#88 closed defect (fixed)
Whole world projections fail at the 180 degree crack.
| Reported by: | Norm Olsen | Owned by: | Norm Olsen |
|---|---|---|---|
| Priority: | minor | Milestone: | 2.0 |
| Component: | Library | Version: | |
| Keywords: | Mercator | Cc: | HuguesWisniewski |
Description
There appear to be several of the "whole world" type of projections that behave poorly at the 180 degree crack.
Note:
See TracTickets
for help on using tickets.
This is true of the Mercator (all three variations), the Eckerts (IV and VI), the Miller, and the American Polyconic. (The Polyconic is not really a whole world projection, but the smae bug was found there so itsd being corrected with the submission associated with this ticket.)
Essentially, the anti-meridian (i.e. the central meridian plus/minus 180 degrees) was excluded from the domain of these projections. Trying to convert a point on the anti-meridian produced a Useful Range" error. This is corrected in all seven instances to consider the anti-meridian a part of the domain and useful range.
Additionally, to solve the specific case in which this defect was discovered, the Pseudo Mercator projection has had an additional change made to it. This change recognizes the special cases where this projection is used; specifically in handling data produced by Google, Bing, Yahoo, and other providers of map data over the Internet. This change applies only to the Pseudo Mercator variation of the Mercator porjection and considers both poles to be within the domain of the projection, and further applies a limit of 85 degrees to latitude values (which is equivalent to 20,000,000 meters in the northing). This change enables user to obtain whole world data which in this projection.
Technically, the poles are sinularity points in the Mercator projection. The normal action of the CS-MAP Meractor implementation in the case of a latitude very close to either pole (about 50 centimeters) is to return a northing on the order of +/-111,000,000 and return a "domain error".