GRS Presentation at IGTF2016 Conference: Ken Stumpf Wants "No Nuts" in His Landsat Imagery
April 15, 2016 - For Immediate Release - GRS Starts Petition at Change.org to Reinstate Option to Use Nearest Neighbor Resampling Algorithm During LandSat Terrain Correction Processing
Ken Stumpf of Geographic Resource Solutions started his presentation by comparing the Free Landsat Imagery currently available from the EROS Data Center with Free Ice Cream.
Stumpf said "Imagine you really like ice cream and a new ice cream store opened in your town that advertised FREE ICE CREAM, as much as you can eat."
If you were me, you would say “WOW that’s great,” and go to the ice cream store to get some.
But when you got there and surveyed the menu you found that every flavor of the free ice cream contained NUTS – Lots of NUTS … and you don’t like NUTS!
It turned out that the FREE ICE CREAM was not so great a deal after all.
That’s sort of the problem Stumpf has with this free Landsat imagery being distributed by the EROS Data Center. What are these "nuts" in the imagery?
Stumpf went on to compare the digital artifacts introduced by the Cubic Convolution resampling algorithm to the Nuts in the Ice Cream and demonstrate that while the Cubic Convolution algorithm could create a "prettier image" it was at the expense of both spectral and statistical distortion. Such distortion is known to introduce errors into analyses and classification/mapping efforts based on this resampled imagery as opposed to Nearest Neighbor resampled imagery. Several examples were shown that illustrated both potential statistical and visual effects of these different resampling algorithms.
In conclusion, Stumpf related that he had started a petition at Change.org to seek reinstatement of the Nearest Neighbor resampling algorithm as an option along with the Cubic Convolution algorithm enabling the image user to have a choice. If you are involved in image classification/mapping, change detection, and/or other image quantitative analyses or support this effort to make Nearest Neighbor resampled imagery available for those who work with Landsat imagery then we suggest you review and sign Stumpf's petition at Change.org.
Links to the petition and to Stumpf's presentation may be found below associated with each image: