ng a resolution for an entire system, you can also think about resolution from an image-specific point of view. In general every image has a ground resolved distance (GRD) that depends on spatial resolution, contrast and texture as well as on the characteristics of the system that was used to capture the image. As an example, image a winter scene and a summer scene of a northern deciduous forest with clearcut areas. The winter scene will have higher contrast (snow vs. trees) than the summer scene (snow vs. green meadows). Thus, for a study trying to identify the size of clear-cuts in the area, a TM image in winter will have better GRD than one taken in the summer, and smaller clear-cuts would be able to be distinguished in the winter scene. As far as ground resolved distance is concerned, our best case so far (with a 1000mm lens) is resolving 2 to 3 meter objects on the ground. These have included building shadows, roads, and boats in harbors. Most photographs have much lower resolution. Since astronauts are traveling so fast in orbit, why don't the photographs look smeared? If you do the calculation for the amount of ground smear expected based on the shutter speed of the camera and the velocity of the International Space Station relative to the ground, you would expect that the ground resolved distance could be no better than about 30m. Because we have much better resolution (i.e. boats in harbors, narrow roads), we know that consciously or unconsciously, astronauts track the Earth when they take a photo, so the resolution is actually better than you would calculate. How can I use astronaut photographs in scientific research? The photographs have a great deal of information value "as-is" because they provide a broad geographic perspective on climate, geology, ecology, oceanography and other disciplines. In addition these photographs provide the longest possible time series of information documenting changes in the Earth. What does the "Features from Machine Learni