Improving GPS Accuracy When Recording Locations

Considering the inherent inaccuracy in most consumer grade gps units (~15 meters), I wouldn't waste time taking three readings.

I use both my old Garmin Etrex Legend and more recently I use my Nuvi 1350. If you simply leave your gps in one place, untouched, for 10 minutes or so you can be relatively certain the lat/long position being displayed will be the most accurate your unit is capable of, likely in the range of 5 - 10 meters.

My two units are only .001 apart on either lat/long and I know the accuracy of my Etrex is +/-3 meters.

In my area (MD just NE of Wash. DC) the Google maps accuracy (for coordinates! Addresses are another, long story completely) appears to be spot on for at least the closest NGS survey marker that I can almost make out in the satellite view.

Try these two links to search for a benchmark nearby that you can visit and 'calibrate' your gps and Google maps to for your area:

http://www.waymarking.com
http://www.geocaching.com/mark/

The one I used close by me is here, for example:
http://www.waymarking.com/waymarks/WM1X5P_SPRINGBROOK_HIGH_S...

@GadgetGuy2008: I am tending to agree that multiple readings hasn't helped much. That one reading after waiting a couple of minutes is sufficient.

@JD4x4: Now that is something that I never thought about!!! There are several (supposed) bench marks close by, but not confirmed. I'll have to check it out.

One of my units (Garmin GPSmap76csx) can average a location for presumably greater accuracy. I've not seen this produce results much different from the initial location.

Expand a bit on exactly how you take three coordinate readings? Are you at the location with a GPS unit? Or, are you using Google and other sources?

With respect to Google (or BaseCamp) coordinates, I think it matters how much you expand the map. It is difficult to point at exactly what point you want when you don't go to the highest map resolution. I see a problem with my clicking on the right point, in other words.

Jim

I wait a couple of minutes before each reading and then press coordinates.

This has been mentioned in another thread, but this is probably the best way to verify that your GPS is as accurate as it claims to be - Use a "Benchmark".

What is a Benchmark? Simply stated, a Benchmark (BM) is a marker that has been placed in a location that has been verified the National Geodetic Survey (NGS) to be at a specific Longitude and Latitude, with a high degree of accuracy. (Yes, our Federal tax dollars fund this) In the old days (before GPS) they were determined by surveying. Today they use very expensive GPS units that cost thousands of dollars and it is much, much quicker and probably more accurete as well.

Anyway, these BMs are used by lots and lots of Gov. agencies and also by businesses. BMs are used mainly by map makers and surveyors (although, surveyors are mostly using GPS these days). They are indicated on the maps that the NGS publishes.

So, to check your GPS, you need to find a BM. It's not hard to do - and no maps to buy!

At the NGS/NOAA site, there is an interactive map that will display BMs in your area for you. It's rather "Clunky" (not as flashy and smooth like Google Earth), but it does work! Go to:

http://www.ngs.noaa.gov/cgi-bin/datasheet.prl?Explain=NgsMap

Be sure to follow the instructions about testing your pop-up blocker. Zoom in to your area and then click the "Display Stations" button. A list will "Pop-up" listing the BMs in the area displayed. You can use the listed L/Ls to go to the BMs and check your GPS.

The BMs are usually rather small and are not always easy to find, but many are located right out in the open on city sidewalks. Once you know the actual physical location of the BM, you can check Google's or Bing's maps as well (in satelllite view). Have fun!

I haven't played with it much, but I will this weekend to see how truthful my 660 is!!!!

I have been planning to do that since reading the post by JD4x4, but it has been raining. This coming weekend may be a good opportunity.

My question was not about verifying the accuracy but about using an "average" to compute a more accurate reading. Nevertheless checking it against a benchmark would be a good test. It will also prove to be an excellent opportunity to test the hypothesis that multiple readings would improve GPS accuracy.

You are assuming that "averaging" the readings will reduce any error. But that would work only if the error in the readings were truly random such that averaging would reduce it. However, there is nothing that I have heard or read that would lead me to believe that a plot of the three coordinates would "center" on the true location.

I also suspect that the time of day of the readings would change them as different satellites would be in view of the unit.

Excellent feedback. Exactly what I was looking for.

Depends on what are we trying to do, if you want to cruise down to the nearest Taco Bell what difference does it makes if you are off by 500 feet, now if you are trying to hit Anwar al-Awlaki then I would worry about accuracy.

Generally agree with the bigger logic. However, 500 feet might put the POI on a different street and then we would have someone complaining that our files were not accurate. So we need to be good enough to get someone easily to the location AND/OR get an alert.

That is one of the issues I have with "addresses" versus the actual recorded location.

We pulled off the interstate to go to a Starbucks. No Starbucks! We drove up and down the main street a bit, never saw it. As we got back on the interstate, there it was. The address coordinates were really off. Too late to do anything about it.

I found a bench mark.
Bench Mark-->N34 44.551 W076 44.828
Average ----->N34.44.553 W076 44.828

That produced an average error distance of 12'. According to the GPS derived coordinates where accurate to within 16'. The very first reading was accurate to within 4'.

I also noticed that with the GPS on the ground and not moving, the readings were fairly stable (within 7'). Once I stood up and moved around a bit (even though I was over the benchmark) the reading were a bit more variable (within 18'). Clearly a stable surface helps.

Thanks for your comments.

Just as an FYI, helpful websites tools.
Calculate distance, bearing and more between Latitude/Longitude points
http://www.movable-type.co.uk/scripts/latlong.html

Convert Latitude/Longitude to Decimal
http://andrew.hedges.name/experiments/convert_lat_long/

Kilometers to Feet (km to feet) Conversion
http://www.metric-conversions.org/length/kilometers-to-feet....

thanks for the info and the links

Here is an easier way to find the Benchmarks that metricman mentioned, just plug in your ZIP to find the ones around you, once you find what is in the area you can look at them using various links including Google, MapQuest and many other maps.

http://bit.ly/qZhCLP

Thanks very much for the link. This has been a very informative exercise.