I recently acquired a recreation-grade GPS (Garmin eTrex 20). It allows waypoint averaging, which would improve the accuracy from the typical instantaneous accuracy of maybe 10 meters down to a smaller radius, maybe one or two meters (still trying to figure that out).
I also noticed that when various methods of converting lat-lon to Vermont State Plane (entering a custom UTM projection into the GPS device, DNRGPS, CORPCON) the results can differ by a little more than a meter. I understand a common rule of thumb is that for one source of error to be considered negligible compared to a larger error source, the lesser error should be no more than 1/3 of the larger error. So one would like a conversion workflow that is accurate to 0.3 meter if one wants to just ignore it. I'm wondering which of the multitude of epochs and ellipsoids should be selected at the various stages of conversion to assure the conversion error is less than 0.3 meter?
The default coordinate system is only described as "WGS 84".
Oh by the way, it really is being used for recreation. The Vermont Center for Geographic Information has shape files for many recreational features such as Wildlife Management Areas, published in Vermont State Plane coordinates, which can be viewed with their interactive map viewer.
> The default coordinate system is only described as "WGS 84".
NAD83, and, by extension, State Plane use the GRS80 ellipsoid. Using the WGS84 ellipsoid where the GRS80 is appropriate would cause a small difference.
> Oh by the way, it really is being used for recreation. The Vermont Center for Geographic Information has shape files for many recreational features such as Wildlife Management Areas, published in Vermont State Plane coordinates, which can be viewed with their interactive map viewer.
It is not likely that your recreational grade receiver is going to produce positions under any conditions that would render anything in the sub meter range significant. Nor is it likely that the GIS data is precise enough to render anything in the sub meter range significant.
Every GPS unit I know of is recreation grade until the data is post processed or processed in real time.
😉
While the NAD 83 and WGS 84 coordinate systems have a horizontal difference of approximately 1 m (depending on where you are in the U.S.). That difference is in the noise of a position determined with a recreational receiver. The GRS 80 and WGS 84 ellipsoids are identical in size and shape out to 2 in the 6th decimal place of the flattening, which is insignificant in the computation of a grid coordinate (either UTM or State Plane).
Fortunately, Vermont State Plane uses a Transverse Mercator projection which is what UTM uses. Changing the defining parameters will yield excellent results. Some states use the Lambert Conformal Conic and the panhandle of Alaska uses the Oblique Mercator; for those areas, Garmin receivers will not work - only Magellan will allow appropriate parameter input for State Plane coordinate display.
I believe Garmin has started to address this deficiency. I looked at a 78S in a store and I think I saw Lambert as an option in the user menu. Several older models I've looked at did not have it.
I've found that a popular application for transforming recreation (and mapping??) GPS waypoints, tracks, DNR GPS, is a bit limited. Also, as with any software, it pays to try out some scenarios with known data to figure out what is really going on.
When you download data from the GPS device to DNR GPS, it downloads the lat-long data and ignores any projection that might have been done by the device. It detects and displays the device brand and model, so you might imagine DNR GPS knows what datum the device uses for its lat-long. You would be wrong. The user selects a projection, such as NAT 83 (NSRS2007) datum projected into VT state plane coordinates within DNR GPS. DNR GPS silently assumes the lat-long received from the GPS device is in the datum chosen by the user, and has no built-in knowledge of which datum various GPS brands and models use (or at least, has knowledge of what datum the Garment eTrex 20 uses).
This means that unless the device can be set to export lat-long in an appropriate version of NAD 83, conversion by DNR GPS will be inaccurate by a few meters (nearly 3 meters for the NGS GPS mark I used as an example). If one wishes to keep conversion error negligible, one wouldn't want to use DNR GPS with a GPS device that uses the WGS-84 datum unless the error of the GPS readings was greater than 9 meters for my example mark. I imagine one could find other marks in the North American plate that are a little worse. So the conversion error is right on the edge of being in the noise.
Of course, gross mismatches between the different layers available from vcgi.org, such as roads, wildlife management areas, tax parcel data, and driveways indicate that the conversion errors really are in the noise for the purpose of figuring out where you are with respect to all (or maybe all but one) of these layers.