High-Frequency Satellite Data to New Level
One of Planet’s mantras is: “You can’t fix what you can’t see.” The satellite solutions company offers high-frequency data collection – collecting imagery around the globe every day to assure that people needing data for agriculture can get the imagery they need when they need it, said Stephanie Giard, Planet’ Senior Product Marketing Manager for Agriculture, in a recent webinar, “Enhanced Vegetative Health Maps From Satellite Imagery Using the Red Edge Band.”
“The Planet platform is integrated, scalable, and interoperable,” Giard said. “It can process terabytes of data on the fly from Planet’s imaging constellation and easily connect with other sources, workflows, and existing platforms.”
Remote sensing isn’t new in agriculture, but until Planet was formed in 2010 there had been limited coverage, Giard said, noting that Planet offers the highest frequency satellite data commercially available.
The company’s newest additions to its PlanetScope Monitoring product are its 80 SuperDove satellites. The past two generations of Planet’s Dove satellites include four bands: blue, green, red, and near infrared. The new eight-band SuperDove satellites include those four bands as well as coastal blue, a second green band, yellow, and a red edge band.
“The most interesting band by far for agriculture is the red edge band,” Giard said. “It is the most sensitive to changes in plant health. Indices using that band don’t saturate as much after canopy closure as those using the near infrared band. So, this makes the red edge band ideal for monitoring variability later in the season after canopy closure.”
Misty Tucker, Global Industry Principal for Agriculture for Planet, says the normalized difference vegetation index should no longer be viewed as the “be-all-end-all index” for crops throughout the entire growing season.
“If we move toward an index that utilizes the red edge band, we can get more canopy penetration. That is a far better tool for monitoring things like pest and disease pressure during the late season,” she said, noting that diseases that start lower in the canopy, such as powdery mildew, can be easily detected.
Expanding on the red edge band, Tucker explained it is a “narrow area of change between red and near infrared, which is extremely sensitive to different phenological changes within the plant.” A few of the red edge indices that are particularly useful are the normalized difference red edge index, the red edge-based chlorophyll index, and the modified chlorophyll absorption in reflectance index, she added.
Giard also discussed some of the improvements Planet has made in its remote sensing data, beginning with a decrease in hot pixels and columns in PlanetScope Monitoring images. “Hot pixels or hot columns are pixels that are a lot brighter than they should be,” Giard said. “Planet has developed a new algorithm to catch and fix hot pixels and columns for all its Dove satellites. The new algorithm detects and corrects about 98% of the hot pixels and columns.”
Another improvement the company made was “sharpening” its SuperDove visual images, Giard said. “We developed and deployed a new sharpening kernel to account for changes in the telescope design in the SuperDove satellites,” she added. “There is now greater clarity and a higher level of detail, and the after-visual image is greatly improved and makes identification of objects much more accurate.”
Planet also improved the SuperDove band-to-band alignment. “The SuperDove sensor plane consists of eight horizontal stripes that capture eight different spectral bands (coastal blue, blue, green one, green two, red, yellow, red edge and near infrared),” Giard explains. “To form one multiband composite of the SuperDove imagery, we combine all the frames for those individual bands. Previously, our technique would sometimes result in a visible misalignment between the bands. However, all images acquired after March 1 of this year are processed through a new compositor and registration process, which produces much better results and reduces misalignment between bands. We are now able to output imagery with alignment errors of less than half a pixel.”