January 15, 2013
NASA, USGS Program Helps Find Patterns In Forest Disturbances
[Watch Video: LandTrenr Analysis of Pacific Northwest]
Lee Rannals for redOrbit.com — Your Universe Online
A new computer program is now able to help find patterns that had been buried within vast amounts of scientific data.
LandTrendr has already led to seeing an obscured, slow-moving decline and recovery of trees in Pacific Northwest forests.
Scientists found after comparing satellite data to ground data, that the cause of this decline is due to bugs, according to Robert Kennedy, a remote sensing specialist at Boston University.
The pattern showed a long slow decline of tree health over years followed by slow regrowth. It emerged in several areas, peaking in 1992 when regrowth began, and near Mount Rainer where the insect outbreak lasted ten years from its onset in 1994.
Kennedy created the LandTrendr program to work with data from the Landsat satellite program. With this new way to view Landsat imagery, it is changing how the Forest Service in the Pacific Northwest operates its yearly forest monitoring program.
Kennedy said that LandTrendr works because of the unique nature of Landsat data. The data embedded in images are a scientific record of the Earth's surface that goes back 40 years.
Each of the images cover an area of 115 miles by 112 miles, providing data for wavelengths of light reflected or emitted from the Earth's surface.
"Studying big areas over many years means handling big data sets and figuring out how to get all the data to work together for meaningful comparisons," NASA said in a statement. "One challenge in particular is finding images from the same time of year where the view of the ground is not hidden by clouds."
Kennedy's breakthrough was to combine cloud-free pixels from multiple scenes of the same area collected over the growing season in late summer. He compares new images for each year to one another, and breaks the scene down into smaller sized pixels.
"We're getting better data use out of what people think of as crummier images," Curtis Woodcock, a remote sensing specialist at Boston University, said.
Kennedy first saw a yearly succession of moderate resolution Landsat satellite images of a 13,000-square-mile area near Portland, Oregon. Kennedy said that it was like looking at a pair of air photos of a forested hillside offset so that with special glasses, the image pops into 3D clarity.
"You sort of squint your eyes and it takes a while and all of a sudden you get that moment -- boom! Oh, my god, it's a landscape!" he says. "I had that same sense [of amazement] when I first started looking at the time series stuff."
Woodcock said that LandTrendr is a remarkable tool for looking at forestation changes over a period of time. He hopes to eventually see a move from the retrospective, to the real time.
"The goal in the long run is to be able to provide land managers information on what's happening as it's happening," he said.