topic Define projection changes position of raster layer in Data Management Questions

Define projection changes position of raster layer

DitlevReventlow — Sun, 05 Jul 2015 14:16:02 GMT

Hey Everyone.

I use define projection on a raster layer that has the spatial reference system User_defined equidistant_Cylindrical and the datum user_defined, and change it into either equidistant_culindrical (Sphere) or equidistant cylindrical (world).

I try to do it into both because I am not sure which of them is the correct one. My thought was then to convert into WGS_1984 afterwards and see which fits best.

However, when I do the define projection, the whole raster drastically changes position. How can this be? Isn't define projection just used to define the projection system, not actually changing the position or drawing. This is at least how it has normally worked for me.

The reason why I am doing this is because I want the data in WGS_1984. Any help on how I could do this is also needed.

I would very much appreciate any help. Thank you very much!

Best regards

Ditlev Reventlow

Re: Define projection changes position of raster layer

DanPatterson_Retired — Sun, 05 Jul 2015 15:11:38 GMT

Define Projection is only used when a file has an unknown projection or...as in your case...when it has been define incorrectly and has to be reverted back to the original projection

Re: Define projection changes position of raster layer

WesMiller — Sun, 05 Jul 2015 15:52:06 GMT

I didn't think define projection would move anything either but the project tool would if you had a projection assigned. Did you accidentally use the wrong one, maybe?

Re: Define projection changes position of raster layer

XanderBakker — Sat, 11 Dec 2021 22:12:13 GMT

In addition to what Dan Patterson explained... the projection is used to interpret the coordinates of the dataset and define the location of the data on the earth. If you look closer to the variables that define the projection you will notice that the falseOriginAndUnits is different for both projections:

Sphere_Equidistant_Cylindrical	World_Equidistant_Cylindrical
-10007700 -10007700 10000	-10018900 -10018900 10000

This would mean a shift of 11200 meter in both X and Y

Python snippet:

import arcpy
sr1 = arcpy.SpatialReference(53002) # Sphere_Equidistant_Cylindrical
sr2 = arcpy.SpatialReference(54002) # World_Equidistant_Cylindrical
print sr1.name, "\t", sr2.name
print sr1.falseOriginAndUnits, "\t", sr2.falseOriginAndUnits

Re: Define projection changes position of raster layer

DitlevReventlow — Sun, 05 Jul 2015 16:22:28 GMT

As can be seen , it changes a lot but the two defined rasters are quite similar (not the same but quite similar)

Re: Define projection changes position of raster layer

DitlevReventlow — Sun, 05 Jul 2015 16:26:25 GMT

Dan, I am not sure what you mean by defined incorrecty. I mean I cant know whether it is defined incorrectly just because it is user defined?

Wes, As can be seen from the drawing both of them changes the position a lot.

Xander, the difference in my case is only about 10 km difference.

Re: Define projection changes position of raster layer

DanPatterson_Retired — Sun, 05 Jul 2015 18:08:38 GMT

Your first sentence...you applied a projection to a data set to see which one was best or worked best. Get the projection information from the source if it is known. If it is a matter that only one of the two is correct, you will quickly find out...as long as you have ruled out the possibility that neither are correct.

Re: Define projection changes position of raster layer

DitlevReventlow — Mon, 06 Jul 2015 08:33:43 GMT

Dan,

Thank you for this. The thing is that both seems to be from. When I convert both the defined rasters to WGS 1984 with "project raster" , the almost dont change - they are still as heavily distorted. Aboth 10 km different as mentioned above, but both many 1000 km off other WGS 1984 rasters.

Re: Define projection changes position of raster layer

DanPatterson_Retired — Mon, 06 Jul 2015 09:11:27 GMT

Is there a chance that you are doing these operations in the same data frame as other data? I would suggest you create two new data frames and place each file into its own data frame with no other data. Then ... for each file, in its own data frame...you right-click on the layer, go to properties and determine the extent values. These will not change regardless of what you have done to the file in terms of defining a projection. Do the extents look reasonable? A file with a geographic coordinate system ( like a GCS WGS84) will only have values in the range -180 to 180 EW and -90 to 90 NS. If they are big numbers then you have projected data and the range in values will give you what type of projection it was or should be in. Ideally, both files should have the same extent if indeed they covered the same extent. If they have the same extent and they are defined differently, then one or both have been improperly defined when using the Define Projection Tool.

