[OpenLayers-Users] R: openlayers can't show an image layer in the right place starting from a center point

[Phil Scadden]

> the problem is this, whereas I am a newbie of this argument. I don't know
> what do you mean with "the image needs to be in the same projections" as
> your base layers. You mentioned to "create a layer in Geoserver from the
> image" or "doing it in the proper image transformation tools". Also I don't
> know what do you mean with " 400x400km on the ground or 400x400 projected
> kms". I am sorry, probably there are some assumptions that I should know
> that I am completely ignoring. What I know is only that I have a png image
> 400x400 pixels, each pixel is a km so it represents a linear square area of
> 400x400 kilometers and I have to draw it centered in a specific lat/lon
> point.
This comes down to really understanding what your image is. A figure 
that precisely represents  400x400 kilometers on the ground could not be 
a square. That's the reality of going from an ellipsoidal object to a 
flat image. So your 400x400 image is a projection of the ellipsoid onto 
a flat plane. The question is which projection - because you need to 
know that to get the image reprojected accurately. It will also depend 
on your own precision needs. If the image is derived from satellite or 
aerial photography then you need to find out what the rectification 
process (if any - though if you know it is 1px=1km then I will bet it 
has some). If the image is a construct (say from modelling), then it 
comes down to what assumption were made building the grid - the implicit 
projection used.

The question about projected meters versus ground meters is at the heart 
of projection.  Consider 900913 - this is a variant of the mercator 
projection. If you measure the a distance on the map at the equator, it 
will correspond well with the distance on the ground. However, if you 
measure a map distance at a latitude of 70degrees, then the map distance 
will be very much larger than real distance. (So Greenland looks much 
larger than it really is on a mercator map). A printed mercator map in 
an atlas will have a scale correction for latitude. You might want to 
look at a projection primer like 
http://www.colorado.edu/geography/gcraft/notes/mapproj/mapproj_f.html My 
bible is Snyder's "Map projections - a working manual".

Regarding tools. You will find the tools in all commercial GIS that I am 
aware of. ERDAS is what we use for photos and remote sensing stuff.  
Also commonly use ArcGIS. You will probably find adequate tools in GDAL 
and GMT in the freeware. I would strongly emphasis however, that you 
need to understand your image first. Normally you would enter the 
coordinates of the 4 corners of the image. At 1px per km, you could 
probably get away with pretty naive calculation of those coordinates for 
model output. Will be tougher if this is remote sensing imagery where 
you might get visible mismatch with google imagery.

Notice: This email and any attachments are confidential.
If received in error please destroy and immediately notify us.
Do not copy or disclose the contents.