms to help us analyze the data, the measurements give us a kind of 'fingerprint' of an item. Layers on layers of images give the information across the wavelengths, and the measurements across the wavelengths generate spectra to help us figure out what we're looking at."

Data can be gathered by supersensitive handheld systems, airborne sensors (used when you can't get into the area you're investigating), or possibly even spaceborne sensors. The researchers can then take measurements in the field or in a lab, depending on the instrumentation tools and the type of data they're looking for. Because of the volume of data and computer-intensive nature of hyperspectral imaging, much of the actual analysis must be done back in the laboratory with high-powered computers. Images are compared to libraries of known hyperspectral signatures to help determine what the sensors have detected.

By working in a controlled setting, the MITRE team is using these sensors to measure visible and non-visible changes in the plants' chemical and biological makeup and compare them to the healthy control plants. They can then begin to build databases of information that will provide valid clues about what's happening in the real world. The project brings together MITRE personnel and outside consultants in the areas of plant physiology, botany, and vegetation chemistry.

At the Root of the Truth

To test their theories, MITRE's vegetation forensics team set up shop on a piece of land in southeast Virginia. The initial tests involved exposing several sets of plants to airborne toxins within fume hoods and measuring the changes in plant chemistry. That test proved successful enough to lead to a more recent experiment, involving a much larger sample and a different exposure method. For this latest investigation, the team simulated illegal chemical dumping by applying gasoline and isopropyl alcohol to the soil surrounding several dozen plants. The team chose philodendron and schefflera, typical of the kind of thick foliage that makes inspection of jungles and rainforests so difficult.

The results were striking. "The roots of the plants grew distinctly differently from those of the control plants—the roots were dying and trying to grow up and away from the chemicals," says Olson. Using hyperspectral sensors known as reflectance spectrometers, the team was then able to measure specific changes in the plants, in this case light-reflection characteristics.

Michael West, a MITRE senior software applications development engineer who works closely with Olson, says that how plants reflect light reveals their health, providing an excellent baseline for experimentation. "The green-leaf vegetation spectrum is well-understood," he says, making it easier to measure changes that seem out of the ordinary.

"The reflectance function indicates the material properties—the actual [molecular] structure—of things," West says. "For instance, we can tell the difference between vegetation and soil, concrete, and so on, by the way something reflects light, including the invisible wavelengths."

Analyzing the spectral signature makes it possible to detect even minute changes from the norm. A noticeable difference in a plant's spectral signature often indicates unhealthy conditions. For instance, in the group of gasoline-exposed plants, the sensors showed measurable differences in certain cell characteristics between the stressed plants and the healthy control group.

"By measuring the spectral characteristics of these plants, we can help identify what's going on: drought or chemical dumping? Natural or manmade? We want to build large [databases of] spectral libraries and imagery that will be useful to a variety of government organizations" West says. "But first, we want to prove that this works."

Branching Out

Since analyzing environmental clues in this way is a relatively new area of research, it's important that the team follow proper experimental protocols. Before they even began their plant-contamination experiments, Olson and her colleagues participated in airborne collections of hyperspectral data over the National Arboretum in Washington, D.C., and Fort A. P. Hill in Virginia. As part of the ground-based team, the MITRE researchers gained valuable experience in gathering and analyzing this type of data and comparing it to known characteristics of plants, much as they would if they were the actual personnel using these methods in the field.

To further enhance its vegetation forensics work, MITRE's Exploitation Systems Lab recently acquired a special instrument called a spectrofluorometer. Plants emit (as opposed to just reflect) light that is not visible to the naked eye. This emittance, called "chlorophyll fluorescence," can give clues to a plant's health; too much fluorescence compared to the norm could indicate environmental changes.

Olson believes this is just the beginning of some promising work. "We definitely know we can see stress—now we want to learn how to classify it: What's happening? Is it human-caused? Is it naturally occurring? Even better: Can we fine tune our system to figure out exactly what's happening?"

One day, MITRE's vegetation forensics work may help provide early warning of a pipeline leak or indications of human activity in a remote location. The success of the initial experiments has given researchers a peek at what lies beneath—a promising new method for analyzing environmental data.

"Our eyes only see a little of the spectrum, but there's so much more there," Olson says.

—by Alison Stern-Dunyak

Related Information

Articles and News

• Analyzing Hyperspectral Data

Solutions That Make a Difference.®
Copyright © 1997-2013, The MITRE Corporation. All rights reserved.
MITRE is a registered trademark of The MITRE Corporation.
Material on this site may be copied and distributed with permission only.

Privacy Policy | Contact Us