+barefootjeff Posted October 26, 2020 Share Posted October 26, 2020 19 minutes ago, HunterandSamuel said: True. But I'm anal about my coords. I personally would not want my cache to be 30 feet off. One of my hides is a multi with waypoints along a series of waterfalls in a gully with very little view of the sky, so GPS reception in there is pretty poor with an EPE typically around 10 to 15 metres. These were the coordinates I got on four separate days for the final location: The satellite image was of no help since all showed were tree tops, so about all I could do was provide helper photos on the cache page to get searchers into the right area and then give strong hints at the preceeding waypoints. It's a nice waterfall in an interesting area that fitted well with the theme of the cache, so I really wanted to put it there. Quote Link to comment
+HunterandSamuel Posted October 26, 2020 Share Posted October 26, 2020 Just now, barefootjeff said: One of my hides is a multi with waypoints along a series of waterfalls in a gully with very little view of the sky, so GPS reception in there is pretty poor with an EPE typically around 10 to 15 metres. These were the coordinates I got on four separate days for the final location: The satellite image was of no help since all showed were tree tops, so about all I could do was provide helper photos on the cache page to get searchers into the right area and then give strong hints at the preceeding waypoints. It's a nice waterfall in an interesting area that fitted well with the theme of the cache, so I really wanted to put it there. Wonderful. Quote Link to comment
+barefootjeff Posted October 26, 2020 Share Posted October 26, 2020 4 minutes ago, HunterandSamuel said: Wonderful. It's received 11 FPs from 19 finds over the five years it's been there, but it's also had a few DNFs from people who didn't look at the cache page before they went in there and, with no data coverage anywhere in the area, couldn't view the photos. Quote Link to comment
+HunterandSamuel Posted October 26, 2020 Share Posted October 26, 2020 7 minutes ago, barefootjeff said: It's received 11 FPs from 19 finds over the five years it's been there, but it's also had a few DNFs from people who didn't look at the cache page before they went in there and, with no data coverage anywhere in the area, couldn't view the photos. Great. 1 Quote Link to comment
+ecanderson Posted October 26, 2020 Share Posted October 26, 2020 3 hours ago, kunarion said: The OP didn't mention how they are getting coordinates, but did mention "averaging", which is not something to play around with, on the assumption that a bunch of points that are by definition NOT where the cache is, will magically be accurate if averaged. They are the average of a lot of points, and the real way to perform "averaging" is an art and very technical. It's not a button click and done. You are right. The OP didn't mention how they are getting coordinates. So saying that their use of averaging "is not something to play around with" is quite premature. In fact, best practice for placing caches incorporates an averaging scheme. Lack of that sort of tool in the 'official app' is one of my bigger complaints with it. If you aren't using any kind of averaging, and repeating it multiple times, you're probably not getting the best set of coordinates for your own placements. Quote Link to comment
+kunarion Posted October 26, 2020 Share Posted October 26, 2020 (edited) 9 hours ago, ecanderson said: If you aren't using any kind of averaging, and repeating it multiple times, you're probably not getting the best set of coordinates for your own placements. An average of a bunch of loose coordinates is NOT precision, it's an average. If you're playing around with "averaging", you're definitely not getting the best coordinates. The technique requires returning over different times of day, different weather and seasons, each set taken in the same way with great care. If you are not doing it like that, might as well not do "averaging" at all. It's way more work than what you need for Geocaching, and if you're doing it because it's tough to get a signal, the "average" Geocacher will also have trouble getting a signal. Instead of averaging in a tough area, consider a cache hint. Sure it's a popular thing to throw out to new hiders. But I see tons of threads about problems people had getting good coordinates, after they in fact DID "average" the coords, just like the OP in this thread. Something is going wrong, probably adding the extra learning curve of "averaging" to the first difficulty of gathering decent points in the first place. Let the App settle out at the cache point, save a few waypoints, check which one returns you most consistently close, done. No "average" needed. Throwing out all but the closest points means you don't have an average, you have a close point. The GPS itself presents the "average" of millions of samples every time it displays one set of coords. But do a lot of caching before you ever hide one. Look at that App display, what does it show? Get a feel for what it's telling you. If it's always off, if it wanders in distance and direction, beware. My iPhone 8 has been "off" for a couple of years with The Official App, so I don't use it for compass guidance at a cache. If the pointer is all over the place, "averaging" is not the answer. You need a precision device to obtain precision. Edited October 26, 2020 by kunarion 1 Quote Link to comment
