Agri Leader

UF pioneer in eco-imaging

Photos

JOHN BUCHANAN
Published: December 5, 2012

The University of Florida Institute of Food and Agricultural Sciences has helped develop an innovative scientific initiative that will allow for more effective water management.

Matteo Convertino, a research scientist in the agricultural and biological engineering department, led a team that created a new mathematical model for extracting information from satellite images of ecosystems.

The importance of the groundbreaking work lies in the fact that making field observations of organisms and ecosystems is often very difficult. Existing methods of observation are limited and travel to remote locations is expensive.

As a result, digital satellite imagery has long been considered a good alternative for assessing the environmental well-being of wild places such as the Florida Everglades.

Until now, however, the use of satellite imagery in ecological studies has been fairly limited. One reason, Convertino said, is "a strong need to improve mathematical formulas for extracting and analyzing data."

Convertino and his colleagues developed a new mathematical model capable of extracting data from even low-quality images.

"The reason that is important," Convertino said, "is that it actually goes in the opposite direction of most work done by other people, who have been working on better sensors that will take more high-quality images. And of course, our model also works with higher-quality imagery, as well. And as the quality of the imagery continues to improve, our model will perform better and better."

In their project, published recently in the peer-reviewed scientific journal PLOS ONE, Convertino and his team, using satellite photographs, achieved 98-percent accuracy in their analysis of the number of different plant species in Everglades wilderness.

The new and improved method for extracting information from images delivers three key benefits, Convertino said — improved accuracy, higher speed and reduced costs. It also allows for more use of existing imagery that dates back for decades, but was not usable for highly accurate analysis until now.

"The benefit of being able to use older existing imagery is that it covers long periods of time and large tracts of land," Convertino said.

To help make that vast library of imagery available to researchers and analysts, Convertino and his colleagues used a 71-year-old probability formula known as the Kullback-Leibler divergence.

For farmers and ranchers, the relevance of their work is that scientists can do a better job of extracting information from imagery that has to do with large water resources such as the Everglades. For example, studies can now do a more effective job of analyzing diversity within an ecosystem. And that helps increase capabilities for responsible stewardship of water resources, an issue critically important to Florida.

The next step in Convertino's ongoing work will be development of an additional component of the model that will allow analysts to identify the precise locations of individual species. "Right now, we can only detect the number of species in a particular area," Convertino said. "We are not yet able to see exactly where they are. So the next step in the process will allow for segmentation by species. And that will be useful in improving analysis of ecosystems and doing a better job of protecting them."

Collaborating with Convertino were Igor Linkov of Carnegie Mellon University and the U.S. Army Corps of Engineers, and Nathan Lowry, Mukund Desai and Rami Mangoubi of Charles Draper Laboratory in Cambridge, Mass.

Mangoubi praised Convertino for his role. "Working with Professor Convertino presents a unique opportunity for addressing important environmental and agricultural issues," he said. "He has a unique flair for selecting research problems that are both relevant and analytically challenging."

Ruth Borger, assistant vice president at IFAS Information and Communication Services, also gave Convertino high marks for his work. "He is a great example of a UF/IFAS researcher who's looking for new ways to address big challenges — in this case, using satellite imagery to provide data on remote ecosystems that would be difficult, and maybe even dangerous, for people to go and visit," Borger said. "This paper is part of a larger, longer-term collaboration and I'm excited to see what other insights his team gains."