[Xastir] UTM vs lat/long, GIS/GPS

[Gerry Creager n5jxs]

Seems like I did this in geodesy 101 a while back.  SOmeo were obviously 
asleep.



Wes Johnston wrote:
> About a year ago, I did a comparison between NAD27 and WGS84... Mind you, I'm
> certianly not on a par with Gerry, but I did stay in a holiday inn express last
> night!!...

So, don't try brain surgery.  There's no correlation with that and 
Holliday Inn Expresses...

> I found that in my part of SC, that the shift from NAD27 to WGS84 was about
> 130'.

OK.  Not unreasonable.  call it 40mCEP.  That's ok.

> Now, two weeks ago, I had a discussion on the phone with our local GIS folks and
> they told me all about state plane.  State plane is just like UTM, but uses feet
> instead of meters.  Just for trivia, SC is one of 5 states that uses the
> "international foot" which is .3048 meters instead of 0.3048006096 meters.  At
> 2million feet (the width of SC) it makes 4' difference.

No.  The GIS geeks either oversimplified, or are wrong... or may be right.

State Plane systems were set up in all the states by National Geodetic 
Survey to provide plane systems for land surveyors to work from, since 
they're easier for 2d coordinate geometry than a curvilinear surface, 
and using spherical or nonlinear geometry.  Use of the international 
foot was a State of SC issue, not an NGS issue which adopted the US foot 
and eventually, the meter.

NOW: It's possible that the officially recognized projection for SC 
State Plane is UTM, International Foot, NAD83, or  some variation thereof.

UTM can be expressed in meters, the norm, feet, furhlongs, or parsecs 
although the math gets a little hairier with the last 2.

> In my search for details on international foot, I came upon
> http://www.ngs.noaa.gov/TOOLS/Nadcon/Nadcon.html .  This shows the distortions
> between NAD27 and NAD83.  I'm sure such a chart exists for NAD27 to WGS84.

This is *representative* of the shift.

NAD27 is classified as a *subjective* datum; NAD83 is an objective 
datum.Subjective datums mean they do not necessarily have a basisi in 
careful measurement... or reality.  Objective datums are based in good 
measurement, error mitigation by repeated observation, rigorous 
least-squares adjustment, and error analysis.

It's possible to move from one datum to another, but moving from a 
subjective datum to an objective datum means you have to do so 
referenced to a particular point.  The US has identified these points, 
thru the NGS.

It's not "chartable" save in very broad terms.

> Previously, I had believed the shift to be a pretty linear shift (if you are at
> point X,Y, just add 130 feet north east), but if you are point x1,y1, the shift
> is 125 feet, and so on....  But it's not.  So the point to all of this is that
> you can't count on a simple "slide the map a little to the left" conversion
> from one datum to another.

That's right.  A 130 foot delta in one part of the country could easily 
be a 22' delta in another... or an error of less than 20' in another. 
It varies from place to place.  The area east of the Mississippi is in 
better shape than the area near the Pacific NW.

> Now, UTM...  every document I've read about Military grid and UTM is quite clear
> about disortions at the corners.  That's why UTM uses zones, and all of those
> zones overlap.  You are supposed to use the zone who's center you are closest
> to.

Correct

> Now to answer the original question... the arguement they pose is like saying
> you get less accurate measurements in meters than feet.  Units are Units are
> Units... doesn't matter if you _display_ numbers in UTM or lat/lon, or feet or
> meters.

Correct.

> WGS84 and NAD27 are models of the earth's spherical shape.  90degrees around a
> circle will give a different measure of distance than 90 degrees around an
> ellipse, even though they may have the same total circumference.

NAD27 uses the Clarke ellipsoid of 1866 (I *THINK*; I'm 'way minus on 
sleep and not sure what time zone I'm either in, was was last in). 
WGS84 uses the WGS84 ellipsoid, which is very similar to the Geodetic 
Reference System ellipsoid of 1980.  The models are different by some 
10s of meters, but the surface difference isn't too great.

NAD83 also uses GRS80 as an ellipsoid.

