University of Nebraska-Lincoln graduate student Thomas Polivka has publicly identified a flaw in new satellite remote sensing technology that monitors temperatures on Earth.
The flaw Polivka found has to do with how data from the sensor detectors are combined into pixels before being transmitted from the satellite to the Earth. Readings taken over hot spots, such as wildfires or gas flares, may be inaccurate because the combined data may mask whether one or more of the sensor detectors was maxed out with more radiance than it could measure, making it saturated.
Polivka was the lead author of a peer-reviewed article published this month in the Institute of Electrical and Electronics Engineers Geoscience and Remote Sensing Letters. The article, featured on the issue’s cover, included co-authors Edward J. Hyer, a scientist with the Naval Research Laboratory’s Marine Meteorology Division in Monterey, Calif.; David A. Peterson of the National Research Council, also in Monterey; and Jun Wang, an associate professor of earth and atmospheric sciences at UNL.
“This work is interesting and important,” Wang said. “It is very, very rare for a student who has not yet defended a master’s degree to have a journal-cover article like this published. It’s the first time I’ve seen it in my career.”
The flaw had been a topic of conjecture by other scientists, Wang said. Polivka began studying the issue during a 10-week summer internship at the Naval Research Laboratory in 2013. Polivka’s article is the first global analysis of saturation artifacts in the detection system’s infrared channels and the effects of sample aggregation. The work is supported in part by NASA.
The space-borne sensor, called the Visible Infrared Imaging Radiometer Suite, or VIIRS, is a feature of the next generation of satellites being launched to monitor weather patterns on Earth, Wang said. It is an update and replacement of aging weather satellite sensors that have been in use for decades.
NASA and the National Oceanographic and Atmospheric Administration first launched VIIRS in October 2011 aboard the Suomi National Polar-orbiting Partnership satellite. In an orbit about 520 miles above the Earth, the satellite circles the planet from pole to pole about 14 times a day, taking readings from an approximately 1,864-mile-wide swath of the Earth turning below. The sensor takes both day and night readings in a spectrum that includes near ultraviolet light, visible light and infrared light. It measures land and sea surface temperatures, clouds and airborne particles and fires.
A second satellite equipped with VIIRS is scheduled to launch in 2017.
“The cool thing about this sensor is you can actually see city lights, you can see wildfires, you can see volcanoes, you can see moonlit clouds and moonlit terrain, even bioluminescence, at night,” Polivka said.
However, because of the Earth’s curvature, sensor readings cover a larger area on the outer edges of the satellite’s scans than they do in the center of the scan, more directly beneath the satellite’s orbit. To reduce the resulting distortion in pixel size, three data samples from the center of the swath are aggregated into one pixel before being transmitted to Earth. In an intermediate zone that begins about 330 miles out on each side, two data samples are combined into one pixel. From about 560 miles out to the scan edge, each pixel contains one data sample.
The potential problem occurs if one or more of the data samples in a combined pixel is saturated, meaning there was more light emitted in that spot than the sensor could measure. It’s like a too-bright photo that shows nothing but white or a 110-degree day with a thermometer that registers no more than 100 degrees.
The sensor doesn’t denote whether the combined pixels contain saturated data samples. Polivka performed calculations to identify data readings that might be suspect, as well as which of the sensor system’s 22 bands might be more prone to the problems.
The flaw could be remedied for future satellite launches by developing software that would flag saturated samples before they are combined into pixels, he said. The problem does not affect the remote sensor’s ability to correctly read surface and ocean temperature. It also can detect the presence of fires and hot spots, but questions about the data prevent analysis of fire temperatures and configurations when using the affected bands.
Source: University of Nebraska-Lincoln