Hi, here’s a “somewhat” theoretical question - but theoretically life saving
Assume a perfectly true mapping of a path, e.g. along a cliff edge, but all GPS tracks along the path are displaced in the same fashion relative to the map. Since someone on the ground does not know their own true coordinates, the perfect coordinates of the map are not useable for them. On the other hand, it seems safe to assume that all GPS tracks continuing along the path didn’t fall off the cliff, whereas the shorter ones … … no matter what the map shows. In that sense, GPS tracks reflect the OTG situation. Wouldn’t it be better to map this path according to the GPS tracks, rather than according to some perfect but inaccessible reference?
A bit more practical: If all aerials/reliable other sources disagree with the GPS tracks (assuming many tracks that are consistent, etc), should one rather map a way based on the (average?) GPS records or according to the theoretically most reliable other source of information? After all, I would want to know where I am given my GPS position, not where all GPS positions should be instead. I can hardly tell a device “you are wrong, shift there” (please ignore GPS accuracy or image shift issues).
Just to add a bit more information for people who may not be familiar with the problem - if you’ve got a location where any GPS receiver only has the same partial view of the sky, then you may get consistent offsets regardless of how many readings you combine.
To answer the question what I’d normally do is to try and provide the “correct” (in this case, not GPS-indicated) location, but include other information (here perhaps elevation) that indicates “how far up the cliff we’re talking about”. Also the usual other detail (“surface” et al) beyond “a path drawn from a basic GPS trace”.
Thanks for the extra clarification. Do you put additional infos in the description (if not regular tags)? In any case, while in the field I still would have to find out that there is additional information and that I need it - might have gone over the edge by then .
Do you have a source for this claim? All the information I can find suggests that while accuracy of any one reading will be reduced, there will be no systematic offset or bias. In other words, you should, with enough readings, be able to average the error out.
In any event, my understanding of the OP’s first message was that the GPS readings were taken from the top of the cliff, where the sky would not necessarily be obscured (unless there happened to be another cliff higher up).
Experience, based on a few visits, somewhere around Matlock Bath in England if I remember correctly.
The device used would likely have been a Garmin eTrex, and would have supported GPS + egnos (waas in Europe). It’s a pretty rare phenomenon - that’s partly why it was noticeable.
If there are no GPS accuracy or image shift issues, how would the two not align…?
I would place the path where it truly is. People won’t fall of the cliff because their GPS signal is off. They will do it for other reasons if they do, and a perfectly aligned GPS position/path won’t prevent that.
Sometimes it’s possible to shift the visual background map. Whether you shift the map or the GPS coordinates has the same effect.
Then you are most likely artificially limiting yourself. Maybe you can find milestones or other remarkable objects to take your position from. If you are certain that you aren’t hovering next to the cliff, you may use that information to get a more precise position relative to the map.
This ties in with the point above and heavily depends on what is considered “inaccessible”. As soon as you add other tools of measurement, you walk into conflicts that basically amount to “which tool will the user have at their disposal” which I don’t think we can nor should have an answer to.
I agree, and that’s why everything is relative (cue General Relativity).
Of course, we generally try to position the objects of the map with the most accurate (somewhat absolute) reading we have, but much more important is that the distances between them and their relative positions are correct, because that is much easier to measure. If you move the pathway around, then that might lead to inconsistencies elsewhere.
So to give you my personal answer to your question
Map relative to the objects that already exist on the map. This would mean that the path should be on the cliff that’s already mapped and not in the water.
If the other objects are mapped in the wrong (relative) position, then it gets complicated. But at best, the GPS coordinates are a hint that something is off; they can’t be the sole basis for correction. Every measurement requires interpretation, as do maps.
Interesting. Given that I assume that you hadn’t surveyed the location in question more accurately with another method, how did you determine that the eTrex had a bias? Was the normal Gaussian distribution of multiple readings truncated/flattened/skewed? In which direction was the bias, toward the cliff or away from the cliff?
Do you have any idea of the specific mechanism that might cause the bias? The only thing I can think of is multi path errors or reflected signals, but I think most receivers should be pretty good at rejecting those.
A systematic GPS shift could ocurr because the local conditions may introduce a GPS error(reflection, satellite visibility). Another situation is that in some places of the world even official maps may not reflect the true coordinates, for lack of data or on purpose.
This is a GPS problem, not a map data’s problem. A map is only responsible for accurately representing reality. It’s their fault for not being able to find where they are.
