I have a raster dataset that is supposed to be in the Equidistant Cylindrical projection - but when I import it, it is also in decimal degrees. I have searched through all the projections available in an attempt to define this correctly, but can't seem to find something appropriate. Equidistant Cylindrical (and it's equivalents like Plate Caree) seem to all have linear units of meters.
No, you wouldn't be able to make any projected coordinate system with "degrees" as the linear unit, unless you made a custom unit. In equidistant cylindrical, for a equal-length box of degree intervals (like 10x10), the north-south distance should be larger than the east-west. I wonder if the data is in pseudo-plate carree. That is, it is in decimal degrees and should be defined with a geographic coordinate system, which are being treated as if they're linear.
Was wondering if a solution came up for this. I think I have a similar problem... I've downloaded MODIS aqua chlorophyll and sea surface temperature from the ocean color website http://oceancolor.gsfc.nasa.gov/.
These exist as hdf files and I used a program called FWTools to convert them to geotiffs. The metadata in the hdf file lists the projection as "Equidistant Cylindrical". However, it lists the bounds and units as degree units which is very perplexing. Melita, have you worked with this data set before? I can't seem to find a projection in Arc that works - all "Equidistant Cylindrical" projections are in units of meters. From the hdf metadata:
:Map Projection = "Equidistant Cylindrical" ; :Latitude Units = "degrees North" ; :Longitude Units = "degrees East" ; :Northernmost Latitude = 90.f ; :Southernmost Latitude = -90.f ; :Westernmost Longitude = -180.f ; :Easternmost Longitude = 180.f ; :Latitude Step = 0.041666668f ; :Longitude Step = 0.041666668f ; :SW Point Latitude = -89.979164f ; :SW Point Longitude = -179.97917f ;
Well, they may be marking it as equidistant cylindrical because that's how you end up viewing it in a 2D display. That is, if the data is georeferenced in degrees, and the software just displays it, it's using a pseudo-equidistant cylindrical (aka Plate Carree) projection. I would define it with a geographic coordinate system and see if that works.
I tried defining the projection as unprojected WGS84. This results in an "inconsistent extent" warning. I've also tried Plate Carree under Projected Coordinate Systems > World (as well as under > Sphere-based). Can I manually go in and change the extents somehow if I use GCS WGS84?
The OceanColor site that serves up this data provides a tool called SeaDAS that allows you to reproject the files - you simply select which projection you want. I tried this and selected Geographic Lat/Lon WGS84. The output lined up with a shapefile I had of the globe. But I need to batch process this for dozens of files and SeaDAS would require learning an entirely different programming language to batch process. So somehow this SeaDAS tool is able to read in and reproject it correctly, but I can't find any information on their forums on what the data is initially in - they just say to check the file metadata for this info which I listed above. It's like a catch-22!