BROOKINGS, S.D. -Sept. 28, 2021 -Scientists working with Landsat satellites used an area east of 22nd Avenue in the Research Park at South Dakota State University to measure the light reflected and heat emitted from the Earth’s surface. Their research will help ensure the accuracy of a new line of U.S. Geological Survey Landsat science products.
“What we are doing here has a global impact,” said project manager Cody Anderson of the USGS Earth Resources and Observation Sciences Center’s (EROS) Cal/Val Center of Excellence (ECCOE), which sponsored the weeklong field campaign. The USGS-EROS ECCOE team, along with the NASA Goddard Space Flight Center Cal/Val team in Greenbelt, Maryland, is responsible for the data quality of the Landsat mission, including Landsat 9, which was launched Sept. 27.
From 2008 to 2019, scientists worldwide downloaded more than 100 million free Landsat images through the USGS user interface. This data provided an estimated $3.45 billion economic benefit to the scientific community in 2017 alone, according to a USGS report.
For the field campaign, USGS EROS had researchers from South Dakota State University’s Image Processing Laboratory, Rochester Institute of Technology’s Chester F. Carlson Center for Imaging Science and USGS service contractor KBR-Wyle compare new technologies with traditional methods of measuring surface reflectance and surface temperature.
Chris Durell, director of business development for Labsphere, came to Brookings through company funding “to be here with world-class calibration teams” and to demonstrate its mirror-based reflectance measurement technology.
In December, USGS rolled out its Level 2 products. “In the past, everything was Level 1, above the atmosphere and scientists had to do their own atmospheric corrections. The Level 2 Landsat products already have the correction component,” Anderson said. This not only saves researchers time and money, but also means the scientists are working with “data processed in the same way—everyone starts from the same point.
“In the past, we focused on finding the best possible site, the clearest atmosphere, to clear out the unknowns,” Anderson continued. “We wanted to remove everything from here to the satellite. Now, we want to get a broader area, different land surface types and atmospheric conditions in different regions throughout the United States. We’re reintroducing those unknowns and using multiple sites to test the algorithms (used for the Level 2 products) and how they work globally.”
Measuring surface reflectance
SDSU Image Processing Lab Director Larry Leigh said, “With Level 1, we were calibrating the sensors (on the satellites). With Level 2 surface products, the emphasis has shifted to coming up with technology to validate this higher-order product.”
During the field campaign, the SDSU researchers used an ASD Field Spec, which they have proven is the gold standard for measuring surface reflectance, and tested the Arable Mark 2 technology, which can be placed permanently at a location and transmit data to a central site.
“It’s not a high-quality instrument, like the ASD which requires a team to take readings at the site, but it’s inexpensive and we can stick out in a field and walk away. It has good potential, but we have to determine is whether it’s good enough,” Leigh said.
As another partner working on surface product validation, the RIT team deployed Headwall instrumentation on a pair of drones to do hyperspectral imaging, which, like the ASD, covers the visible and near infrared bands, and multispectral imaging, as does the Arable technology. Hyperspectral imaging divides the wavelengths into fine segments, while multispectral takes a broader approach with fewer bands.
“A lot of time has been spent putting the sensors together, buying the instruments and integrating the sensors so they talk to the drones,” said senior scientist Aaron Gerace, who leads the RIT team. The drone program, which began through more than $1 million in internal RIT funding, is in its second year.
SDSU Distinguished Professor Emeritus Dennis Helder, who worked on the field campaign through USGS contractor KBR-Wyle, said, “These are expensive technologies, but they have the potential to cover more territory and thereby produce a larger aerial image.”
In addition, Durell and RIT doctoral student David Coran worked with Planet, a commercial satellite company, to demonstrate the capabilities of Labsphere’s FLARE, which stands for Field Line-of-Sight Automated Radiance Exposure. The technology uses convex mirrors to redirect the sun’s rays toward the satellite and take radiometric (reflectance) and geometric measurements.
“We are showing the USGS, NASA and RIT teams that this is a really quick, inexpensive way to do the same things they’re doing,” he said, noting that in one day as many as 10 satellites passed over the field campaign site.
Helder said, “If they can show the mirrors work, this technology would make it possible to validate a 10-by-10-mile (or larger) area as long as the atmosphere is the same over that area. The potential is significant.”
Addressing thermal side
The RIT team works on the thermal side of the Landsat mission. “We’ve used an existing algorithm tailored for the Landsat thermal infrared sensor to measure surface temperature in two thermal bands,” Gerace said.
On the ground, the RIT researchers deployed a D&P FTIR spectral radiometer—the gold standard for measuring surface temperature—and a broadband radiometer called the Apogee, which is a commonly used instrument for measuring surface temperature. Furthermore, this is the first time RIT has tested a longwave infrared thermal camera, known as the FLIR, aboard drones.
“We are using this sensor data along with some image processing techniques to measure the surface temperature of water, vegetation, sand and asphalt,” Gerace said. “Our hope is this can be used to support Landsat calibration and characterization of our experimental sensors to validate Landsat’s surface temperature products.”
Another ground-based RIT technology based on thermopile detectors holds promise for improving surface temperature accuracy over land. Gerace and doctoral student Lucy Falcon developed the devices, which are the size of a nickel and have eight spectral channels optimized for Landsat.
“Ultimately, we can place them permanently in a field and take temperature measurements along with other data, such as wind speed, direction and rainfall, which can be transmitted to a website through which the free data can be downloaded,” Gerace said. By changing the filter configuration, the RIT researchers can also adapt the instrument for other companies’ satellites.
USGS’s Anderson concluded, “Everyone is going after automation, but before we can trust the automation we need to go out and do the legwork.”
