Found markers, GPS accuracy?

[+oisact]

Okay, I set out to find some local markers to get a feel for the accuracy of my GPSr (eTrex Venture Cx). All of them I've found so far have scaled location, so I can't compare that. However all of the altitudes are adjusted. The altitude my GPS reports has been off by at least 17 feet, and has been off both to the the positive and negative. The first marker listed below is the most concerning to me, because my GPSr was reporting an accuracy of +- 6 ft, yet it differs from the marker's entry by 17 feet.

 

FZ0720

FZ0705

FZ0707

 

I'll probably run by a couple of these again today to see what my GPSr is reading now.

 

Is this normal for GPSr? Adjusted altitudes should be extremely accurate, correct? As in what, within inches?

 

Dan East

[+PFF]

Hi, Oaisat, and congratulations on your first benchmarks!

 

Your area seems to have an abundance of marks with SCALED coordinates. So, unless you want to go onto the grounds of the nearby knitting mill or take a hike up to the High Rock lookout tower, there is no way to check your GPS receiver for horizontal positions.

 

Meanwhile, if the elevation reading on your receiver ever matches what is published for the disk, it is strictly a coincidence. Our consumer models simply are not accurate for vertical positions. It's just something we have to live with.

 

There are "work arounds" you can use. If you wish to see how much elevation you've gained or lost while hiking, you can set a waypoint at the beginning of the hike, and compare the elevation with your reading at the summit.

 

The same technique works for horizontal distances, as well. I recently had to find a survey pin which was 1,250 feet from the centerline of a highway. I set a waypoint on the road. Then I put the GPS unit into the GO TO mode for that waypoint and walked until it indicated I had 1,250 feet to go. I was pleasantly surprised that the bearing was accurate, also. I found the pin without much difficulty, and it was a lot easier than pulling a tape!

 

Those of us who have checked various models of GPS receivers against ADJUSTED benchmarks have found the accuracy to be excellent. So, even if it is a while before you get to put your unit to the test, it is reasonable to assume that it's accurate to within a few feet, with this disclaimer:

 

Readings can be thrown off by blocked signals (such as next to a building), and by multipath reflections of signals bouncing off stop signs, buildings, automobiles, etc.

 

Do not put your unit on top of the benchmark to take a reading. The mark, itself, can create a reflection. Instead, stand over the mark and hold the unit at face level.

 

Back to those pesky SCALED coordinates. You'll notice in other Forum threads that many of us look up a mark on a topo map before heading out. If there is an "X" near the spot, we put the cursor over the "X" and then read the coordinates. Those coordinates are what we put into the GPS memory. I use the TOPOZONE feature available here on GEOCACHING.COM. It's a radio button on the right side, under the small map. It will bring up a topo map with a red dot at the published coordinates. Look around the area for a printed "X". If you see it, go to the left side of the screen and make two changes. Adjust the scale to 9K, and set the coordinates to the scale you prefer (probably DDMM.MMM). Then click exactly on the "X" and read the coordinates at the top of the screen. (Try this with FZ0720 and FZ0705.)

 

Wintertime (Patty) is the Queen of predicting coordinates from topo maps. Perhaps she will join this thread with some tips, or you can drop her a note. Again, it is amazing how many of the marks in your area have SCALED coordinates. But since that's what you have to work with, I hope these suggestions give you an advantage!

 

-Paul-

Edited September 21, 2007 by PFF

[+oisact]

Thanks for the reply! So am I correct in understanding that the accuracy the GPS reports applies to location and not altitude?

 

The smokestack at the knitting mill is still in good shape, but is not accessible. That property is for sale, and is fenced and gated. I don't even know who to ask to gain access, because no-one stays at the site. Maybe I can catch someone mowing sometime (they do keep it mowed nicely though!).

 

I was planning on heading to the top of the mountain soon, hopefully today. There are a couple caches up there too I'd like to grab.

 

I have heard a few mentions of "averaging" a location, and seen pictures of a GPS (Magellan I think) that was in some sort of averaging mode. My GPSr doesn't seem to have such a mode. If my GPS is reporting a good accuracy (like the +- 6 ft I got yesterday) am I correct in assuming it is really that accurate? Should I specify the accuracy the GPS is reporting when I log a marker location? At the post office I couldn't get better than +- 30 ft because the marker was against the side of the building. The railroad station marker was even worse, because it was under a roof and against the building. In cases like that do you try to get a reading out in the open, measure the distance to the marker, and translate the reading? Or is it simply not that important to be that accurate?

