SMARTER STRATEGIES

Cloud Democratizes Smart Planet Imagery

R. Colin Johnson | Date: 04-06-11 | Comments

A U.S. Geological Survey policy recently democratized access to satellite imagery, and now easy-to-use cloud-based software brings smarter-planet analytics to the people.

The government has made its Landsat archive of global satellite imagery available to cloud-based services offering free access to sophisticated pattern analysis software, allowing almost anyone to perform smarter-planet studies.

Since 1972 the U.S. Geological Survey has been cataloging millions of satellite images that track global changes in agriculture, cartography, geology, forestry, regional development, education, national security and plain-old surveillance. The latest Landsat 7 satellite scours the globe in eight spectral bands with resolutions down to 15 meters that are updated every 16 days. The next-generation Landsat, due to be launched in 2012, will combine nine spectral bands in the shortwave portion of the spectrum (visible, near infrared and shortwave infrared) with two longwave (thermal) spectral bands.

The Landsat 7 satellite offers Earth images with resolutions down to 15 meters. (Source: NASA)

Traditionally, access to the imagery was restricted and the software analytic tools for deducing trends like deforestation from the database required thousands of dollars' investment and extensive training, but free access to the imagery has prompted free cloud-based analytics to be performed by untrained users�basically anybody with an Internet connection.

"When the U.S. Geological Survey made its data free, all of a sudden this whole new world opened up to us," said professor Matt Hansen at South Dakota State University (SDSU). "Landsat data went from $600 per image�which measures 185 kilometers square�to being free."

Hansen, the co-director of the Geographic Information Science Center of Excellence at SDSU, worked with Google last fall to help launch its Google Earth Engine, a cloud-based service that can access the entire Landsat database.

Google's cloud computing capability is set up to quantify all kinds of smarter-planet parameters, from deforestation to natural disasters to cropland area and urbanization. As an example, Hansen and postdoctoral researcher Peter Potapov tested out the Google Earth Engine by combining 50,000 images into a detailed map of Mexico. Without the free access afforded by the U.S. Geological Survey to those Landsat images, that map would have cost upward of $30 million to assemble. In addition, special software for processing such huge data sets and skilled operators would have been required on a project that Hansen and Potapov were able to perform on their own.

Landsat data here in the thermal ranges reveal the island of Hawaii's hardened lava flows (black), a thin grayish plume of smoke rising from the Kilauea volcano (near the island's southeastern shore), lush tropical forests (dark green) and patchy areas of sugar cane (light green). (Source: NASA)

With democratization also comes responsibility, according to Hansen, who is promoting a collaborative effort between academics, government scientists and private industry to process, characterize and validate derivative satellite data sets.

Hansen is currently developing accuracy assessment tools for these derivative data sets, such as those predicting global warming, to help ensure the accuracy of what he calls "map products"�to ensure that they maintain the accuracy of the imagery in revealing how Earth is changing.