Applications don’t need GPS to be precise either. They can snap it to the nearest =path correctly.
GNSS can already be improved with SBAS eg WAAS before. L5 frequency for dual-band GPS is not uncommon now.
RTK is more expensive, but not prohibitive and completely unaffordable for enthusiasts. Lack of free coverage, and thus the subscription cost (or a 2nd set to set up your base station instead) is more of a burden.
Thanks for the comments, but the point is whether we want to map according to available information OTG - which I suggest GPS also is - or to theoretically perfect survey data whose quality cannot be known in the field. In practice I would of course use additional information like landscape features.
Just for fun I’ve been recording tracks on my bike commute every day for over a year. (With my phone, so probably not super reliable compared to a dedicated receiver.) So this is like 300 tracks of the same route.
I won’t share the dataset because it would be obvious exactly where I live and work
However, I can say that while in most locations, the average of GPS tracks is spot-on, there are a few places along the route where there is a systematic shift in the GPS positions of say, up to 5-10 m as compared to OSM. (OSM in my city is really accurate due to government data and 7.5 cm-resolution imagery being available.) Usually these are places where there are lots of buildings around.
Also, when I leave the office building where I work and turn on the tracker, it consistently is off about 10 m to the south. There are lots of glass facades around there, so I’m assuming this has something to do with reflections.
Anyway, back to the OP’s question: IMO we should definitely map with the best possible positional accuracy and disregard whatever OTG positioning influences there may be. After all, OSM is not only used OTG by people with GPS receivers.
Definitely, but it is what you would get OTG in the example, even if it is wrong in absolute terms. A GPS would often be good in some place and off 1 km down the road. How would you know that? Setting it to snap to the nearest path can be dangerous and not recommended in not so well mapped regions, and certainly there might be better GPS system, too. My question is what to map: the reality of the information available OTG as GPS data or the reality of the absolute coordinates supplied (or not) by some official survery, whose data are not available to me out there.
True, w/o GPS it is a non-problem. One would just use local features relative to the map, not knowing if any distances are correct in absolute terms (at least I don’t usually measure them).
Nice example! thanks. So anyone with a map app would see the building 10m off its true position.
So that means there is either GPS inaccuracy or image (map) shift.
I have had my GPS track perfectly follow the ortophoto path, but consistently a few meters off. And we have a well-known situation of ortophotos being misaligned by a few meters as well (they now paint alignment markers on the roads to mitigate that).
In reality, either one can be true, and both can be true. If you place more trust in the ortophotos, trace the path according to them. That seems to have been the premise of the question. If you place more trust in the GPS tracks, trace the path according to them.
Maybe I’m reading too much into your statement, but it’s interesting that you say
To some extent that would disqualify that specific data set, because it fails the verification rule. If it cannot be verified OTG, then it can’t be used, even if we otherwise know that it is perfect.
In reality, it gets fuzzy because there are lots of tools with which to verify the map, and there are different margins of error and error sources accordingly.
As such - as someone else already said - it’s not an or, but an and. Out of all available information, we have to find out what is there and where it is. If there is conflicting information, the conflict has to be resolved by getting more data to determine which one is correct. There is no magic silver bullet to decide this.
This comes back to my question regarding what is “available information OTG” and would likely fall under Mapping for a specific use case. I believe this is the answer to your theoretic example. It’s not allowed because travel isn’t the only reason to use OSM.
Right, if you are in the field, one may have a great map but without GPS you would not know your true coordinates unless there is some feature around you that you can identify on the map. Since we have many areas that are less than perfectly mapped on OSM, sometimes next to good ones, somewhere in the wild the user would not know if the map is wrong, things have changed since the mapping or your GPS is off. One would need to research beforehand that GPS is off at that particular position. So OTG one relies on either the map (incl. landmarks) or GPS and wouldn’t know which is true.
My survival instinct would probably rely on other people’s (positive) experience and trust the GPS, that is the OTG data.
Certainly true while in the woods.
From a mapping perspective, I agree with the conflict quoted above: OSM makes a big point for validation on the ground in numerous discussions, but in this case the sentiment seems to be to ignore this requirement and use the official survey data. archie and I recently found out that older orthophotos and an official map in Sweden are misaligned to the newest ones and sometimes even with each other.
As you said: no silver bullet, apparently.
… no matter what the map shows. In that sense, GPS tracks reflect the OTG situation. Wouldn’t it be better to map this path according to the GPS tracks, rather than according to some perfect but inaccessible reference?