 

Dan

[holograph]

A handheld GPS unit can have two ways of determining altitude. One is by using barometric pressure, and one is by using the satellite signals. Not all models have a barometric altimeter, but if they do, then that will usually be more accurate than the altitude determined from the satellite signals. However, a barometric altimeter has to be calibrated before use, because weather systems cause changes in pressure.

 

The altitude determined from the satellite signals is highly unreliable, partly because the geometry of the satellites doesn't lend itself to accurate determination of altitude, and partly because the signals tell the receiver its ellipsoidal height, but the receiver then has to estimate the geiodal height from that. The quality of the conversion varies by receiver model. (You can see the difference on the datasheet -- the "geoid height" tells you the difference between the ellipsoidal height and what we normally call elevation. That difference varies from place to place, and the GPS receiver has to estimate it for any location.)

 

If your receiver has a barometric altimeter, you can calibrate it extremely accurately from any of the vertical control points you found. The altitude will be pretty accurate for the remainder of the day, as long as the weather doesn't change. The vertical control points themselves have elevations accurate to within a couple of centimeters.

 

Note, however, that the receivers with barometric altimeters often have some kind of "self calibrate" mode that uses the satellite signals to calibrate the altimeter. That won't be very accurate, because the satellite signals don't give an accurate altitude. However, the barometric altimeter will still let you accurately compare the altitudes at two locations. Without a barometric altimeter, you can't even do that very accurately.

 

If you don't have a barometric altimeter, there's not much you can do except compare what your GPS says and learn not to trust it's altitude reading very much. You'll see that it probably isn't even consistently inaccurate -- meaning that if it is 20 feet too high at one mark, it may be 20 feet too low at the next.

[+oisact]

No, it doesn't have a barometer. I have a Yaesu VX-7R handheld amateur radio that I can get a barometer module for, but I don't know how accurate it would be.

 

I had simply assumed that the elevation would be as accurate as the location, but geometrically I can see why that isn't so. I don't need accurate elevations any more than any other casual GPS user, but it's very good to be aware about the accuracy issue.

 

Dan

[+frex3wv]

PFF wrote: "if the elevation reading on your receiver ever matches what is published for the disk, it is strictly a coincidence."

 

I am curious if this applies to USGS marks also. I found one of those recently and the elevation listed on the mark exactly matched the elevation on my GPS. I find it hard to believe that that would merely be a coincidence.

 

Anyone have any thoughts?

 

(just as an aside - I have searched out and found scaled marks that my gps got to within 50ft of. I know it could be much more off then that - but my experience is that they are closer then you may think. As for adjusted marks - my Mag. has gotten me down to within a few feet at times - pretty sweet!)

Edited September 21, 2007 by frex3wv

[+oisact]

I just noticed on the topo map a BM within a few hundred yards of where I'm at right now, so I walked on over and found it. After I located it I looked up the designator online, and found its altitude is 2361.16 ft. My GPSr reported 2362. LOL

 

FZ0721

 

Dan

[ArtMan]

oisact,

 

As others have said, consumer GPS units are notoriously inaccurate on elevation. They give you a ballpark figure, but not much better than that. If you go on a hike and return to your starting point, the elevation might be 100 feet different or more.

 

Yes, you're right: the displayed accuracy, e.g. +/- 14 ft, refers to latitude and longitude, not elevation. I don't know of any consumer GPS receivers that display accuracy estimates for elevation.

 

I have the same GPS you have, Venture cx, and you definitely CAN take an average reading. I don't have my unit with me, but as I recall you page through to the screen of icons where on the top row one is labeled MARK. Select that, and at the bottom of the next screen will be a box labeled AVG. Click on that, and it will take one reading per second and average them for the coordinates of your position.

 

I wonder, however, if the average really represents a significantly more accurate position in terms of ground truth, since over the course of a few minutes you would be using the same satellites in approximately the same orientation.

 

As to finding adjusted coordinates, even if you don't have any NGS marks in your area, there may be others, placed by a municipality, county government or other entity. You might check with your city or county surveyor or public works agency. My county — Arlington, VA — has maps and datasheets (with diagrams!) online; yours may be in a file cabinet.

 

-ArtMan-

 

PS - This is my 1,000th forum post! (I gotta get a life....)