Now on to projection... When you use the Project Tool, always put the result into its own dataframe since a file that is projected will project-on-the-fly to try and match the projection of the dataframe...and as you have noted, when the file was defined wrong, then projected, it flies off to somewhere you don't want it to be.

The whole process of adding data to a data frame since ArcMape is to try to "be helpful"... In the "old days" you would get the warning that things didn't match and sure enough, the files would fly off into their corners spatially separated because the projection files were wrong, undefined or mismatch. In order to fix this in the "helpful" environment, you need to examine their properties...understand their possible extent values for a given coordinate system/projection and then act accordingly.

So in summary, I have no clue what the real coordinate systems of the input files were, but IF they were defined by some other source and they were redefined correctly...set them back and ensure that they are correct. Once they are in a know, verified coordinate system...you can proceed to Projecting the data to a different coordinate system. Then...and only then...do I introduce other data into the dataframe. If all the files play nice...then they should overlap or abut perfectly.

Re: Define projection changes position of raster layer

DitlevReventlow — Mon, 06 Jul 2015 14:27:07 GMT

I get this error message I have attached to this message when I try to add either of the defined projections to a newly created data frame. I have shoen the extent of the two defined layers as well, As can be seen they are the same.

The extent values are far higher than they should be for GCS. So I guess that they must be projectied data. But this also fits with what they have been defined as since both World_equidistant_cylindrical and sphere_erquidistant_cylindrical are projected coordinate systems.

I am confused about what you mean when you say that the extent will not change when you define a new projection. Because at the same time you say that if the two layers were defined different (which they are ) and have the same extent then one or both have been improperly defined.

This to me sounds reverse.

Re: Define projection changes position of raster layer

DanPatterson_Retired — Mon, 06 Jul 2015 14:32:48 GMT

I will try again...

You do not define the projection of the new data frame...you just add one file to it...without specifying anything. If you get an error message...the file has been defined wrong.

If you define the projection of the data frame THEN add data...you are defeating the whole purpose of what I am trying to get you to do. So I will defer to the guidance of others to try another approach to solve your problem.

At some time, everything will line up and you will have identified the problem...please report in detail what was done to rectify it so that others don't go down the path again.

Good luck

Re: Define projection changes position of raster layer

DitlevReventlow — Mon, 06 Jul 2015 15:12:51 GMT

Ok i understand. But I have done as you said, I havent altered the data frame properties, I have only defined the projection for the layers that I have added to them. And since I get an error in each of the cases where I add one of the two defined layers, I guess that both the defined projections are wrong.

I will make sure to make a description when/if I solve it.

Re: Define projection changes position of raster layer

ChrisDonohue__GISP — Mon, 06 Jul 2015 15:16:09 GMT

Let me see if I can help clarify part of this, as it is not obvious. ArcMap takes the first data set added to it and inherits that data sets projection/coordinate system for the Data Frame in ArcMap. For example, if I open a new MXD and then add a shapefile that is in "UTM 10N NAD 83" in ArcMap, the Data Frame will now be automatically set to " UTM 10N NAD83".

When you are checking data to figure out what its coordinate system should be, you want to add it to a new ArcMap mxd that has not had any data added previously or had a projection/coordinate system defined. This way you can get a real sense of the data's location. If, however, one adds it to a MXD where it has already inherited or been assigned a projection/coordinate system, ArcMaps "project-on-the-fly" will be in operation, and the results will be not representative.

Chris Donohue, GISP

Re: Define projection changes position of raster layer

DitlevReventlow — Mon, 06 Jul 2015 16:36:15 GMT

Okay, in my case when I add the raster which I have defined the projection for, to a new arcmap file, I get the error message listed above, and as Dan said this means that the wrong projection has been attributed to the data.

