0

I am trying to extract data from gridded weather datasets such as GLDAS from pixels in which weather stations are located. I created an ArcGIS shapefile of weather station points. These weatherstation point coordinates were in GCS_North_American_1983.

My GLDAS raster data has a strange geographic coordinate system (see below). To properly extract the pixel values of my raster underneath the points in my shapefile I need to reproject the shapefile to be in the same GCS as the GLDAS rasters. When I attempt to project the weather station points to the GLDAS GCS I get the following error: "000151 : The geotransformation(s) is not valid."

I am using the Project tool in ArcMap. I select the input coordinate system to match the data to be reprojected (GCS_North_American_1983) and I select the output coordinate system to match my GLDAS raster. The spatial reference for the raster data is as follows:

GCS_Coordinate_System_imported_from_GRIB_file Authority: Custom

Angular Unit: Degree (0.0174532925199433) Prime Meridian: Greenwich (0.0) Datum: D_unknown Spheroid: Sphere Semimajor Axis: 6367470.0 Semiminor Axis: 6367470.0 Inverse Flattening: 0.0

I did not create that GCS, that's what the raster came with. Projecting to the above GCS fails yet I have been able to project the same weather station points to this subtly different output coordinate system from a similar gridded weather dataset called NLDAS:

GCS_Coordinate_System_imported_from_GRIB_file Authority: Custom

Angular Unit: Degree (0.0174532925199433) Prime Meridian: Greenwich (0.0) Datum: D_unknown Spheroid: Sphere Semimajor Axis: 6371200.0 Semiminor Axis: 6371200.0 Inverse Flattening: 0.0

I can't seem to figure out why I get an error for the first GCS projection and not the second.

  • Welcome to GISse! Some additional information would probably be helpful. Please tell us more about where the data is located spatially. Where is the output coordinate system coming from? It says custom, so is this something you determined from looking at another dataset? You say the weather station locations are xy NAD83. What does this mean? What projection are they actually in? Do you actually mean lat/lon NAD 83, or an actual projected coordinate system? The more detail you can provide about your data and your goal, the more likely you are to get a helpful answer. – Get Spatial Oct 31 '15 at 23:10
  • I wish the weather modelers would improve the models enough to use WGS84! /grumble_off – mkennedy Nov 1 '15 at 13:41
0

I don't know why the one works but not the other. Neither one should work.

You have two choices. One is to use the Create Custom Geographic Transformation tool. Set it up like the Project tool and import the output coordinate reference system (CRS) from the raster data. Because you're getting the same geographic CRS name from both datasets, make the transformation name reflect which sphere or dataset it should work with.

Try using the Null method. If there's a way to check it, and you don't think the data is overlaying correctly, try again using the geocentric translation method. In the first case, the longitude and latitude values are unchanged. In the second case, the latitude values are converted between NAD83 and the sphere.

Once you do that, you should be able to use it in the Project tool, or in ArcMap itself.

Alternatively, you could add the raster layer to a new ArcMap first so that it sets the data frame's coordinate system, then add the station layer. If you don't want to perform a transformation (AKA Null works), right-click the station layer and select data, export data. In that dialog, select to use the data frame's coordinate system.

  • I added the raster layer (set the data frame's coordinate system) then added the station layer. I then performed the null projection on the station layer and the points appeared exactly where they were originally. However, when I performed the reprojection using geocentric translation the points appeared very near the points produced by reprojecting the NLDAS coordinate system. I suspect the geocentric translation was the correct method in this case. – Philip Blankenau Nov 2 '15 at 1:21

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.