GPS Questions Part 2

[Driver8]

I hope this Google Earth piggy-back question is ok, since it's related. As a still-new peakbagger, I am interested with each summit to find the correct high point on my climbs. A favorite resource for hiking intel is Google Earth. It appears that G.E., sometimes, has the correct summit placement, which sometimes is not the commonly reputed and accepted summit. Other times, though, G.E. is, to my eyes, clearly incorrect. Also, its elevation report for a summit is often off from the U.S.G.S. record. My question is, does anyone here know enough about Google's methodology to explain these differences? Is G.E. sometimes right where the conventional wisdom is wrong about a summit location? If G.E. relies on satellite pinging for elevation, I could see it being off a few feet from the U.S.G.S., which would raise the question as to which is more accurate.

Some examples of what I'm talking about:

1. Bear Mountain, on the A.T. in Connecticut, has a "summit" monument, near ground that is 2316', according to my phone GPS and to the U.S.G.S. This clearly, though, by a few feet, is not the summit - the A.T., northbound, descends a few feet in its final 75 feet or so NOBO to the monument, and the correct high ground, clearly, is about 60 yards NW of the monument. My GPS, and G.E., have it closer to 2323', which is I number I've heard elsewhere. Google Earth has the proper summit marked pretty close to correctly, maybe off by 5 feet to the southeast.

2. Moosilauke and Eisenhower both show, in G.E., a summit at a good remove from their commonly accepted summit. When I hiked Eisenhower, though, it seemed abundantly clear that the high ground was right where the big cairn is in the middle of the dome. I looked for neighboring rocks whose heads might poke above that level on a recent visit, and though I didn't take measurements or do an exhaustive survey, the visual scan didn't yield any candidates. IOW, Eisenhower's accepted summit seems correct, not Google Earth's. Moosilauke's proper summit does appear, from my two visits there, to be atop a rock about 5-10 feet NW of the famous orange sign. Google Earth suggests that it's around 0.1 NW, just west of the A.T. Looking at photos I took last May in that direction from the summit, I considered that maybe G.E. was right, as the ground seems to swell upward to the NW. On my second visit, last month, it seemed pretty clear to me, approaching from the north, that the correct summit is near the sign - you're clearly looking up to the summit area from the area of the G.E. "summit" - the ground between the two mostly rises, though dips a foot or two just NW of the rocky summit area, then climbs a few feet more to the summit rocks. It was a bit confounding, though, as I took a break at the summit, that the ground seems to rise to the NW from the summit, but I chalk this up to the optical illusion of a long, fairly level area seeming to "rise" as it gets further away - to my eyes, coming south, I was clearly looking up to the summit, where looking from the summit NW it just felt vaguely up.

3. Mount Washington is accepted as 6288' above sea level. G.E. has it at 6291'. There are numerous instances of such variance - rare is the peak I've looked at in Google Earth whose elevation matches exactly with the U.S.G.S. number.

4. Google Earth 3D imaging of famous places like the White 4Ks appears to be pretty true-to-life. I have found this to be less the case with less famous places, such as Metacomet Range peaks like Mt. Tom, West Peak, and Talcott Mountain. Sharply defined cliff faces, for instance, are often rounded in the 3D representation. I figure Google puts a premium on getting the famous places right, which would be understandable, and doesn't sweat less exalted venues, figuring on improving these images with time.

So anyway, some more cartography/GPS-related material to chew on. I hope it's welcome in this thread - please forgive if not.

 

[DougPaul]

I hope this Google Earth piggy-back question is ok, since it's related. As a still-new peakbagger, I am interested with each summit to find the correct high point on my climbs. A favorite resource for hiking intel is Google Earth. It appears that G.E., sometimes, has the correct summit placement, which sometimes is not the commonly reputed and accepted summit. Other times, though, G.E. is, to my eyes, clearly incorrect. Also, its elevation report for a summit is often off from the U.S.G.S. record. My question is, does anyone here know enough about Google's methodology to explain these differences? Is G.E. sometimes right where the conventional wisdom is wrong about a summit location? If G.E. relies on satellite pinging for elevation, I could see it being off a few feet from the U.S.G.S., which would raise the question as to which is more accurate.

