Welcome to GPSFileDepot!

Main Menu

How to deal with x,y displacement

Started by hwstock, June 07, 2013, 05:23:13 PM

Previous topic - Next topic


Howdy -- I've made a img (garmin) format map for an area of Baja, Mexico, in the bounds of 30 to 32 degrees north, 115 to 116 degrees west.  I used the process given in this site's tutorial (downloading data from the USGS NED server, converting gridfloat to geotiff via usgs and gdal utilities, stepping through Dem2topo, then gpsmapedit and cgpsmapper). I have some exact gps waypoints from the area, and can confirm that the countours are displaced approximately 200 feet west of the actual positions.

I assume that somewhere in the process, I can use obe of these clever programs to displace the contours 200' east, so they will coincide with the actual terrain.

Any idea where I would add this displacement (and how)?  It looks like the affine translation GPSMapEdit will work, but it looks bizarrely ungainly for my needs-- I really don't need 3 pair of points, I just need to displace the map in a well-known direction.  I don't want the imprecision of picking points off the map in GPSMapEdit.

I'm sure this error is real, because I previously bought the cartografia garmin-compatible maps for all of Mexico, and also looked over the competitors' garmin-compatible maps for that area.  Both were displaced by about 200'. I contacted the technical people at cartografia, and they confirmed the 200' displacement in that area, and were kind enough to fix the maps and send me the update.  However, I've found that some of the areas of interest for me in Mexico have similar displacements, probably due to some mismatch in datum or whatever. I'd like to make custom maps of those small areas, and correct them for the displacements.
EDIT: I figured out how to do it in GPSMapEdit -- pretty simple in the end.  Just make a gps file with 3 waypoints near edges (and not colinear!), and make 3 more that are displaced by the exact same amount.  Then add that gpx file to the project, and use the points to do the affine mapping by drag and drop. Worked like a charm.


I would try to figure out why the contours don't line up. My guess is that they don't have the same datum and when you imported the DEMs you selected the wrong datum.

Of course, your pragmatic solution is just fine.


Datum issues were my first guess; but the misalignment doesn't match the expected offset (for e.g. NAD27 vs WGS84), and the datum is supposed to be wgs84 or close equivalent-- datum seems to be hard-wired into these files, and by the normal processing, I don't get the option to change it.

I just tried the NASA ASTER dem, and the misalignment disappears -- but the data give a blobbier image, even though they are supposed to be 1x1 arcsecond^2 as well.


I believe this is the same as the well-known misalignment of SRTM data, for example, see the osm2garmin documentation (search for "SRTM contour shift correction").  I've never seen any real explanation of it, though, just speculation.

It's more or less standard when generating contours with Srtm2Osm to use something like " -corrxy 0.0005 0.0005", though my understanding is that the amount of error depends on the location.



Truly bizarre -- in the last section, I got a degree by degree block that spanned USA and Mexico. Contours were OK in USA, but were displaced over the border.  Two degrees over in Mexico, contours were fine.


You might want to post your question on - this is really right up their alley. Some amazing knowledgeable people read and post to the forum.