[+PFF]

When you go to High Rocks, try to recover the reference marks, as well as the main disk. Here are some coordinates you can plug into your GPS unit.

 

RM1 N36 54.451 W081 04.109

 

RM2 N36 54.450 W081 04.117

 

These were calculated using a NGS computer program and they should allow you to walk right up to the reference marks. Let the group know how it goes, and take lots of pictures!

 

-Paul-

[+PFF]

I have a Yaesu VX-7R handheld amateur radio that I can get a barometer module for, but I don't know how accurate it would be.

 

In view of your other hobby, I think you need to go after these nearby benchmarks:

 

FZ0674 HAM

FZ0673 HAM RM1

FZ0675 HAM RM2

FZ0677 HAM RM3

 

73,

Paul (K5PF)

[+Black Dog Trackers]

There's a vertical control disk near my house. I went there with my GPS receiver a couple of years ago and sat down near it and watched the GPS receiver screen. Despite my sitting nearby, it had good satellite views. As I watched it gave lots of different elevation readings. It was pretty much as if I was riding in an elevator of a small building. It would drift up several feet, and then down several feet. It never settled down. So, if you get a reading that's close to what the mark's elevation is, just keep watching.

 

As for a GPSr that has a barometric altimiter in it, that sounds really good, especially for finding USGS marks in hilly country, since the to-reach instructions only have distance and elevation as coordinates on USGS sheets. I don't have one of those and I wonder how accurate they seem in the field.

 

Of course, if there's a change in weather (barometric pressure rising or falling) on the day you're hunting benchmarks, the calibration you do will degrade after time. Does anyone have a web reference for barometric pressure change in millibars per foot of altitude?

[+oisact]

In view of your other hobby, I think you need to go after these nearby benchmarks:

 

FZ0674 HAM

FZ0673 HAM RM1

FZ0675 HAM RM2

FZ0677 HAM RM3

 

I noticed those. The HAM was because of a "HAM BARN" that used to be there. All the repeaters are up on mountains around here, but I'll run over and try to find these since they haven't been logged.

 

73s,

 

Dan

[Bill93]

>I wonder, however, if the average really represents a significantly

>more accurate position in terms of ground truth, since over the

>course of a few minutes you would be using the same satellites in

> approximately the same orientation.

 

My take is that the longer you average the better it gets, down to the internal roundoff and display precision of the unit, but it takes more averaging than you usually want to do in order to get that good.

 

If you average for a very few minutes, you will do a lot to overcome the variations due to receiver noise, and also signal reflections because the reflections change fairly fast as the satellites move.

 

There are other effects, primarily the ionospheric path delays, that don't change that fast so you will still be at their mercy unless you average for hours, or more likely take readings several hours apart and manually average them. I really wish my unit had one more display digit so that I could average on multiple days without the display rounding degrading the average.

 

Several 5-minute averages one after the other will agree pretty well, but if you come back the next day the averages won't match the previous day so well. Thus in a given hour you can do better in measuring the distance between a disk and its RM, or a landmark and a disk, than you can the absolute coordinates of a location.

 

The indicated accuracy of many handheld units is a 95% confidence value, meaning that one time in 20 you will be further off than it says.

[+oisact]

I have the same GPS you have, Venture cx, and you definitely CAN take an average reading. I don't have my unit with me, but as I recall you page through to the screen of icons where on the top row one is labeled MARK. Select that, and at the bottom of the next screen will be a box labeled AVG. Click on that, and it will take one reading per second and average them for the coordinates of your position.

 

Aha! I hadn't noticed that small button. I did a quick test, and the averaged results were the same as the "instant" results with +/- 14 ft accuracy. I'm sure in some cases it will make a difference.

 

I did some experimenting to figure out how to display the location in a large font on this device, so it can be seen clearly in photographs. The only way I've found to achieve that is to go into the Trip Computer, specify Big Numbers, then change the data fields. I have Location, Elevation and Accuracy, and just those three fill the entire display, so it's nice and big.

 

PS - This is my 1,000th forum post! (I gotta get a life....)

 

And just think, you wasted it on me!

 

Dan

[Wintertime]

Wintertime (Patty) is the Queen of predicting coordinates from topo maps. Perhaps she will join this thread with some tips, or you can drop her a note. Again, it is amazing how many of the marks in your area have SCALED coordinates. But since that's what you have to work with, I hope these suggestions give you an advantage!

