I am preparing a hiking database, showing distance, start/end elevations, elevation gain, anticipated time, etc. for various hiking legs. I have defined 86 waypoints, and some 200 possible sequential connections among them. I have GPS coordinates for each waypoint, but elevations are problematic because I cannot trust the elevations from the GPS, a Garmin 64ts.

How can I determine elevations from GPS coordinates? For example, the door to my home is at 20°18.237'N 103°15.804'W. My GPS gives varying altitudes for the same locations on different days, sometimes differing by as much as 100--200 feet.

The Garmin can display elevation from both barometric and GPS sources. The barometric readings are calibrated to a "known" altitude before every hike: according to Google Earth, the altitude at street level next to my door is 5,204 feet. I realize that Google Earth's elevations are just an interpolated average between topographical lines and in hilly terrain where I live this is likely to err by hundreds of feet. Presently I am using elevations that are the average of as many as 10 different samplings from the GPS over a period of months. Just sitting on my desk as I type this, the GPS elevations shown have varied by 26 feet barometric, and 37 feet GPS.

One large contributor to the inconsistencies I get is that there is a considerable lag between arriving at a waypoint and the GPS settling down to a reliable (?) elevation reading, particularly if there is a large change in elevation in the previous few minutes. It can take five minutes or more for the readings (both barometric) and GPS) to stabilize. If I have just arrived at a waypoint that is, say, 250 feet higher than the previous waypoint, the GPS will generally show an elevation that is about 50-100 feet lower than the true elevation. Over the next 5-10 minutes, I can watch the elevation numbers advance, climbing one foot at a time until they finally reach the correct elevation. Except that if I wait another 5-10 minutes, they might (or might not) continue climbing past that elevation while I stand perfectly still, watching the GPS screen.

The GPS distances seem to record pretty accurately, and the GPS coordinates seem to be spot-on, at least when compared to Google Earth’s coordinates for known locations. The elevations… not so much.

So, I’ve got a set of believable GPS coordinates. How can I get reasonable elevations for those points? Absolute accuracy is not what I’m looking for. Every one of them could be 400 feet off, as long as it was the same 400 feet in every case. But I want the elevation difference between point A and point B to be the same every time I measure it, plus or minus 10 feet if possible, 25 feet at most.

How can I do this?

tanstaafl.

"How can I do this?"

This Wikipedia pages has a lot of math I don't understand:

https://en.wikipedia.org/wiki/Dilution_of_precision_(GPS)

but the two pictures speak to the issue. "DOP" depends greatly on the angle from your location to satellites in view. The example in the canyon show the effect of what I would call narrow angles. For altitude/elevation, low DOP depends more on getting signal from satellites whose angle is closer to a tangent to earth curvature, so lower-lying features like hills and trees make it hard to get signals from stallites that can provide a flatter geometry. Airplanes flying at altitude don't have the same problem of acquiring and keeping enough satellites to provide low VDOP.

This doesn't answer the question (and I made some of this up) but there may not be a good answer. No free lunch as they say.

Yeah, that much I kinda figured out. But I am not looking for a way to improve the accuracy of my GPS. Instead, I am looking for a piece of software or a website where I can enter my GPS coordinates (which are) accurate and get an elevation reading that even if not totally accurate will at least be consistent.

tanstaafl.

Well, either the altitude info is coming from a geospatial database that has the elevations at close enough accuracy and 2D frequency to map usefully to your recorded GPS waypoints or you need an altitude accurate device to take with you to each location.

I don't know if there is a sufficiently accurate ground profile database for your area, nor have I any idea how one would get access to one if it exists.

Consumer handheld GPS is not intended to provide high degrees of accuracy for altitude. The primary focus is on x-y accuracy and speed of determining current location. I think there are some high accuracy handheld GPS products that have better altitude accuracy.

Perhaps the GPS products sold for land survey work would provide the necessary accuracy. No idea on cost or how usable such a unit might be for your purposes. This land survey model is apparently accurate to 2cm vertical.

My Magellan Crossover GPS has a screen that shows altitude. I always assumed it was calculated.

You are only going to be able to do this if someone had done a high resolution/precision survey of the elevation data in the area you are interested in. Given where you live, I suspect you've got less chance than average.

Google do provide API access to their elevation data, but as you've guessed it typically is rough and ready in its precision and accuracy:

https://developers.google.com/maps/documentation/elevation/intro

Here is an example bit of UI using Google's data:

http://www.geocodezip.com/v3_elevation-profile_distance.html

In some parts of the world much better data sets are available:

https://environmentagency.blog.gov.uk/20...risk-knowledge/

I've you want I could knock you up a quick web page to paste your coordinates into to give you Google's idea of what the elevations are.