So I don't know how to move from this. As mentioned, the original raster file had the following reference system User_defined equidistant_Cylindrical and the datum user_defined.

The only two I could think of that I could be would be equidistant_culindrical (Sphere) or equidistant cylindrical (world). And since both of them apparent are wrong (since I get the error message when I open them in a new mxd file).

I guess that the only thing I can know for sure is it is a projected coordinate system(because of the extent). But that is the only clue I have.

Re: Define projection changes position of raster layer

MelitaKennedy — Mon, 06 Jul 2015 20:30:53 GMT

Ditlev,

Can you post the complete, original definition? Please make sure it includes the spheroid (ellipsoid) information as well.

I think the projection parameters may be different. The "Sphere" and "World" definitions are generic and using zero for the central meridian and 60 for the standard parallel.

What happens if you just overlay this data (with its original coordinate system) with a base map or other reference map? Does it line up or not?

Melita

Re: Define projection changes position of raster layer

DitlevReventlow — Tue, 07 Jul 2015 08:03:13 GMT

Melita,

The original raster definition is as follows:

User_Defined_Equidistant_Cylindrical

Authority: Custom

Projection: Plate_Carree

False_Easting: 0,0

False_Northing: 0,0

Central_Meridian: 0,0

Linear Unit: Meter (1,0)

Geographic Coordinate System: GCS_User_Defined

Angular Unit: Degree (0,0174532925199433)

Prime Meridian: Greenwich (0,0)

Datum: D_User_Defined

Spheroid: User_Defined_Spheroid

Semimajor Axis: 6378137,0

Semiminor Axis: 6378137,0

Inverse Flattening: 0,0

I create a mxd file with three data frames, one with the original raster and one with each of the newly defined raster. Then I add a basemap to each of these. This results in that for the original raster the dataset fits with the basemap. For the two where I have used define projection they don't align with the basemap, and they are very much off.

Re: Define projection changes position of raster layer

MelitaKennedy — Tue, 07 Jul 2015 22:25:15 GMT

Hi Ditlev,

So that was the issue. When you used the Define Projection, you changed the metadata, but not the actual coordinates of the raster, so it no longer reprojected correctly. You should try the Project Raster tool. Weirdly enough, the GCS/datum/spheroid/ellipsoid is a sphere with the same value as WGS84/GRS80 semimajor axis.

Hmmm. Actually, it may be easier to reproject the raster in ArcMap. Add it to ArcMap and set the data frame's coordinate system to what you want the new raster to use. Right-click the raster's name in the table of contents and select data, export data. In the tool, select the data frame's coordinate system. This will mean you won't have to create a custom geographic (datum) transformation to convert from the "GCS_User_Defined" to WGS 1984. I don't think it's needed.

Melita

Re: Define projection changes position of raster layer

DitlevReventlow — Tue, 07 Jul 2015 23:00:53 GMT

Thank you for this. I converted the coordinate system used by the dataframe to WGS_1984. This worked but it didn't shift any data around.

Then I went into data--> export data --> and changed the spatial reference from raster layer to dataframe and save. This worked in the sense that if I right click on the raster layer and select properties --> source, then the reference system states that it uses wgs_1984.

However, the file remains in the same position as the original raster. It didn't change around. Also If I look in data --> export data on the output raster, the spatial reference box still has raster dataset selected (and not data frame) although it is obvious that things have been changed around in the rest of the "export data menu".

What do you think about all of this? Do you think that it could be correct that the position of the cells are indeed the same in the original "defined" spatial reference system and WGS 1984?

Re: Define projection changes position of raster layer

MelitaKennedy — Wed, 08 Jul 2015 21:45:28 GMT

What does the raster's properties say about its extents? If you add the new raster into ArcMap, ArcMap will reproject all layers to the data frame's coordinate system--so the two rasters should still line up. Is that what you're seeing?

Re: Define projection changes position of raster layer

DitlevReventlow — Wed, 08 Jul 2015 22:06:27 GMT

The two raster layers have complelety different visible extents. But they have the same full extent.

Yes the two raster are aligned when I open them in the same dataframe.