GPS confusion

37 46' 15.5" N 122 24' 41.9" W +50'
Since the site went up, the DNA Lounge Directions page has had a little map on it, with the ICBM coordinates of the club below it. I got these numbers by digging them out of the URL of some mapping site or other (I no longer remember which.) Well, the other day I got mail from someone who said that he had punched my coordinates into his GPS box and:

Bart Smit wrote:
I discovered the following:

As no datum (geodetic reference system) is given, one is tempted to interpret these coordinates as WGS84 (the standard datum for gps). However, this would let the DNA Lounge sit on Harrisson towards 10th, which is more than 90 meters off.

If I assume that the coordinates are referred to the North American Datum of 1927 (NAD27), which has been the standard datum for USGS maps for a long time, and then plot the position of the DNA lounge on a map, the result looks much more plausible. See attached image, which is a small section of a scan of the SF North Quadrangle map from USGS's 7.5 minute series. The circles represent the margin of error of a GPS receiver.

Perhaps you could clarify on the web site that the given coordinates are NAD27 (are they?), or else use the WGS84 coordinates, 37 46.252' N, 122 24.761' W

A ``geoditic reference system'' is apparently a way of mapping the spherical coordinates (latitude and longitude) to points on the surface of the earth: this is not a 1:1 mapping since the earth is not a sphere. It seems there are a number of different ``geoids'' used for doing this, and different maps use different ones.

So I'm trying to figure out what the ``right'' ICBM address for the club is. GeoURL lists a few ways to find your coordinates. You can find yourself in Yahoo Maps or Tiger Maps and dig the lat/long out of the resulting URL; or you can go to Poskanzer's super cool ACME Mapper which uses the TerraServer data, and lets you type in lat/long, among other things.

I found DNA's coordinates in various places, plus netik brought his GPS down and got numbers from that. These numbers are (well, I want to say ``all over the map,'', ha ha, but actually are just) slightly off.

  • When looking at these numbers in the ACME Mapper, Yahoo and Tiger are in agreement with ACME, so presumably they all use the same system. Is this NAD27?
  • My old numbers and netik's GPS numbers are pretty close to each other, and both put show up in the middle of the CostCo parking lot down the street.

  • Bart's numbers, which he says are WGS84, show up basically next door (inside the glass block place.)

I am confused about A) what the ``right'' reference system to use is, and B) about what reference system is being used by the various mapping sites. Is it the case that the mapping sites use a different reference system than GPS receivers use? That would be pretty silly, if true...

  Source:   Lat (dec):   Long (dec):   Lat (deg):   Long (deg):   Via ACME:
Bart 37.7709 -122.4127 37 46' 15.24" -122 24' 45.72" Glass Block
Netik 37.771008 -122.411750 37 46' 15.629" -122 24' 42.3" CostCo
jwz 37.77097 -122.41164 37 46' 15.5" -122 24' 41.9" CostCo
Yahoo 37.770960 -122.412794 37 46' 15.456" -122 24' 46.058" DNA
Tiger 37.7710686 -122.4127808 37 46' 15.847" -122 24' 46.011" DNA
ACME 37.771019 -122.41273 37 46' 15.668" -122 24' 45.828" DNA

(See also RFC 1876, where you can include your lat/long in DNS, to make tools like XTraceRoute work better!)



Dan Egnor wrote:

TIGER (since 1995) uses NAD83, not NAD27 (except for Hawaii and Pacific islands). Many commercial mapping data suppliers (such as those which the likes of Yahoo Maps use) start with TIGER data. The Census Bureau point out that the "likely coordinate shift between NAD83 and WGS84 is much smaller than the nominal accuracy of the TIGER/Line files".

Based on my experience TIGER/Line and commercial mapping engines intended for casual use (as opposed to professional GIS, or USGS maps) are simply not very accurate -- I mean, why bother sweating the small stuff as long as the map looks OK? I suspect that's the cost of most of the variance you're seeing, not datum mismatches.

So this means that the truthful answer would be to use the coordinates we got from the GPS, which are in WGS84, even though those coordinates show us to be almost 90M east when fed into the ACME, Yahoo, and Tiger sites.

Kinda lame!

Tags: , , , , ,

6 Responses:

  1. ch says:

    GPS receivers let you set the datum to the map you are using. Check to see how the unit is set.

  2. ciphergoth says:

    WGS84 is probably the right choice - if I see latitude, longitude and altitude without a datum specified, I think I should be able to assume WGS84. However, it'll probably say your club is below ground, or far above ground, or something.

  3. naturalborn says:

    Yet another reason to pave the earth, to make it a nice perfect sphere.

  4. susano_otter says:

    At least there's a good chance the ICBMs will miss the club by 90 meters or so.

    Which is why I plan to use a Thomas Guide and a suitcase nuke the next time I visit.

  5. netik says:

    Just to let you know, I just took my GPS out of the car and it's set to:

    Location Format: hdddd (degree) mm.mmmm'
    Datum: WGS 84
    North Reference: TRUE

  6. fo0bar says:

    $ host -t loc is an alias for LOC 39 28 5.952 N 119 48 50.580 W -1427.68m 10m 10000m 10m

    I need to dig out my GPS unit and take it to work for similar reasons. And yeah, my GPS's output agrees with yahoo and tiger, but terraserver is off.