Hello List:

Was wondering if anyone has experienced anything similar to the following.
Let me state initially this is probably my lack of understanding not a
software problem.  I have an .ERS file that is several USGS quads in width
and height.  The pixels reflect elevation heights and the regions within the
.ERS file reflect the 7.5' neatlines of the quads.

When building a USGS DEM for instance, it is important to ensure that the
center of each pixel is at a UTM coordinate that is evenly divisible by the
cell or pixel resolution (i.e. 30 or 10 meters).  Typically after I generate
the expansive, multiquad DEM, I choose a pixel and using the geoposition
wizard, I assign that pixel (i.e. 1275.5, 2345.5) a known point UTM position
(i.e. 375650, 4625010).  This effectively shifts the grid a fraction of a
pixel and assures my pixel centroids are in the correct UTM position
relative to being evenly divisible by the resolution.  I then burn in my
regions for the individual quad neatlines.

I use a formula to turn all pixels outside the region to null and leave all
pixels inside the region alone and save as a new algorithm.  I crop the
edges of the algorithm window and save to a new, smaller .ALG.  When I load
that new .ALG and overlay the neatline polygon from which I built the
region, everything fits perfectly.  If I use the geoposition tools in the
VIEW option and key in any pixel (i.e. 1275.5, 2345.5) I get a UTM with all
zeros to the right of the decimal point.  Perfect.  However if when I use my
formula and save as a new .ERS my grid shifts slightly and my UTM values at
pixel centroids no longer are evenly divisible by the resolution.  They have
lost that "registration point" information.

Obviously, I should just save individual quads as .ALGs.  Well here is the
rub.  When I export the .ALG or even reposition the .ERS, and then export it
to .XYZ, I get extra (non-null) pixels on some quads, typically along the
north and east edges.  This is random and intermittent and typically only
one extra pixel at the end of each column.  If I have 15 regions of adjacent
quads in an .ERS and extract all to .ALG and then go directly to export XYZ,
30% - 50% will have extra pixels.  If I re-introduce the same regions to the
same .ERS file and convert to .ALG and export to XYZ, several will now not
have extra pixels.  I simply keep doing this until none have extra pixels.
Am I missing something here?  I never re-shift the original .ERS file and
all the neatlines are exactly correct and fit perfectly side by side.  Does
ERM change position of the data ever so slightly upon export to .XYZ?
Perhaps I am not understanding subtle issues with projected data and ERM.

I run into this problem on every WINNT and WIN2000 machine I have used
running ERM 6.0 - 6.21.  Any thoughts would be very helpful.

Greg


-----------------------------------------------------------

To make changes to your subscription, please visit our website, 
http://www.ermapper.com/technicl/ermapperl/index.htm

Reply via email to