[Geotiff] Creating GeoTIFFs using IDL
Brent Shaw
brent.shaw at wni.com
Wed May 4 10:56:31 EDT 2005
Greetings,
I am new to GeoTIFF. I am a meteorologist, and would say I have a
better than average understanding of projections than many
meteorologists, but still am unsure of a lot of the cartographic
terminology used in the GeoTIFF specification. I am currently able to
create georeferenced images of gridded data sets using an application
called IDL (Interactive Data Language, a 4DL used for data visualization
and exploration). My gridded data sets are generally either Lambert
Conformal, Polar Stereographic, or Mercator projections. For these data
sets, I have routines that allow me to define the grid based on 1 known
point in the grid, defined by its (i,j) index and a latitude and
longitude, the "standard longitude" (the longitude line that is parallel
to the grids y-axis), the standard latitude(s) (the latitude(s) at which
the grid spacing is exact), the grid point spacing (in meters) at the
standard latitude, and the dimensions of the grid. From that
definition, I am able to compute the latitude/longitude of any I/J point
and vice versa. In my convention, the grid (i,j) values originate at
the southwest corner of the grid.
>From this definition, I am also able to define a an IDL "mapping" window
from this information, such that IDL has a georeferenced canvas upon
which it can contour the data, plot points, etc. Within IDL, I can
determine the latitude and longitude of any pixel location. IDL
supports writing GeoTIFFs, but currently the GeoTIFF tags have to be
manually set. My problem is that I am having trouble figuring out the
appropriate values. I have looked at the GeoTIFF spec and some of the
Lambert conformal examples, but still cannot seem to set things
correctly. In particular, the whole concept of "false northing/easting"
has me stumped. I don't know the Pixel scale values, etc. It would
seem that if I can provide one or more "tie points" ,where a tie point
is a known pixel location ( [i,j] in raster units) and the
latitude/longitude, along with the projection parameters (type,
standard_parallels, standard longitude), that everything one needs to
establish a transformation is present, without having to know what the
pixel spacing is, etc. For example, I have a tangent Lambert conformal
grid with a standard longitude of -95.0 deg, standard latitude of 25.0,
and a known "cone factor." I also can provide the latitude/longitude of
all 4 corner points. Is that enough info to encode a geoTIFF, and if
so, which parameters should I be setting? Or, do I have to write some
code to compute things like "false origin", "false easting" etc.?
Sorry for the novice question, but after all my searching I am still not
clear. Thanks in advance for any help you can provide.
Best regards,
Brent Shaw
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.maptools.org/pipermail/geotiff/attachments/20050504/e1d70d7c/attachment.html
More information about the Geotiff
mailing list