Projection transformation problem (from GCS_WGS_1984 to WGS_1984_Albers)

[kimberleyPerry]

Hi,

I am wondering if someone can shed some light on a small problem.

I have got three sets of data that I need to use in a hydrological model and they need to be in the same projection. There is a DEM and soil grid file in Geographic co-ordinate GCS_WGS_1984 and land use data in WGS_1984_Albers. I thought it would be a good idea to use the transformation tool to make DEM and soil both WGS_1984_Albers. But now the maps do not line up- if I can call it that; for instance they do not overlap now in the model.

Thanks for any help,
Kim

[markdenil]

Your vocabulary is a bit confusing.

Coordinate transformations usually referes to transforming the underlying datum,
not reprojecting the coordinates.

In this case, your datums are the same (WGS 1984),
so no transformation should be required, just a projection.

Conceptually, features are projected, not from the earth, but from a generalized figure (a globe)
The datum defines the globe.
Using differently defined globes (datums) can lead to very different locations for features,
so transformation from one dataum to another can be significant.

It is true that a projection 'transforms' the globe onto a plane
(sometimes with intermediate steps such as a developable surface (a cylinder or cone)
or with rotaions or skews...), but we usually refer to it as projecting rather than transforming.

How far apart are the locations that should be coincident?
In what direction do they lay?
and most important:
What exactly did you do to each of the (geographic coordinate) data sets?
(tools and parameters, please)

[WhitynOwen]

This post deals with possible Transformation issues between GCS WGS84 and GCS NAD83.


I've been scratching my head at this for a few days and I now think there may be a bug at 10.2.1 (possibly 10.2 also) with transformations in general. Here are a few tests and my results.

Test 1:
Shapefile with a single point in GCS WGS84 (lower 48 US)
Add XY Coordinates tool to retrieve WGS84 coord values
Copy/Paste those values to new field to preserve them for comparison later

Project from GCS WGS84 to GCS NAD83 using ITRF00 transformation --> output to a shapefile
Add XY Coordinates tool to retrieve 'transformed' coords. (they match the original but they shouldn't?)
Copy/Paste those values to new fields for comparison

Project the other direction, i.e. GCS NAD83 --> GCS WGS84 using ITRF00
Add XY Coordinates tool to retrieve 'transformed' coords.
All three sets of coords match and they shouldn't


Test 2:
Again, Shapefile with a point in GCS WGS84 (lower 48 US)
Add XY Coordinates tool to retrieve coord values
Copy/Paste those values to a new field for comparison

Project to GCS NAD83 using ITRF00 transformation --> output to a file GDB
Add XY Coordinates tool to retrieve 'transformed' coords. (the coords appear to have been transformed correctly)
Rename those field headers for comparison

Project the other direction, i.e. GCS NAD83 --> GCS WGS84 using ITRF00 and output to a shapefile
Add XY Coordinates tool to retrieve values (Coords don't appear to be transformed; coords match the NAD83 values in the GDB not the orig)

Can anyone else repeat these results? I need to run more tests to see if it is just the output to a shape or if there are other issues as well. Also, the pop-up warning when adding data in a different Datum than the data frame projection seems to be inconsistent, i.e. sometimes I get it sometimes I don't.

Thanks.

[kimberleyPerry]

Hi,

Sorry about confusing vocab- I have little experience in GIS. Can I give a sort of step-by-step account of what I did with my data, and if you can help that would be great:) I would not be surprised if it ends up being something really simple Ive done wrong.

I have a land use file that is Datum: D_WGS_1984. Spatial reference WGS_1984_Albers. But this did not have an attribute table (which I needed) so I used a tool 'copy raster in data management tools' and I specified a different pixel type (from 32 bit signed integer to 32 bit signed) which worked and gave me an attribute table for my data. I dont know if this may have now altered my data maybe? But the copied data still has the same spatial reference.

The dem data that I am using is datum D_WGS_1984 and spatial reference GCS_WGS_1984.

These files as they are, are fine and all overlap. But I need to have them all in the same projected coordinate system. So I used a tool in data management to 'define the projection' of the DEM data to WGS_1984_Albers to be the same as the land cover.

When I do this however, the two sets of data now do not overlap. And when I look at the two sets- I see that their 'extents' are different? Perhaps that is the problem?

[kimberleyPerry]

Hi,

It now seems to be working.  I went to ArcCatalog and used a different tool- still in data management but rather: Projections and Transformations>>Raster>>Project Raster. This was instead of the tool Projections and Transformations>> Define Projection. And I changed both data files to a completely different projection (Africa_Albers_Equal_Area_Conic). Now all files are seem to be projected and all overlap:)

[WhitynOwen]

The varying results of "Tests 1 and 2" I posted above have been traced to the source .prj file format. However, I'm still getting consistent results where the transformation is not applied when converting from GCS WGS84 to GCS NAD83 using the default ITRF00 transformation in 10.2.1.

A short test has been described in a new post "Transformation issues between GCS WGS84 and GCS NAD83 using 10.2.1"

I'd appreciate it if anyone can duplicate the results or help me identify my error.

Thanks!