+HoochDog Posted October 26, 2020 Share Posted October 26, 2020 (edited) 1 hour ago, kunarion said: An average of a bunch of loose coordinates is NOT precision, it's an average. That's semantics, but it's hard to argue that averaging is not beneficial to improving accuracy. Let's run a simulation as an example. Take a random number. Say 57value = 57 We have a hypothetical GPS device that isn't very accurate. In fact, that device can only tell you the coordinates to value within 50 places. It will return a value randomly anywhere between 7 and 107. Not very accurate at all. Now, let's take 200 random readings of "value" where the results can wildly range from 7 to 107 and average them together. If we run this simulation multiple times (just for fun!)...57.020301980532984 56.6082896193402258.36645537356805 Now, let's take a "precision" device. that gives us coordinates to within 15 feet and grab a single value that we think looks accurate.70 66 52 My personal experience shows GPS averaging gives more accurate coordinates than trying to pick out a single reading. Is it possible that a single reading can be more accurate than the average? Yes, it can, but most of the time, it probably won't be as good as an average of many samples taken. Edited October 26, 2020 by HoochDog formatting Quote Link to comment
+ecanderson Posted October 26, 2020 Share Posted October 26, 2020 1 hour ago, kunarion said: Let the App settle out at the cache point, save a few waypoints, check which one returns you most consistently close, done. No "average" needed. Throwing out all but the closest points means you don't have an average, you have a close point. The GPS itself presents the "average" of millions of samples every time it displays one set of coords. No one suggested "An average of a bunch of loose coordinates" as a solution. I didn't make any blanket statements of that sort. I'm sure that Garmin isn't doing that on their devices, nor phone apps that do real waypoint averaging. We're talking about 2 different averaging techniques here. Quote Link to comment
+ecanderson Posted October 26, 2020 Share Posted October 26, 2020 13 minutes ago, HoochDog said: From personal experience, GPS averaging, by and large will give you far more accurate coordinates than trying to pick out a single reading. Again, we may be talking at crossed purposes here. My observation of how waypoint 'averaging' works in Garmin and phone app code isn't so simplistic. There's a good reason why the device's estimation of the quality of the fix can go DOWN during measurement in addition to going up. It's not just a simple average collected over time. Quote Link to comment
+arisoft Posted October 26, 2020 Share Posted October 26, 2020 55 minutes ago, ecanderson said: My observation of how waypoint 'averaging' works in Garmin and phone app code isn't so simplistic. Could you enlight us by telling what averaging algorithm they use in their devices. Quote Link to comment
+ecanderson Posted October 26, 2020 Share Posted October 26, 2020 1 hour ago, arisoft said: Could you enlight us by telling what averaging algorithm they use in their devices. Wise guy, eh? Would be fun to know exactly, but there is definitely some handling of outlier data, accounting for drift, and smoothing going on. If you observe in the field how Garmin devices respond with their % confidence number during 'Waypoint Averaging", there's some clear indication, especially as regards any drift in the instantaneous computed location. One interesting test is to set up a track and then (to make it easier) take a video of the % confidence and current estimate being reported during the averaging, and then compare it on a timeline to the track that was built. No way to exactly reverse engineer the algorithm, but you do see some interesting correlations. Quote Link to comment
+ecanderson Posted October 26, 2020 Share Posted October 26, 2020 Of course, one shouldn't just perform even THIS function just once. DOP being what it is, even Garmin recommends repeating the procedure 4 to 8 times, "waiting at least 90 minutes between samples." Quote Link to comment
+kunarion Posted October 26, 2020 Share Posted October 26, 2020 (edited) 3 hours ago, HoochDog said: That's semantics, but it's hard to argue that averaging is not beneficial to improving accuracy. Let's run a simulation as an example. Take a random number. Say 57value = 57 We have a hypothetical GPS device that isn't very accurate. In fact, that device can only tell you the coordinates to value within 50 places. It will return a value randomly anywhere between 7 and 107. Not very accurate at all. Now, let's take 200 random readings of "value" where the results can wildly range from 7 to 107 and average them together. If we run this simulation multiple times (just for fun!)...57.020301980532984 56.6082896193402258.36645537356805 Now, let's take a "precision" device. that gives us coordinates to within 15 feet and grab a single value that we think looks accurate.70 66 52 My personal experience shows GPS averaging gives more accurate coordinates than trying to pick out a single reading. Is it possible that a single reading can be more accurate than the average? Yes, it can, but most of the time, it probably won't be as good as an average of many samples taken. That’s not what the OP got. The coords are 30 feet off. That's what it averaged out to. That's not the worst I've seen for coordinates. But if the OP is comparing distance anyway, the OP might as pick a couple of the closest saved waypoints and see which is the best. Saves the OP from doing a lot of "averaging" for average coordinates. Edited October 26, 2020 by kunarion 1 Quote Link to comment
