[Geotiff] Linear Units of Projections Parameters

[pw]

Hello,

That's a good question. I don't think I've really taken much notice
until this question was posed.

I use primarily ENVI (RSI) and Geomatica (PCI) and as far as I
remember I have never had a GeoTiff image that wasn't in a meter
based projection.

These programs, however, tend to declare the projection units in the
file header as part of the image geography.

In ENVI a separate text header file contains a 'map-info' declaration
with a pinned geographic position relative to one pixel
(usually the upper left corner) and the units of measurement including
the X scale and Y scale in those units.

In PCI the PIX file contains geographic information internal to the
format. (the geo-segment) This segment, much as in the ENVI header
contains the relative X and Y pixel scales based on the projection units
for the image.

The geographic units may be defined as meters or feet, and this is how
binary and PIX images are loaded and displayed by default.

Now, whether the end user adjust the program *Display* to translate
from feet to meters, or vice versa, is a completely different question.
That is a user preference.

Being a file format, and not an application, it seems logical that
the format should not make assumptions regarding the function of 
applications.

/*................................................................*/
Another issue that I thought I should mention is the definition
of pixel size. This should be defined in specific units at least
once for each dimension X and Y.

The X scales and Y scales are defined separately, since the pixel
dimensions of much remote sensing data may not be square, which can 
cause mapping problems later if programs make assumptions regarding
square pixels.

  Along the same vein, it may be relevant to define multiple pixel
sizes by position in X and Y. This is because many satellite images
have *bow tie* (scanners) effect where the pixels in the center of the 
swath (nadir) are, say, 1 Km and the pixels are 2 km at the outer edges
of the swath. They overlapp eachother at the edges of the images.
This is due to the curvature of the earth across track,
and other factors in the satellite oribital geometry.
   This is not the same thing as applying additional tiepoints to
merely position an image on a map.


Peter


Frank Warmerdam wrote:
> 
> Peter,
> 
> I wonder if you know of packages that interprete the false easting
> and northing as always being meters, or packages that (properly apparently)
> to be in the projection linear units.
> 
> Best regards,