I have boundary data in WGS'84 lat/long and data in a custom generated CRS. The data in the generated CRS appears to be offset by a particular vector compared to WGS'84. I want to attempt to match the generated CRS projection boundary to the WGS'84 boundaries (seen overlayed on Landsat 7 imagery WGS'84 UTM zone 37N, 3 census datasets, yellow is WGS'84 and purple and red are both defined as the custom CRS on features)

3 census datasets, yellow is WGS'84 and purple and red are both defined as the custom CRS The custom CRS is listed here:

+proj=laea +lat_0=0 +lon_0=20 +x_0=0 +y_0=0 +a=6378249.145 +b=6356514.96582849       +towgs84=-160,-6,-302,0,0,0,0 +units=m +no_defs  

What datum is this?

  • Hi, I wonder how that custom CRS came into existance? Is it possible that QGIS just wasn't able to pick the right one when loading? In this case, reprojecting would not be the solution but maybe simply assigning the "correct" one ... – Bernd V. Mar 19 '15 at 11:06

The transformation parameters you're using are for Kenya and Tanzania. If the data is only one of the countries, perhaps try another transformation? The parameters are:

Kenya: -157,-2,-299,0,0,0,0 Tanzania: -175,-23,-303,0,0,0,0

There's also one for Burundi: -153,-5,-292,0,0,0,0

Neither transformation has very good accuracy, but might give a better a fit. If that doesn't work, you could try this. Break down the offset vector into x and y components and add them to the +x_0 and +y_0 parameters. Unproject the data to lat/lon, WGS84. Now see how it overlays. You may need to do some tweaking to get this method to work properly.

  • Thanks, that's right these are census sublocations for Kenya. Could you clarify "break down the offset vector into x and y components" ? – XNSTT Mar 30 '13 at 19:52
  • Tried the suggested transformation for Kenya but it does not shift the sublocations boundaries visibly at 1:164178 scale – XNSTT Mar 30 '13 at 20:06
  • even at 1:30000 scale no transformation visible – XNSTT Mar 30 '13 at 20:30
  • @XNSTT, Let's say you measure the offset and it's 5 meters to the northeast. Figure out two offsets instead: east and north. In this example, it might 3 meters east, and 4 meters north. – mkennedy Mar 30 '13 at 20:34
  • this works well for the sublocations near to the offset I measure but is less effective in all directions farther away. Would it be sensible to split up the sublocations? And if so by what magnitude in your opinion? – XNSTT Mar 30 '13 at 21:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.