Thank you for the nice words, Paul, but you covered the topic very well. The only thing I would add is that one needs a calibrated map to do that. Paul mentioned the online Topozone maps, and in another thread yesterday, I mentioned downloadable DRGs (which are available for most states).

 

I have a Yaesu VX-7R handheld amateur radio that I can get a barometer module for, but I don't know how accurate it would be.

Good heavens, am I ever out of the loop about the current state of HTs! I had no idea that such a thing was available. In fact, until I looked at Yaesu's website just now, I had no idea that any waterproof HTs were available for the ham bands!

 

Patty

N6BIS

[+Black Dog Trackers]

On the subject of GPS receiver accuracy, there are several websites that discuss experimental investigations on that. Here is a very good example. It includes graphs on vertical measurements too.

[+oisact]

After I located it I looked up the designator online, and found its altitude is 2361.16 ft. My GPSr reported 2362.

FZ0721

 

This time I got 2368, and the value was floating up and down.

 

Dan

[+Klemmer]

oisact,

 

One more take on consumer GPS receiver accuracy:

You seem to be taking very seriously the accuracy estimate (EPE) that your GPS gives you. I take those with a large grain of salt. I've owned (own?) three different manufacturer's GPS receivers, and the three differ quite a bit at essentially the same location and times. I've also sat down next to an adjusted mark, with the Magellan GPS position only off by 8 feet or so from the mark coordinates (pretty good!), but the EPE was 135 feet. Strange. Those numbers are not reliable, just a very rough guide, IMHO. I also swear that the marketing department has major input into the algorithm that calculates those numbers....

 

Klemmer

[+PFF]

Bill93 wrote:

 

I really wish my unit had one more display digit so that I could average on multiple days without the display rounding degrading the average.

 

Good news! You can take things to an extra digit by using decimal degrees (DD.DDDDDD). It will allow you to see the decimal places which normally are not displayed.

 

Multiply the digits to the right of the decimal by 60 to convert back to DD MM.MMMM. Repeat for Degrees Minutes and Seconds. Or use one of the handy on-line coordinate calculatiors.

 

(This does not increase the accuracy of your GPS unit. It simply allows you to see farther past the decimal point, so you can view the figures being used for rounding.)

 

-Paul-

[+Black Dog Trackers]

My GPSr offers only

 

hDDD.DDDDD°

hDDD° MM.MMM'

hDDD° MM' SS.S"

 

i.e. no 6th decimal on degrees.

The 5th decimal of degrees is about like a third of a 0.1 second.

Another way of looking at that is that the decimal-degree mode is like reading to 1/30 of a second instead of 1/10 second.

 

My GPSr acts 'funny' with the 5th deimal of degrees. When I set it to a value, it often changes it by 0.00001 degrees. So, not much advantage in the degree decimal mode, so I stick with the 1/10 second mode.

 

Like PFF, I figure a unit's intrinsic accuracy is really the same in any of these modes.

 

My GPSr also does not average vertical elevation.

[+GEO*Trailblazer 1]

That's what I use my GPSTrackmaker for.

 

You can get a graph chart of the elevation average.

 

It also plots the average of the readings.

It does a whole lot more but have yet to figure it all out.

[+Black Dog Trackers]

If you get an average, also get the standard deviation.

[+oisact]

I just found another, and my GPSr was within .8 feet of the elevation when I set it down on the marker. However after it set there a bit it started inching up foot by foot over the course of 100+ samples during averaging. It seems to me that when the GPSr is in motion the elevation is more accurate then when stationary, which is counterintuitive to the way the location value behaves.

 

This is (finally) a location adjusted marker (FZ1351). The official location:

N 36° 57.119 W 081° 05.148 (NAD 83)

My GPSr location:

N 36° 57.119 W 081° 05.153

It was reporting accuracy of +/- 5.6 ft. How much distance does .005 represent?

 

Dan

Edited September 21, 2007 by oisact

[+Black Dog Trackers]

oisact -

 

I recall doing a post on the sizes in feet of the GPSr limits of precision and accuracy, but instead of looking that up just now, partly because it varies depending on what latitude you're at, a neat trick to do is put in 2 coordinate sets in your GPSr, one differing by .005 or 0.1 second, or whatever you want, and then use your GPSr's go-to function (on mine it is called Reference or something) and read the distance from one location to the other in feet.