You have to find out where the elevation number comes from--there are several possible sources:
* If there is a benchmark at the summit, then the elevation will be surveyed. (This is generally the published peak altitude.)
* Otherwise, the elevation is likely from a DEM (Digital Elevation Map) which gives the altitude for 30x30 to 90x90 meter squares. This number is unlikely to match published peak altitudes. (When you move the pointer around and the machine gives you a numerical elevation, it is usually from a DEM...)

See http://www.vftt.org/forums/showthre...-electronic-maps&p=77470&viewfull=1#post77470 for more info.

Note also that the summit marks on many maps are not at the actual summit locations--you may have to think of them as labels rather than summit marks.

Doug

 

[DougPaul]

This is a bit off topic but it is a GPS reading that you don't see very often. The waypoint was captured at the location shown but the photo was taken later in camp. View attachment 4658 Note: the Garmin GPSmap 76CSX will not permit manually loading in the N90 location, I tried. It will allow 89Deg, 59 Min 59.9 Sec. To get to 90 Deg North, you actually have to go there

Sure you can set a waypoint at 90 deg, 00 min, 00.0 sec from the front panel--you just have to be suitably devious. You just can't enter it as 90deg... (And no, neither I nor my GPS has been anywhere near the North Pole.)

Doug

 

[Driver8]

You have to find out where the elevation number comes from--there are several possible sources:
* If there is a benchmark at the summit, then the elevation will be surveyed. (This is generally the published peak altitude.)
* Otherwise, the elevation is likely from a DEM (Digital Elevation Map) which gives the altitude for 30x30 to 90x90 meter squares. This number is unlikely to match published peak altitudes. (When you move the pointer around and the machine gives you a numerical elevation, it is usually from a DEM...)

See http://www.vftt.org/forums/showthre...-electronic-maps&p=77470&viewfull=1#post77470 for more info.

Note also that the summit marks on many maps are not at the actual summit locations--you may have to think of them as labels rather than summit marks.

Doug

Thanks for your reply, Doug. In Google Earth, is appears clear and consistent, peak to peak, that they put their little light green "peak" marker at what they have as the highest elevation for that mountain in their database, which appears to be the DEM as you say. Move your cursor around a summit area and you continuously get an "elevation" figure at bottom right, and that number is largest, i.e. elevation is highest, at or very close to the green triangle summit marker on several mountains I've examined using it.

As such, G.E. "thinks" the summit of Moosilauke is 0.1 or so NW of the actual summit, with an elevation of 4820'. It has the proper summit at about 4804', and has ground just northwest of it that you step down to from the summit rocks at 4806'. It unquestionably is inaccurate, except perhaps as to the correct elevation of the actual summit.

Likewise for Mount Washington - the green "peak" marker, with "elevation" 6291' is in the gravelly area between the actual summit and the Adams building, a good 10-12' lower, in the real world, than the true summit, which is billed at 6289'. (For that matter, in a non-elevation aside, Google Earth lists the Old Bridle Path up Lafayette and the Tuckerman Ravine Trail up the cone of Washington as segments of the Applachian Trail.)

I guess the upshot is that Google Earth, enticing tool that it be, is still very much approximate.

 

[DougPaul]

I guess the upshot is that Google Earth, enticing tool that it be, is still very much approximate.

As I said earlier, all mapping methods have errors (ie they are all approximations)... The fact that we have multiple sources of data and high-accuracy hand-held navigation devices simply makes the errors visible to the lay user.

The DEM data is generally from the SRTM (Shuttle Radar Topography Mission). http://www2.jpl.nasa.gov/srtm/ http://srtm.usgs.gov/.

USGS topo maps were generated by a combination of ground surveying and stereo (3D) aerial photography. They are available on paper and in DRG (Digital Raster Graphics, ie scanned) form or DLG (Digital Line Graphic) form. http://topomaps.usgs.gov/

Some companies/people have generated topo maps from the DEM data, but the maps are generally inferior to the USGS maps. Generating DEM maps from USGS topo maps is very difficult because of ambiguities in the maps and difficulty reading them. It isn't generally done--SRTM data is much easier to use.