A bit of Googling suggests that most of their elevation data it is interpolated from a survey with 30 metre resolution

https://productforums.google.com/forum/#!topic/earth/3Th8MuHzKtE

https://www.quora.com/How-accurate-are-relative-elevations-in-Google-Earth

So in summary, the data you want probably doesn't exist. We don't know nearly as much about the detail of surface of our planet as we might think we do...

...and your GPS receiver could capture the data you want, but only by tediously sitting for extended times at the locations of interest to get a decent average.

Tha would be an unwarranted amount of work on your part. Since there are only (at present) 86 waypoints I have to deal with, it is easy for me to just use Google Earth to navigate to each point and write down the elevation from the screen. I have some of the 86 done that way already and will collect 14 more this morning, and I can assign varying degrees of credibility to them based on whether they are on top of a ridge or down in an arroyo, and whether the barometric readouts (particularly when I can work from an average of multiple data readings) seem not too far out of line with what the GPS satellites are saying.

Thank you very much for the offer. However, if the idea of playing with this really appeals to you, go ahead and I can compare my manually collected elevations to what your results show as an error check.

tanstaafl.

Assuming that you meant Garmin 64st...

Looking at the specs online, that unit looks like it's a good high-tech handheld receiver with good circuitry and reception. So I don't think you're going to find anything more accurate in a handheld unit.

The fact that the GPS elevation data takes a while to settle down is troubling. Question: do you find that the elevation data settling down is correlated with its DOP readings? In other words, a better DOP means less movement of the elevation data?

Also does your GPS report its current VDOP number? The vertical precision indicator? Perhaps by digging into one of its configuration screens? If so, pay attention to that. If that number goes good and your altitude settles down, you probably have a "good enough" representation of altitude for your purposes.

No.

But, to reiterate, I am NOT trying to improve the accuracy of elevation readings of my GPS. That is a lost cause.

Instead, I want to find a way to generate good elevation numbers based on the GPS coordinates that I have recorded. Elevation numbers created by something with more precision and repeatability than a hand-held GPS unit.

Google Earth does it, and while their numbers are quite consistent in that the same GPS coordinates will always show the same elevation, in mountainous terrain the number shown can be woefully inaccurate.

Today I recorded elevation information for 14 waypoints that was comprised of: (1) The barometric reading; (2) The GPS satellite reading; (3) The average of all previous readings; and (4) the elevation for those coordinates as reported by Google Earth. I think what I'll do is take the average of those four numbers and call it good. Maybe if one of the four is hundreds of feet different from the others I'll ignore it and average the remaining three. If I end up with elevations that are within 50 feet of the true elevation (how will I ever know?) I'll be happy.

tanstaafl.

Just get rid of the barometric readings from the data. They really just don't add any value unless you go to immense effort to normalize them hourly against the weather etc..

A typical hike might run three hours, and I always calibrate the GPS to a known altitude at the beginning. I'm probably getting better accuracy from the barometric readings than I am from the satellites, because just about half of my waypoints are down in arroyos. Garmin by default uses the barometric readings, you have to dig into the menu structure to even find the GPS elevation readings.

tanstaafl.

To achieve that, you would need:

- Certainty that someone, sometime, has done an accurate survey of the area in question. If the area is remote, I'm betting that's unlikely.

- Knowledge of where to retrieve the data from that survey.

Google Maps is not likely the source of data you're looking for. My guess is that the data you're looking for isn't publicly available at all. Your handheld GPS might be the best data you've got access to right now.

Hm. I wonder how much used surveying equipment is on Ebay?

Unfortunately, I am afraid that Google Earth is the only available source of data. If it agrees within reasonable limits with the data I have gathered myself (say +/- 100 feet) I will feel comfortable adding it into the mix.

The reason I am trying to gather these data is so that I can come up with a reasonable estimate of total cumulative elevation gain, distance traveled, and elapsed time before leading a group hike, so people can evaluate whether or not that particular hike is suitable to their abilities.

We have a fairly active hiking group, sometimes as many as 30-40 people show up and choose among the three hikes offered that day: Short, Medium, Long. The short hike usually runs about two hours with less than 1,000 feet elevation gain. The Medium hike about three and a half to five hours, less than 3,000 feet elevation gain. The Long hike... well, anything goes. We had one where we missed a trail turnoff, ended up with about 4,000 feet gain, eleven hours. And one memorable hike where a "shortcut" didn't work out so well and the hikers spent the night on the mountain. Only two hikers were on that one, we would never take a group out on an exploratory hike.