> I think the guy that wrote that article pulled the old double switch-a-roo... if
> you compare apples and apples (UTM and lat/lon both using WGS84), you'll find
> they match very closely, until you get to the edges of the UTM zone.  If you
> compare a given postion in UTM or lat/lon to WGS84 and NAD27, you'll find they
> are both off by the same number of meters - in my case , 130 feet).

UTM is a stinking projection, with most distortion at the corners. 
Lat/Lon is an unprojected representation, with distortion nonuniform 
throughout, but most pronounced at the linear edges.  The projection you 
select is supposed to be based on one that will minimized the overall 
distortion for your area of interest.


> I think the fundamental confusion here is that most paper maps are in NAD27.  So
> the author's only source of data for UTM was a paper map.  When he compared that
> map to his GPS using WGS84, he switched to lat/lon at the same time... he
> probably didn't know he could change his GPS to read UTM.

Plausible.

> Further, I know I can change the eillipseoid model in my GPS independently of
> the units that are displayed.

Usually correct.  There are some "low cost" exceptions.

> Quoting "Curt, WE7U" <archer at eskimo.com>:
> 
> 
>>I'm having a discussion over on the sarcomm list and haven't found
>>any GIS-literate people to get some questions answered in order to
>>back up my points.  Since this is probably of interest here as well,
>>I'll see if Gerry Creager or other GIS-literate person might see it
>>here.
>>
>>Start with this:
>>
>>http://www.uaex.edu/Other_Areas/publications/HTML/FSA-1032.asp
>>
>>In that document they say this:
>>
>>
>>  "How Much Error? For most receivers the default datum option for
>>  screen display probably will be WGS  84 or NAD  83. If the paper map
>>  datum is NAD  27, then NAD  27 should be selected from the GPS
>>  receivers option list. The position errors caused by choosing the
>>  wrong datum can be very serious for differentially corrected
>>  position data. How large the errors are depends on whether Lat/Long
>>  or UTM coordinates are being used for the GPS/map work. The errors
>>  may not be important for travel or recreation purposes, where the
>>  receivers screen-displayed maps are being used, or where precisely
>>  computed coordinates are not extremely critical.
>>
>>   For lat/long, the position discrepancy will be less than 100 feet.
>>   For UTM coordinates, the discrepancy will be more than 650 feet."

Horse Hockey.  He can't convert between feet <-> meters.  He's confused 
and attempting to propagate it to the masses.

>>I say that's BS.  Unfortunately some SAR guys on that list have
>>latched onto this as gospel and think that they get large errors
>>with UTM.
>>
>>The fact that tick marks near the corners of the topo maps don't
>>line up with the lat/lon might tend to make them think they have
>>reinforcement for their view.
>>
>>My points:
>>
>>*) They're confusing datum-shift errors with UTM/Lat-Long.  They are
>>totally and entirely independent.  UTM does not give you ANY error
>>as compared with lat/long.

Correct.

>>*) The above document is just plain wrong.  Wrong wrong wrong!

You are Right right right!

>>*) Some of the SAR guys are claiming errors of "two ridges over from
>>where they were supposed to be" or "up to 1/2 mile error when using
>>UTM".  They most likely had an incorrect datum for their paper or
>>electronics maps plus might have had lat/long format problems
>>compounding it.  You can be off by quite a bit if you think you're
>>given coordinates in one lat/long format but they're really in
>>another.

10s to 100s of meters... datum error is the most common cause of blunder 
in adding data to maps using various forms of geospatial data collection.

>>*) UTM gives you meter resolution in the numbers.  Nice.  Easy to
>>get to 10 meters or so plotting on a map without extra tools.

Correct

>>*) UTM gives you the ability to calculate distances in your head, A
>>squared plus B squared = C squared.

Correct.

>>Can anyone shed some experienced light on this and help me get these
>>guys on the right track?

Feel free to post something, quoting me if required.  I've gotta get 
some sleep before I tackle something like this with a response to the 
group in general and make it look authoritative.  Maybe later today, 
maybe tomorrow...  Have to see.

73, gerry
-- 
Gerry Creager -- gerry.creager at tamu.edu
Texas Mesonet -- AATLT, Texas A&M University	
Cell: 979.229.5301 Office: 979.458.4020
FAX:  979.847.8578 Pager:  979.228.0173
Office: 903A Eller Bldg, TAMU, College Station, TX 77843