Contact: Aga Haupt
Kudzu, the lush imported vine that can grow up to 60 feet a year, creates green sculptures as it covers trees, telephone poles and abandoned buildings in the South.
In timberland, it also creates economic losses by delaying harvests and damaging tree health.
To help lessen kudzu's impact, two Mississippi State University faculty members are combining their areas of expertise to tame the fast-growing weed. Lori M. Bruce, assistant professor of electrical and computer engineering, and John D. Byrd, research professor of plant and soil sciences, are applying the space-age technologies of remote sensing in a research effort they hope will produce a fully automated detection system to precisely locate kudzu.
"The advantages of applying remote sensing technology to the problem of kudzu include the ability to detect infestations in remote areas, as well as eliminating man-hours of ground surveying," Byrd said. "The images generated by remote sensing equipment also allow us to monitor changes in infestations over time."
Remote sensing captures data from aircraft- or space-based platforms. With funding from MSU's Remote Sensing Technologies Center, the team is preparing to analyze the first publicly available data from a 2000 NASA satellite launch. The data will be validated by ground observers.
"Our ultimate goal is to find a cost-effective way to establish the amounts of kudzu in the state and to pinpoint the areas where it grows," Bruce explained.
Native to Japan, the vine was introduced in the United States in 1876, and while it no longer remains on the list of noxious weeds targeted by the U.S. Department of Agriculture, "it is important to help eradicate the vine to spare landowners lost revenues and expenses related to spraying," Byrd said.
To do that, the research team is using a mathematical technique called the wavelet transform to help distinguish between classes of vegetation in remotely sensed imagery. An imaging technique based on the wavelet transform can tell the difference between kudzu and other vegetation.
"We are using remote sensing data to locate kudzu based both on spectral and spatial information," Bruce said, adding that the spectral data--an analysis of how light reflects off of objects on the ground--helps distinguish between vegetation and non vegetation, as well as between different classes of vegetation. The spatial data, including size, shape and texture, help pinpoint location.
"For example, in airborne data, kudzu looks smooth like a carpet, while trees have a bumpy texture," Bruce said.
NASA's data will be a starting point for the team to apply the space agency's technologies to a major vegetative headache, as well as to build on MSU's continuing remote sensing work in weed science and other agricultural applications.
"We hope that detecting kudzu patches that are one-quarter acre in diameter also will provide additional tools for helping farmers identify weeds that are less than four inches tall," Byrd said. "What we learn in this project can contribute to strategies for a regional kudzu eradication program in the Southeast."