We hike in a group twice a week, and I go out solo on Sundays with Amber, my apricot colored standard poodle who positively lives for these outings in the mountains. The intriguing thing about the group is that the five fastest, strongest, best hikers in the whole group, the ones that nobody can stay with, are all women in their 60's and 70's. I am one of the strongest male hikers, and there is no way that I can stay with what I call the "Fearsome Five." Oh, it is so humiliating to be humbled by those "gurrls". I tease them unmercifully, and they just smile and walk away up the mountain, leaving me in the dust. I love it!

tanstaafl.

One thing I learned from the Titanfall 2 campaign is to never trust a shortcut. Especially one proposed by a giant hulking murder bot.

Google Earth isn't the only available data, but it's probably the easiest to digest. If you dislike Google Earth's data for some reason, there are two different types of datasets you can download -- one is called GIS (geographic information system), the other is DEM (digital elevation model). To make any use of those, you'll need some GIS software (QGIS works for me). The GIS datasets you can find will most likely be contour lines -- but at what resolution depends on the data set. The DEM files will be at a much higher resolution, but you'll have to do some work to get the elevation out of them. And honestly, that's probably all the same work that Google Earth (and ArcGIS) have already done.


Oh, have you tried Gaia GPS, then? It lets you plot a route, and provides an elevation profile of that route, along with estimated trip length.

Generally the federal government has this data on record somewhere, because it's useful for all kinds of things. For example, the data for Mexico will be here, somewhere (but since I don't read Spanish, I couldn't tell you where). And if they don't, you can often find pointers to it from universities or other sites dedicated to GIS.

They might not make it publicly available (or, if they do, they might charge for it), though. For example, the last time I looked into it (which was about 10-12 years ago, admittedly), Canadian DEM files were freely available at a certain horizontal resolution (maybe 50m?), but you had to pay if you wanted, say, 10m resolution data.

I fed the data for Whistler into an OpenGL demo app I found; it looked pretty cool. This was before Google Earth, btw.

Holy cow, yes, that https://www.gaiagps.com/map tool that Canuck linked looks like exactly what you're looking for.

So each country maintains its own set of databases for the info, correct?

It looks like all the data Doug could want is there, on that site, freely available (I downloaded the elevation data for a few random grid squares).

Sadly it would take a lot of work just to work out where the relevant data file for Doug's area is

Sigh... All I want to do is key in a pair of GPS coordinates, and have a reasonable estimate of elevation show up on the screen.

So far Google Earth comes the closest, if I can live with the inaccuracies inherent in their algorithm that interpolates the true elevation between the actual "hard" data points.

I think what I will do is use an average of my data acquisitions and Google Earth, discarding any data that IMHO falls outside the believable range based on how many more or less consistent data I have acquired, my knowledge of where the reading was taken (on top of a ridge or down in an arroyo), how close the barometric vs GPS readings are, and whether the readings were taken on a solo hike where I could spend a minute or two waiting for the numbers to stabilize.

This is hardly a sterling example of Scientific Method, but with the "soft" data combined with my experience and intuition, it will probably be close enough.

tanstaafl.

Why don't you just overlay your trail map on that grading map on gaiagps.com?

Kick off the willpower to go to space and look back as a nation, or better yet an entire species, and then keep funding that drive for decades. Then, utilize the public data used to produce more precise elevation data useful for things like this: https://www.youtube.com/watch?v=MJgXaqW3md8

Note, you may end up with more elevation data then requested and may need to move your favorite hiking spots to within the path of an eclipse.

From reading this page:

http://blog.gaiagps.com/plot-a-precise-trail-with-the-gaia-gps-trail-finder/

It doesn't look like their elevation data will be any better than what Doug is getting with Google (though probably quicker to obtain).

Many countries do, but I wouldn't vouch for all countries. It's a reasonable starting point for whatever country you live in, but probably not the easiest starting point if you're looking for data on countries where you don't speak the local language.

In the US, at least some states also maintain a set of state-level databases, and I've even seen county-wide and city-wide databases -- though less frequently.

There's also a lot of sharing that goes on (the Canadian CDSM database started with NASA/JPLs SRTM database), and there's a lot of variation in available resolutions, too. For example about 70% of England has been scanned with LIDAR, but last time I looked, California has very spotty coverage -- predominantly corresponding to wildfire locations.

I've only been looking at this stuff in the last year, but I think things have changed for the better -- for example, the full resolution of the SRTM dataset (30m) was made freely available in 2015.

Iknowrite? Isn't that awesome?!