Contour/Water alignment problem on my first topo map

[masdangkolo]

Help,
I'm trying to build my first topo map.  It is of the island of Guam in the pacific.
I'm using the instructions from
http://home.cinci.rr.com/creek/garmin.htm
as best as I can.  I used this tutorial instead of the one on this site because I only have 1 file for the water data and thought this method was shorter.

Since Guam is not one of the 50 states I had to get some of the data in slightly different ways, although I don't these differences would affect the map outcome.


Downloaded DEM data from usgs seamless server
2 sections
http://seamless.usgs.gov/index.php

Converted tif to mp using dem2topo (2 files)

Loaded both .mp into GPSMapEdit, set levels/layers and saved as new mp

Requested hydrography data for guam from nhd.usgs.gov

Loaded NHDFlowline.shp and NHDWaterbody.shp into GPSMapEdit.

When zooming in to an edge of the island, it seems that the outline of the island (which is in the NHDFlowline.shp file) does not match the contour lines.  I looks like the flowline data is about 1500 ft to the east and 500 ft to the north.

The .prj file from the DEM data says
Projection    GEOGRAPHIC
Datum         NAD83
Spheroid      GRS80
Units         DD
Zunits        METERS
Parameters

The .prj file from the NHD data says
GEOGCS["GCS_North_American_1983",
DATUM["D_North_American_1983",
SPHEROID["GRS_1980",6378137.0,298.257222101]],
PRIMEM["Greenwich",0.0],
UNIT["Degree",0.0174532925199433]]

Does anyone know how to fix this?  I tried to use many other datum settings, but none seem to line up.
At this piont, I'm not really sure which is correct, the contours, or the water data.

Attached is a section of the map that illustrates this problem (hopefully it posts).

[-Oz-]

The interesting this is that both of those datums are the same.  Both using the same projection, units, and datum.

It does definitely appear to be a datum error though.  The .mp file created by dem2topo is always in wgs84 (which is close enough to nad83 to not matter).  I can get seamless to load right now so I can't download and check but if you could post the link the nhd gave you to the water data that would help me.

I wonder if guam has its own set of data.  To check which is correct I would find coordinates for a summit or something of that sort and see if it appears in the water or on land (which land really).

[masdangkolo]

I've been fighting with this on multiple fronts.  Hopefully I'm keeping it all straight.
The NHD data link is:
ftp://nhdftp.usgs.gov/NHD136010.zip

I have loaded the NHDFlowline.shp fie into GPSMapEdit specifying NAD83 as Datum

then from TIGER 2007 data:
ftp://ftp2.census.gov/geo/tiger/TIGER2007FE/66_GUAM/66010_Guam/
I imported fe_2007_66010_edges.shp specifying Guam 1963 as Datum

Both of these shp files seem to compare well (within about 100 ft), but are significantly different than the contour data.

Will look into finding a reference point to try and determine which is correct.  Is it possible to tell dem2topo that the tif is in datum other than WGS84?

[-Oz-]

I downloaded the edges from the census data and converted them to nad83.  Let me know if they line up with the contours or water or what.  Knowing that will definitely help us determine where to go from here.

I attached the modified file to this post.

[masdangkolo]

Dan,
Thank you for your efforts.

This is still somewhat puzzling to me but I have an idea. When importing the converted edges file you provided (census) seems to be about 50 feet different from the data I got from the NHD.

As for the contour data, I finally have figured out how to use FWTools (I think, at least it produces output).  Using gdalwarp, I assumed the geotiff data was in Guam_1963 datum and converted it to NAD83 (even though the seamless server claims to be NAD83). Then generated contours with dem2topo.  Loaded the mp file and both the census edges and NHD data into GPSMapEdit (census is in black and NHD is in blue).  The HND data seems to match the contours quite well.

I have two data points from a Garmin 60csx that seem to match quite favorably with the contour data.  I am still trying to get some better points to verify the contour data.

What I still don't understand is why the edges from census and flowlines from NHD are so different.

Can you sed some light on my use of gdalwarp?  I used:
gdalwarp -s_srs EPSG:4675 -t_srs NAD83  input.tif  output.tif

Thanks,
Roy

[-Oz-]

which edges are which?  the one that takes off toward the top of the screen is which thing?

The census data includes boundaries for counties I believe.  These boundaries are normally in the water for islands.

[masdangkolo]

Dan,
I've annotated the lines on the attached jpeg.  I guess my color choices didn't provide enough contrast.
Roy

[-Oz-]

i'd be willing to say the census data is not detailed enough.  Or, they announced they had some error in a few counties, guam is probably one of those.  The flowline looks right though

[masdangkolo]

Is it possible to find out what the erroneous projection is and then correct for it?   I thought it was just an incorrectly specified datum, (Guam  1963) but after converting it still seems that it is still in error.  The error appears to be different as latitude increases.