March 13, 2020 - A Landsat Milestone: One Hundred Million Downloads
When the world began downloading the first freely available Landsat images on Oct. 1, 2008, a lot of people wondered just how much imagery would fly out of the U.S. Geological Survey’s (USGS) web-enabled doors.
The answer, it turns out, was a lot.
Return to all Landsat Headlines
Ten months after that October start, the one millionth free scene was downloaded. Three million were downloaded by August 2010, and 6 million by August 2011. On March 9, 2020, an image acquired almost three years earlier by Landsat 8 of New Zealand’s North Island claimed the status of the 100 millionth download from the USGS’ user interface tools (EarthExplorer, GloVis, and LandsatLook) in a little more than 11 years.
For those who lobbied for free and open Landsat data, who track its usage today, and who rely on it for the science they pursue, 100 million is awe inspiring, yet not entirely unexpected.
“It’s a gigantic number, so it’s kind of hard to fathom,” said Barb Ryan, who as the USGS’ Associate Director of Geography signed off on a policy paper with NASA’s Mike Freilich on Jan. 2, 2008, that led to the decision by the USGS, the Department of Interior, the Office of Management and Budget, the White House, and Congress to make Landsat data free and open.
“But I think what it means to me even more than the downloads,” Ryan added, “is that these images are finally getting used.”
Landsat’s $3.45 billion annual impact
Used in a big way, in fact. Landsat images provide domestic and international users a $3.45 billion economic benefit each year. That estimate is based only on images downloaded through the USGS. It doesn’t include scenes downloaded by cloud vendors. Nor does it take into account other downstream economic benefits.
That said, the Landsat images from the USGS are invaluable for measuring global water consumption and assisting in monitoring forest health and agricultural production. History has shown that each new improvement in Landsat science expands the use of technology by operational and scientific groups looking at such things as urbanization, homeland security, disaster mitigation, and more.
Before free and open Landsat data, Senior Research Scientist Mike Wulder with the Canadian Forest Service said a researcher was fortunate to get a single clear image to represent an entire growing season. When the Landsat floodgates opened in 2008, scientists began building algorithms that removed impediments like clouds, shadows, smoke, and haze to create all kinds of clearer imagery, Wulder said. In time, those algorithms got even stronger and better.
In his case, the availability of all Landsat data brought Wulder valuable new insights into how forests change over time. For example, forestry scientists like himself mapping land cover changes can capture not only where and when a disturbance such as wildfire or timber harvest occurred, but how these areas return to forests over time.
“Before, it was a big deal to get any clear imagery for a given year, let alone getting two clear images for change detection. Now with so many images we are able to get beyond detection of simple change and speak to trends and dynamics over large areas,” said Wulder, who is a member of the Landsat Science Team. Then when access to a proliferation of free images became standard, “the line I like to use,” he said, “is that now we’re only limited by our imagination; not by the data and its availability.”
Landsat images once ran as high as $4,400 each
Prior to the free-and-open policy, users paid for each image. When a commercial venture called EOSAT operated the Landsat program from 1985 to 2001, that fee ran as high as $4,400 an image.
Needless to say, there weren’t a lot of acquisitions sold at that price, said Kristi Kline, the project manager for the Landsat Archive at the Earth Resources Observation and Science (EROS) Center. At the highest point after the launch of Landsat 7 in 1999, maybe 25,000 images were purchased annually, Kline said.
While that fee structure enabled the government to recoup processing and other costs, the reality is that most of the purchases were actually made with Federal government dollars, Ryan said. When the USGS looked at who was buying images, it found it was largely other Federal agencies, universities funded by the National Science Foundation, or contractors getting their funding from the Defense Department.
“So, at the Federal government level,” Ryan said, “we were just taking money from one pocket and putting it into another. And every single agency was incurring administrative costs to buy and sell those images, so that was the argument we made when we said, ‘Just stop.’ ”
She said a lot people as far back as 1972—when the first Landsat satellite was launched—argued that the data needed to be broadly and widely available. But the problem in those pre-Internet years was that the USGS and Department of Interior simply weren’t resourced sufficiently to handle user demand.
When the 2000s dawned and the Internet took off, the World Wide Web became the answer to the naysayers, Ryan said. “With the rationale that there was no incremental cost for delivering an additional Landsat scene” because of the Internet, she said, “we used that to finally make the argument that the policy needed to be changed.”
Still, even when the images became available for free, there were hindrances to overcome. Just because scientists now had access to much more imagery didn’t mean they had the bandwidth to download all of it, or the computing power and disk space to pursue big data projects.
Unleashing the power of the cloud on Landsat
That’s all changing now that cloud technology stands on the precipice of unleashing the informational power of 100 million Landsat images, Kline said. The move by USGS and EROS to move a copy of the Landsat archive into a commercial cloud and provide users access to it will enable them to run their algorithms against the data in the cloud faster and more efficiently than ever before, she said.
“Today, we’re talking about the 100 millionth download, but in a few months, when we move a copy of the archive into the cloud, we’ll be providing a capability where users really don’t have to download,” Kline said. “I’m not sure we’re going to be talking about the 200 millionth download for a while. But we might start talking about how much data is actually getting used after it moves to the cloud. I think that’s going to be the significant change that’s made.”
From a science perspective, that big data will become even more usable in the super-computing world of today because of decisions made long ago, Wulder said. The early Landsat engineers and decision-makers were shooting for more than just big numbers when it came to imagery, he said. They wanted an archival record calibrated across sensors and missions so that spectral values or reflectances acquired at one place in, say, 1992 provided the same accurate information 20 years later when a new and improved Landsat and its sensors passed over the same swath of land again.
“Early decisions were made with regards to having calibrated radiometry that have enabled opportunities and not regrets,” Wulder said. “If we hadn’t gone with calibrated radiometry, all these images wouldn’t be as useful. They wouldn’t be able to be compared through time. They wouldn’t allow us to develop algorithms that are portable both over space and time. Without that calibrated radiometry, a lot of what we do now wouldn’t be possible.”
But it is possible, and today he and others can look through the portal of remotely sensed time and space and use those millions of images to not only chronicle what was on the land, but monitor what is there now and project what may be there in the future. While there are certainly other important remote-sensing missions that are valuable when it comes to Earth observations, Landsat and its archive really have provided the example for how to go about doing this, Wulder said.
With its calibrated radiometry, its systematic global data collection in concert with international partners, the placement of its data into an archive in a timely fashion, and its move towards sophisticated Analysis Ready Data products, Landsat has shown other space programs and data vendors what they need to do to make their data more useful, he said.
“You can launch a new satellite, but you can’t put something up in the past,” Wulder said. “Landsat will always be that historic record that new satellites can tie themselves to.”