The location of every pixel in a 2-dimensional satellite image to a specific spatial location on the ground. Geometric calibration may be done by either re-sampling the image in a specific projection or by storing the geometric location information with each pixel in the image. If pre-launch values of the instrument “MTF” have been accurately enough made prior to launch, then an MTF correction can also be made as part of the geometric ground processing. Resampling for TM-type imagery is usually done using cubic convolution algorithms with an accuracy of ±50 m for 30 m ETM+ pixels from inputs that include “definitive” ephemeris and known characteristics of the non-uniform scan rate. This geodetic accuracy refers to how close the knowledge of the pixel’s location is to the actual location on the Earth. MODIS sensors on the Terra and Aqua satellites store calibrated imagery as geolocated information for each pixel. Systematic geometric corrections correct for sensor-related factors, either with or without the use of ground control points for the specific image. Such systematic products can be significantly improved if topographic information is available and used. In TM-era products the approximate 8-bit radiometric precision was reduced with each geometric re-sampling, often by simple rotation and translation, to a new 8-bit product, and therefore it was desirable when radiometrically or geometrically re-calibrating an image, to normalize it to best estimates, to start from raw uncalibrated imagery. Also see modulation transfer function and scan line length.