Every object in the universe above negative 459 degrees Fahrenheit emits a unique heat signature in the form of infrared light, or longwave radiation. Scientists use remote sensing to capture these measurements and identify objects, and now a Penn State-led research team is further developing these techniques to better identify infrared signals.
Nittany Radiance is a government-sponsored, Penn State-led data collection campaign that aims to better classify and characterize objects using hyperspectral longwave remote sensing. Hyperspectral imaging captures wavelengths and divides them into tens or hundreds of smaller bands, unlike conventional imaging that focuses on three bands — red, green and blue — or in the case of multispectral remote sensing, about a dozen. The researchers used Harris Corporation’s “Blue Heron,” a high-resolution instrument that divided the wavelength into 256 bands about one-fifth the thickness of a human hair.
“We’re looking at it with such a high resolution, each pixel will be able to tell us exactly what it is made of,” said Guido Cervone, associate professor of geoinformatics and associate director of the Institute for CyberScience at Penn State. “This is unique data, and it is very difficult in academia to obtain longwave hyperspectral data with such good resolution.”
Whereas traditional remote sensing analyzes one image at a fixed angle at one time, the Nittany Radiance researchers collected data from different angles. Using a combination of multiple scenes and machine learning algorithms, it should be possible to better classify objects, said Cervone.
The researchers set up panels coated in reflective paint and a wading pool at three locations around the University Park campus. They then took different thermal readings of the targets using field equipment and a drone, while an airplane with the Blue Heron instrument circled the campus at 10,000 feet. They also released acetone, ammonia and ethylene, which have distinct infrared signatures that the researchers could easily identify.
“Some gases and materials have very distinct features in the infrared. While they cannot easily be recognized in other parts of the spectrum, they can easily be spotted in the infrared,” according to Guido. “We’re able to say extremely precisely what kind of chemicals are present in the air, so potentially it’s very important for pollution and to identify types of gas leaks. It also works at night. It’s not dependent on the sun.”
Mark Salvador of Zi Inc. and co-primary investigator of the project, used a field spectrometer to collect ground data relative to the targets and the atmosphere. Gabriele Franch, a visiting doctoral student from the University of Trento in Italy, used a handheld reflectometer directly on the targets to get a precise emissions reading.
If Earth had no atmosphere, then what Franch measured on the ground would be the same as the measurements taken by the plane, according to Cervone, who also holds an appointment in the Earth and Environmental Systems Institute. However, the atmosphere emits and absorbs longwave radiation, causing the radiation measured on the airborne sensor to fluctuate.
The wading pool helped resolve this issue. The team used a pump to circulate the water and ensure an even temperature across the pool. Since water has a nearly flat infrared emission, dips and highs in the measurements taken by the plane helped the researchers identify atmospheric interference and better calibrate their computations.
In order to better characterize the atmosphere, Fangcao Xu, a doctoral candidate in the Department of Geography and member of Cervone’s lab, will correct the data then simulate it under different atmospheres and geometric angles, estimating thousands of files. Then the researchers will feed the data into a machine learning system and teach it to identify the different emissions from the targets and the atmosphere so they can perform precise target detections.
Over 20 undergraduate and graduate students and postdoctoral scholars in the departments of geography and geosciences participated in Nittany Radiance.
“I thought it was a great educational experience for the students,” said Carolynne Hultquist, a graduate student in geography who recently defended her doctoral dissertation. “Having a group of them working together and talking about spectral signatures and how functionally this works was very practical.”
The researchers said the project allows for better characterization of the atmosphere and, in turn, detection of gases and solids. Cervone said he hopes to conduct the experiment again next year to see how changes to the campus might change the readings, to better characterize the atmosphere, and to make difficult target detection possible.
Also participating from Penn State were Courtney Jackson, a graduate study in geography, Martina Calovi, a postdoctoral scholar in geography, and Jian Sun, a postdoctoral scholar in geosciences.
The Defense Advanced Research Projects Agency, Air Force Research Laboratory, the U.S. Army Corps of Engineers Engineer Research and Development Center and Harris Corporation supported this project.