when does projection matter?

612
5
Jump to solution
01-15-2014 10:34 PM
Ianmiller
New Contributor III
Hello all, this question may seem rudimentary, but Its been bugging me and I want to confirm.

If say I have 2 polygon layers, zipcode and city boundary for example, and I want to clip the zipcodes using my city boundary, does it matter what projection I do it in (assuming both layers have the same projection)?  I.e. if Zipcode X is inside my city boundary using web mercator, it will still be inside my city boundary if I switch both layers to any other projection?. The projection will only change physical measurements such as distances and areas but not the relative relationship between my polygons.

thank you all for the help
0 Kudos
1 Solution

Accepted Solutions
markdenil
Occasional Contributor III
Given common projections for both data sets
Only under specific circumstances will a particular projection choice be significant.

For example: if your area of interest spans a projection interuption or the 'edge' of the projection
(say, if the polygons run across the 180° longitude, a cylindrical projection interupted
on that line would not be appropriate)

Somewhat less obviously, if your polygons have very long undensified sides
(for example, a rectangle with nodes at the corners and no intermediate verticies)
it may not project faithfully
The corners will be in the right place but the sides may take on paths from corner to corner
that do not represent the original polygon's ground footprint.
This is especially noticable in high latitudes with (as mentioned) long undensified polygon sides.

In any event, these concerns are unlikley to be issues for anyplace with a zipcode
using any projection appropriate to that area.
Neither zipcode zones nor cities usually have the very long sides I have identified as problematic.

View solution in original post

5 Replies
RobertBorchert
Frequent Contributor III
Projection always matters.

Generally your best bet it to have all features in the same projection.   I like to use personal geodatabases a lot as you can import features into a common feature dataset and have them converted to the same projection.

However, sometimes the best projection for one feature is not the same as for another.

If both features are in the same projection then do it in their projection.

QUOTE=ian31590;358548]Hello all, this question may seem rudimentary, but Its been bugging me and I want to confirm.

If say I have 2 polygon layers, zipcode and city boundary for example, and I want to clip the zipcodes using my city boundary, does it matter what projection I do it in (assuming both layers have the same projection)?  I.e. if Zipcode X is inside my city boundary using web mercator, it will still be inside my city boundary if I switch both layers to any other projection?. The projection will only change physical measurements such as distances and areas but not the relative relationship between my polygons.

thank you all for the help
0 Kudos
Ianmiller
New Contributor III
So my end goal is to get a list of all zipcodes within my city boundary. What I am asking is, both layers start with the same projection, it shouldn't matter what that specific projection is, the list of zipcodes that fall within the boundary will be the same every time. If I do the clip with both layers in web mercator and then I do the clip with both layers in UTM zone, zipcode X, Y, and Z will always be within the boundary. Projection will stretch, shrink, and rotate my polygons depenidng on what I use, but a point within a polygon will always be within that polygon assuming that point and polygon start out with the same projection.
0 Kudos
markdenil
Occasional Contributor III
Given common projections for both data sets
Only under specific circumstances will a particular projection choice be significant.

For example: if your area of interest spans a projection interuption or the 'edge' of the projection
(say, if the polygons run across the 180° longitude, a cylindrical projection interupted
on that line would not be appropriate)

Somewhat less obviously, if your polygons have very long undensified sides
(for example, a rectangle with nodes at the corners and no intermediate verticies)
it may not project faithfully
The corners will be in the right place but the sides may take on paths from corner to corner
that do not represent the original polygon's ground footprint.
This is especially noticable in high latitudes with (as mentioned) long undensified polygon sides.

In any event, these concerns are unlikley to be issues for anyplace with a zipcode
using any projection appropriate to that area.
Neither zipcode zones nor cities usually have the very long sides I have identified as problematic.
JoeBorgione
MVP Emeritus
Take a look here to get a background:

http://resources.esri.com/help/9.3/arcgisengine/dotnet/89b720a5-7339-44b0-8b58-0f5bf2843393.htm

Here is a link to what I consider the guide to understanding projections.

http://kartoweb.itc.nl/geometrics/map%20projections/understanding%20map%20projections.pdf
That should just about do it....
0 Kudos
Ianmiller
New Contributor III
Thank you all for the responses and linked resources.
0 Kudos