arcmap circle draw usage error?

2948
9
03-05-2016 09:03 AM
BillNewman
New Contributor

In Arcmap...I create a circle at a particular point in dd.  Then I move to a specified radius (or even type it in). 

The result is a circle where the points on its circumference are not exactly in the right place (and its not just an

offset due to mouse errors)...Example (all in dd):

pick center point -121.400648, 38.300225.  Move out to radius of 0.21037.  You get for NESW points on the circle:

                    38.476373

     -121.626374              -121.175917

                    38.123647

N-S should equal .42074 but it is .353

E-W should equal .42015 but is .450

There might be some mouse errors, but this is way off.   What projection issue have I missed?  I'm pretty

sure the altitude doesn't vary more than 200' in this whole area.

Tags (1)
0 Kudos
9 Replies
curtvprice
MVP Esteemed Contributor

What's the data frame coordinate system set to?

Just because the locations report in degrees does not mean you are drawing the circle in degrees.

BillNewman
New Contributor

For this and the next rresponse:   From the data frame properties coordinate system I have...

WGS_1984_Web_Mercator_Auxiliary_Sphere

WKID: 3857 Authority: EPSG

Projection: Mercator_Auxiliary_Sphere

False_Easting: 0.0

False_Northing: 0.0

Central_Meridian: 0.0

Standard_Parallel_1: 0.0

Auxiliary_Sphere_Type: 0.0

Linear Unit: Meter (1.0)

So obviously I don't know what I am doing, but your response

"Just because the locations report in degrees does not mean you are drawing the circle in degrees." is what I feared the most!

Why doesn't the tool tell me I'm an idiot if I ask to just draw a circle? 

And yes, the line in the original post that read

E-W should equal .42015 but is .450

should read

E-W should equal .42074 but is .450

0 Kudos
DanPatterson_Retired
MVP Emeritus

you cant draw a circle in degrees using the same radius.  Unless you are projectin your data, a unit of a decimal degree has a different distance in the NS direction than in the E W.  you are probably drawing the semi-major and semi-minor axes of an ellipse.  if you need geodesic shapes use geodesic tool or use projected coordinates.  regardless, you will get erroneous values if your coordinate system is undefined

BillNewman
New Contributor

Okay, so you are saying that the MAS projection that I'm using (see previous post) will distort that "circle" that I drew with using the draw menu.

I'd expect some distortion, but over a 30 mile diameter circle in the CA central valley will I really see any deviation?  My expectation was that I wouldn't, but I get .350 NS to .450 EW   -   25%!   If I can justify this with some math that would be good.

I'll follow with what I am really trying to do which is gather statistics from a circle around a point on a relatively flat part of the world.  If you don't draw a circle, then could you point me in the right direction (so-to-speak!)?

It gets worse than this, I took that "circle", converted it into a feature, then published it and used it online to pull population attributes out of the 2015 tapestry.  That chain of commands worked to my surprize, although the circle came out as 25 mile radius, not 30...   To ESRI's credit, I may have missed some errors when converting this to a feature or publishing it so I could use it.  You might tell me if this is a totally wrong way to approach the problem.

0 Kudos
DanPatterson_Retired
MVP Emeritus

Did you see this help topic?  Creating a circle—Help | ArcGIS for Desktop

0 Kudos
DanPatterson_Retired
MVP Emeritus

cos(lat)*1 unit north south = 0.788  In short, the distance between two points on the EW axis is about cos(lat) the same spacing in the NS assuming you are doing something like 1degree EW vs 1 deg NS.  As you head towards the poles, the lines of longitude converge, at the equator cos(0) = 1.0 so the spacing of 1 degree of longitude is about the sameish as 1 degree of latitude.

BillNewman
New Contributor

Thank you!  This is easily demonstrated just by drawing a circle using the draw tool, and looking at its size when setting radius using a west or east extent, versus using a north or south extent....In the NS the circle is much larger...as you have pointed out (yes I'd have to say 21%!).  

This pushes me towards another question...Is there an arbitrary LCC projection option for the data frame, where I pick my lattitudes?  Why can't this be automatically calculated?

I can't mark this question done, since the "Creating a circle " pointer you gave, although the documentation is there, and looks like the right thing, I don't have such a toolbar (Production Feature Builder), nor can I find any relevant online instructions on how to enable it.... so close!  (I'm running 10.4 ArcMap...maybe it has recently been renamed?)

0 Kudos
DanPatterson_Retired
MVP Emeritus

The simplest thing to do is load a base map for your area, If it is in decicmal degrees, you could at least locate yourself.  then you could set the data frame projection to your UTM zone (there are online maps to find your utm zone given your longitude), then make your point on the map and produce your buffer.  Now, since you said feet... most projections use meters so there are 5280 ft in one mile (at least I think so), so you will have to do the math, although there are ways of setting alternate coordinates in the data frame (look under the data frame display properties) to match what you want.  Good luck

MelitaKennedy
Esri Notable Contributor

Bill,

The point you gave is in State Plane California III. There are both US survey feet and meters version of this coordinate system, which uses Lambert conformal conic.

Most map projections don't preserve distances, but using a coordinate system designed for a few counties should work pretty well.

State Plane zones were generally designed to have distortion due to the projection of no worse than 1 part in 10000.

Melita