[+oisact]

put in 2 coordinate sets in your GPSr, one differing by .005 or 0.1 second, or whatever you want, and then use your GPSr's go-to function (on mine it is called Reference or something) and read the distance from one location to the other in feet.

 

I get 25 feet. The ACME Mapper reports 28 feet. Definitely not less than the 5.6 accuracy the GPSr stated.

 

I asked this earlier in my mess of questions, but no-one answered. When a marker has an adjusted elevation or location, exactly how accurate is it? Within inches, feet, yards? Does the precision vary from marker to marker?

 

Dan

[+PFF]

I just found another, and my GPSr was within .8 feet of the elevation when I set it down on the marker.

 

Don't set it on top of the marker. Hold it in your hand. Use a tape measure to determine where to hold it such that it is five feet off the ground. Then subtract 5 feet from the reading.

 

As someone mentioned earlier, the elevations which are adjusted are precise, just as the horizontal positions which are adjusted are precise.

 

Fortunately for those of us in this wacky hobby, we seldom find disks based upon elevation. Since that's the least-accurate function of a GPS receiver, we'd be lost if we had to get the elevation exactly correct!

 

By the way, surveyors in the field are getting second-by-second corrections from ground control stations. Sometimes, these come via radio. Or, with slightly older equipment, they take the unit back to their office and run the readings against the correction history.

 

Glad we don't have to be THAT precise!!

But the point is, they're using a technology which far surpasses what we have.

 

-Paul-

[holograph]

I asked this earlier in my mess of questions, but no-one answered. When a marker has an adjusted elevation or location, exactly how accurate is it? Within inches, feet, yards? Does the precision vary from marker to marker?

 

Rule of thumb: adjusted elevations should be accurate to a few centimeters, and adjusted locations should be accurate from a few millimeters to a meter or more, depending on the "order" of the station. If you see a note in the datasheet that the position is "GPS OBS" (observed by GPS) then the location will probably be accurate to about a centimeter. Geodetic survey quality GPS equipment with post-processing of the data can give accuracies of a centimeter or less.

 

It all depends on when the position was measured, how it was measured, and how accurately it was measured. All that information is in the datasheet, but it takes some effort to understand it. The newest datasheets contain estimates of accuracy directly, but only for stations that were observed by modern GPS equipment. All the older stations still use the "order" system.

[+Black Dog Trackers]

... and the 'order system' only has specifications in terms of distance accuracy between stations. In other words, it does not answer the question: How close (expected value) to the true lat-lon coordinates are these coordinates? Instead it answers the question: If I used the (adjusted) coordinates of this station and the (adjusted) coordinates of station X over there, how accurate of a distance between here and station X is a distance calculation based on the 2 sets of coordinates?

 

The first question is an absolute accuracy question, while the second question is a relative accuracy question.

[+Klemmer]

In simplistic terms, an adjusted benchmark's published position is WAY WAY more accurate than your consumer GPS.

[Papa-Bear-NYC]

Bill93 wrote:

 

I really wish my unit had one more display digit so that I could average on multiple days without the display rounding degrading the average.

 

Good news! You can take things to an extra digit by using decimal degrees (DD.DDDDDD). It will allow you to see the decimal places which normally are not displayed.

 

Multiply the digits to the right of the decimal by 60 to convert back to DD MM.MMMM. Repeat for Degrees Minutes and Seconds. Or use one of the handy on-line coordinate calculatiors.

 

(This does not increase the accuracy of your GPS unit. It simply allows you to see farther past the decimal point, so you can view the figures being used for rounding.)

 

-Paul-

I have a Garmin 60CS and it 's screen displays DDD.DDDDD (5 places) also.

 

However If you set the display to UTM it displays to the nearest meter. I checked this against some datasheets for tri-stations (which have UTM down to the millimeter) and the Garmin displays the number correctly to the meter.

 

This is probably the best precision you can display. Certainly a lot better than .1 second.

 

My rule of thumb is a degree is 60 miles. This of course only works for longitude at certaing latitudes (I think around 40 degrees North or south). But it's an estimate thats' pretty easy to remember. So a minute is 1 mile and a second is 1/60 of a mile or approximately 30 yards (29.33), so .1 sec = about 3 yards. I usually estimate .1 sec to be 10 feet.

 

But with UTM, there is no guesswork. It's on the datasheet and with meters east or north it's very easy to visualize on the ground.