When GIS was young and computers had limited power, it was advantageous to limit the use of the projection equations. This meant that you would reproject a data set at most once, and work with that data set in the same projection as the map. The USGS traditionally used UTM coordinates for its data sets, leading to challenges on the boundaries.

As computer power grew, the GIS software could reproject on the fly. This led to a desire for seamless data, with no problems at the edges of the data sets, and geographic coordinates to store the data. USGS went this route when they rolled out NED and Seamless Server; NGA had always use geographic coordinates.

Geographic coordinate maps (plat caree) should only be used for global grids when significant distortion cannot be avoided; for large scale maps, there must be differential scaling in x and y.

Table 1. General case

Data Set, Grid coordinates | Data Set, Projected coordinates | Data Set, Geographic coordinates | Map, Projected coordinates | Map, screen coordinates |

Integer, window into earth | Float | Float | Float | Integer, Window into map |

DEMs, start 0,0 in SW
corner Satellites, start 0,0 in NW corner |
Equal spacing x and y, in projected meters | Equal spacing x and y, in degrees, minutes,
seconds Very rarely adjust spacing for tiles near poles |
Projected meters | Stats 0,0 upper left |

SPCS (TM, LCC) UTM AEA PS |
||||

NLCD | NED SRTM DTED ETOPO |
|||

---------------> Simple translation, scaling (corner, dx, dy) |
---------------> Data set projection, inverse equations Expensive to compute |
---------------> Map projection, forward equations Expensive to compute |
---------------> Simple translation, scaling, rounding (corner, dx, dy) |
Display |

<------------ Simple translation, scaling Can round, or interpolate |
<------------ Data set projection, forward equations Expensive to compute |
<------------ Map projection, inverse equations Expensive to compute |
<------------ Click on map Simple translation, scaling, |

Projection and datum shift calculations

last revision 11/28/2018

Table 2. Least expensive option, boundary problems

Data Set, Grid coordinates | Data set and Map, Projected coordinates | Map, screen coordinates |

Integer, window into earth | Float | Integer, Window into map |

---------------> Simple translation, scaling |
---------------> Simple translation, scaling, rounding |
Display |

<------------ Map projection, inverse equations Expensive to compute |
<------------ Click on map Simple translation, scaling, |

Table 3. Seamless case using geographic coordinates

Data Set, Grid coordinates | Data Set, Geographic coordinates | Map, Projected coordinates | Map, screen coordinates |

Integer, window into earth | Float | Float | Integer, Window into map |

---------------> Simple translation, scaling |
---------------> Map projection, forward equations Expensive to compute |
---------------> Simple translation, scaling, rounding |
Display |

<------------ Simple translation, scaling Can round, or interpolate |
<------------ Map projection, inverse equations Expensive to compute |
<------------ Click on map Simple translation, scaling, |