2

I am working in ArgGIS 10.1 with a field (raster layer in tif format) in which I have placed two points (.shp layer).

I have two versions of both layers, resulting four layers: one tif+shp set in CRS WGS84 Web Mercator and the other tif+shp set in CRS WGS84 UTM zone 15N.

If I establish the CRS of the Data Frame to WGS84 Web Mercator, the distance between the two points is 138 meters, but if I establish the CRS of the Data Frame to WGS84 UTM zone 15N the distance is 99 meters, which is a huge difference. However, the point layers are displayed absolutely one above the other, I mean: the two points in one CRS haven’t any displacement in relation to the other two points; and the same for the rasters.

  1. Why don’t I see both pairs of points misplaced? Is that “on the fly” projection by ArcGIS? If it’s so, does it mean that ArcGIS projects “on the fly” even between data with CRSs of different geographic coordinate system? (as far as I am concerned, ArcGIS only projects “on the fly” between data with CRSs with the same geographic coordinate system but different projection).

NOTE: Both projected CRSs (WGS84 Web Mercator & WGS84 UTM zone 15N) correspond to these two different geographic coordinate systems: in ArcGIS, Geographic Coordinate System > Spheroid-based > WGS 1984 Major Auxiliary Sphere & Geographic Coordinate System > World > WGS 1984, respectively.

  1. Another major doubt is which of the two CRSs is then better for working in the south of Minnesota State (US)?
Improve this question
3
  • 4
    Web Mercator projection does not suit at all for distance measurements en.wikipedia.org/wiki/Mercator_projection.
    – user30184
    Aug 15, 2014 at 14:01
  • Agreed with the above. If at all possible, use UTM 15N, which for a local focus in Minnesota is far better for distances and shapes.
    – Kevin R Dyke
    Aug 15, 2014 at 16:41
  • 2
    Well, you should use WGS 1984 Web Mercator (Auxiliary Sphere) instead which would put the data on WGS84. However, due to a quirk in how WGS 1984 Web Mercator is used, there's no geographic/datum transformation needed between it (WGS 1984 Major Auxiliary Sphere and WGS 1984. "On-the-fly" will convert between data that's using different GeoCRS but setting a transformation may be needed.
    – mkennedy
    Aug 15, 2014 at 17:07

1 Answer 1

Reset to default
2

1) Your "As far as I'm concerned" quote is incorrect. Since as early as 9.2 it has not been this way. See the help menu and see the quote below from ESRI help. I think the better way to see it is, if ArcGIS can project on the fly then it will.

2) Use UTM.

"If I establish the CRS of the Data Frame to WGS84 Web Mercator, the distance between the two points is 138 meters, but if I establish the CRS of the Data Frame to WGS84 UTM zone 15N the distance is 99 meters, which is a huge difference. However, the point layers are displayed absolutely one above the other, I mean: the two points in one CRS haven’t any displacement in relation to the other two points; and the same for the rasters."

Well this should give you confidence not worry. If I had to guess the WGS distance error it would be about 30% - 40% so I would expect Web Mercator to approximate this. Use this calculator to enter your point locations and I'm sure you find the answer to be around 100 meters.

The first layer added defines the data frame’s coordinate system. This is true whether the data is projected or geographic. For example, if the first layer added contains a Lambert Conformal Conic projected coordinate system, all other layers will project on the fly to match this. Similarly, if the first layer added to the data frame contains data that uses a WGS84 geographic coordinate system, all other layers will adjust to match this. Even data that uses a projected coordinate system will (sic) project on the fly.

Improve this answer

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