Doug

 

[Driver8]

As I said earlier, all mapping methods have errors (ie they are all approximations)... The fact that we have multiple sources of data and high-accuracy hand-held navigation devices simply makes the errors visible to the lay user.

The DEM data is generally from the SRTM (Shuttle Radar Topography Mission). http://www2.jpl.nasa.gov/srtm/ http://srtm.usgs.gov/.

USGS topo maps were generated by a combination of ground surveying and stereo (3D) aerial photography. They are available on paper and in DRG (Digital Raster Graphics, ie scanned) form or DLG (Digital Line Graphic) form. http://topomaps.usgs.gov/

Some companies/people have generated topo maps from the DEM data, but the maps are generally inferior to the USGS maps. Generating DEM maps from USGS topo maps is very difficult because of ambiguities in the maps and difficulty reading them. It isn't generally done--SRTM data is much easier to use.

Doug

Thanks again, Doug. Very informative. Among you and many others here run deep rivers of understanding of the world of mapping and GIS. As an amateur cartophile, I am so pleased to make your acquaintance in that regard. Now and then I may want to pick your brains along these lines, I hope unobtrusively. Thank you again and in advance.

 

[Nessmuk]

I guess the upshot is that Google Earth, enticing tool that it be, is still very much approximate.

Well I don't know about flagging summits, but I find Google Earth extremely accurate with registration to GPS and to the real world.

I use GE to plot my route down the Yukon River for the 440 mile and 1000 mile canoe races in the Yukon and Alaska. I plot the precise route and turn points for our route to follow during the race, 738 carefully selected waypoints at critical turn points and permanent islands in the case of the 1000 mile. I use GE to "fly" the route to memorize it during winter months of paddle machine training. I've done those races four times now, and find GE accuracy to be surprisingly exceptional. Except for where the river has altered the landscape (which it does in places every year), pre-plotted GE points loaded into the GPS and reality typically coincide to within a boat length. GE imagery maps of the Yukon River are generally up to date to within 2-5 years of the present time. The most current topographic maps are all decades old and virtually worthless due to river erosion and altered oxbows.

 

[Driver8]

Well I don't know about flagging summits, but I find Google Earth extremely accurate with registration to GPS and to the real world.

I use GE to plot my route down the Yukon River for the 440 mile and 1000 mile canoe races in the Yukon and Alaska. I plot the precise route and turn points for our route to follow during the race, 738 carefully selected waypoints at critical turn points and permanent islands in the case of the 1000 mile. I use GE to "fly" the route to memorize it during winter months of paddle machine training. I've done those races four times now, and find GE accuracy to be surprisingly exceptional. Except for where the river has altered the landscape (which it does in places every year), pre-plotted GE points loaded into the GPS and reality typically coincide to within a boat length. GE imagery maps of the Yukon River are generally up to date to within 2-5 years of the present time. The most current topographic maps are all decades old and virtually worthless due to river erosion and altered oxbows.

It probably is super-serviceable for purposes where the level of precision required is in the 10's of feet to 10's of yards - the placement and shape of islands in a riverbed, general boundaries. It's clearly wrong about the summits I've pointed to and to others I haven't, where the level of precision I'm looking for is in the under 10 feet range. Wonderful tool, for sure - good to know its limits, as with any tool.

 

[DougPaul]

It's clearly wrong about the summits I've pointed to and to others I haven't, where the level of precision I'm looking for is in the under 10 feet range.

The nominal accuracy of a consumer GPS is typically 95% probability of being within 10 meters (33ft) of the correct value. (Professional mapping GPSes can achieve sub-centimeter accuracies.) Given the accuracy of your GPS, I wouldn't worry about 10 ft. I would think more on the lines of 30--50 ft.

Wonderful tool, for sure - good to know its limits, as with any tool.

Agreed, only I would strengthen it to say: "important to know its limits".

Doug

 

[Driver8]

The nominal accuracy of a consumer GPS is typically 95% probability of being within 10 meters (33ft) of the correct value. (Professional mapping GPSes can achieve sub-centimeter accuracies.) Given the accuracy of your GPS, I wouldn't worry about 10 ft. I would think more on the lines of 30--50 ft.

I've found that the GPS - an app relying on the phone - and USGS maps and GE agree pretty closely most of the time, but there are times when the GPS is off in the 10's of feet. It's clear to me that, for the time being, GE can't be relied upon for locating a mountain's summit.

One thing I've seen so far with the GPS, which I haven't checked rigorously, is that whether the elevation number it gives for a place is correct or not, I've not seen it showing a lower place as having a higher elevation than a clearly higher place in the same vicinity. It doesn't make the same mistake I've pointed to GE making on Moosilauke's and Washington's summits. It could be off for all points in an area by 10-20 feet, but it appears to be off consistently across the area, say a quarter mile radius.

 

[DougPaul]

I've found that the GPS - an app relying on the phone - and USGS maps and GE agree pretty closely most of the time, but there are times when the GPS is off in the 10's of feet. It's clear to me that, for the time being, GE can't be relied upon for locating a mountain's summit.

To determine the accuracy of a GPS you have to leave it at a surveyed location for 24hrs or longer while recording a continuous track (by time--1 point every 30 or 60 seconds will do). You can then compute the average or RMS (root-mean-square) error. (Make sure you use matching datums or convert to matching datums.)

Lacking a surveyed location, the best you can do is to record a track as above with the GPS at a stationary location. This will enable you to compute the spread in estimated locations, but not the actual error.

Long-term averaging will generally yield a more accurate location than a single reading.

One thing I've seen so far with the GPS, which I haven't checked rigorously, is that whether the elevation number it gives for a place is correct or not, I've not seen it showing a lower place as having a higher elevation than a clearly higher place in the same vicinity. It doesn't make the same mistake I've pointed to GE making on Moosilauke's and Washington's summits. It could be off for all points in an area by 10-20 feet, but it appears to be off consistently across the area, say a quarter mile radius.

You first have to ask what kind of altimeter you are using--the Garmin "sensor" units have barometric (air pressure) altimeters the "non-sensor" use the altitude from the core GPS.
* GPS altimeter: accuracy depends on the satellite constellation (positions). On the average, the vertical accuracy is about twice the horizontal accuracy, so ~20 meters (66 ft).
* Barometric altimeter: accuracy depends critically upon the calibration. (Choices: set the altitude, set the pressure, use the GPS altitude (auto).) Garmin gives a spec of +-10 ft if properly calibrated. I personally use the auto calibration--the GPS altitude (high short-term error, small long-term average error) smoothed by the barometric data (small short-term error, large long-term error) gives the best of both worlds. I calibrate during the drive to the trailhead (typ 3 hrs) (and of course continue calibrating during the hike) and generally get altitudes within 10 ft of the published peak altitude.
Note: auto calibration takes some time--turn the GPS on a half hour to several hours before relying on the altitude reading.

Warning 1: The GPS computes the height above the center of the earth and subtracts the mean sea level to get the altitude. Geoid models are used to estimate the local mean sea level--to make completely accurate altitude comparisons, you have to account for differences in the geoid models, if any...

Warning 2: The barometric sensors can detect the altitude difference between your head and feet, so hold the GPS on the benchmark or peak when comparing altitudes or subtract the GPS height above ground from the altitude reading.

Warning 3: Barometric altimeters read out the cabin pressure of a pressurized aircraft, not the actual altitude. (You can get a GPS altitude from a sensor unit--the method varies with the model, but it is generally from the satellite page.)

Doug

 

[bikehikeskifish]

Put your GPS in a ziplock and squeeze it (while shut, full of air) and watch the barometric elevation plummet GPS and GE are remarkably accurate. You can always wander around and find the highest pebble, if you are not within obvious visual distance from the summit. Where GE really rocks is finding old woods roads when doing untrailed peaks.

Tim

 

[Driver8]

Put your GPS in a ziplock and squeeze it (while shut, full of air) and watch the barometric elevation plummet GPS and GE are remarkably accurate. You can always wander around and find the highest pebble, if you are not within obvious visual distance from the summit. Where GE really rocks is finding old woods roads when doing untrailed peaks.

Tim

Thanks, Tim. I'm not that far along. GE is also fun when you're stuck home and can't hike - helps vivify the daydreams. ...

Thanks, Doug, as well.