+kunarion Posted October 26, 2020 Share Posted October 26, 2020 12 minutes ago, ecanderson said: Of course, one shouldn't just perform even THIS function just once. DOP being what it is, even Garmin recommends repeating the procedure 4 to 8 times, "waiting at least 90 minutes between samples." Take a few waypoint snapshots, choose a close one and test it. The result is about the same as averaging, without all the averaging. The OP may be better off picking a point than going through all that averaging, to end up with results that are similar with any method. Especially since with the OP, averaging isn’t working out. Quote Link to comment
+ecanderson Posted October 26, 2020 Share Posted October 26, 2020 @kunarion The amount of effort required to get a good coordinate depends upon a lot of factors, including the physical location being checked. Standing out in the middle of an open field in the middle of Kansas when the DOP is good really isn't much of a challenge. There are, however, LOTS of more difficult areas than that, and to get really good coordinates, it takes a bit more effort. Note what the OP actually said in the original post: "I downloaded a gps coordinate app for cache placement and did the averaging trick... popped the resulting coord into Google maps and was off by probably 30 feet. Another coord from the same method was almost spot on." What we don't know is what phone app was being used, nor the quality of the algorithm being used by that app. But then Garmin doesn't suggest that taking an average ONCE is a good idea, either. The OP made a discovery in that regard. Why should that be a surprise? There's "best practice" and "what the heck, close enough". You can choose. Meanwhile, in many areas, certainly in mine, (back to the original thread), GE is a good 2nd check. Quote Link to comment
+kunarion Posted October 26, 2020 Share Posted October 26, 2020 (edited) 18 minutes ago, ecanderson said: What we don't know is what phone app was being used, nor the quality of the algorithm being used by that app. But then Garmin doesn't suggest that taking an average ONCE is a good idea, either. The OP made a discovery in that regard. Why should that be a surprise? There's "best practice" and "what the heck, close enough". You can choose. Meanwhile, in many areas, certainly in mine, (back to the original thread), GE is a good 2nd check. That's why I didn't go into the various averaging Apps. I'm GUESSING T. If you take a few points, and one is consistently real close, you have your cache coords already. I'm guessing the OP used a phone. If it's an iPhone with the glitch I have, there's the problem. But you may have to trust an averaging App in that case. If finders log some coord suggestions, maybe consider changing the cords. It's really popular around here to suggest "averaging" without the specifics of all the work that averaging requires. If what the heck close enough is in fact close enough, it saves a lot of work "averaging". And then I don't have to explain "did you try again in 90 minutes that nobody told you, and now do even more that you weren't told about". Instead, you take a few waypoints, and see which one is closest. Edited October 26, 2020 by kunarion 1 Quote Link to comment
+arisoft Posted October 26, 2020 Share Posted October 26, 2020 29 minutes ago, ecanderson said: Would be fun to know exactly, but there is definitely some handling of outlier data, accounting for drift, and smoothing going on. You may be right about how Garmin bypass some functions when averaging. For example, I have noticed sometimes, how the position depends on the direction I am approaching the GZ. Mobile Phone Apps have limited control over this kind of functions. I guess that an averaging app in a mobile phone just calculates the average of current coordinates received from the operating system and nothing else. When I do averaging, I move a lot. For example, many turns around the GZ to get rid of smoothing effect etc. and the average is the center of the circular track. Quote Link to comment
+ecanderson Posted October 26, 2020 Share Posted October 26, 2020 21 minutes ago, kunarion said: It's really popular around here to suggest "averaging" without the specifics of all the work that averaging requires. If what the heck close enough is in fact close enough, it saves a lot of work "averaging". Perhaps because it's been explained SO many times before? This isn't a new topic. And if your "If what the heck" implies a big IF. If you get lucky, then yes, you get lucky. If not, you may annoy a lot of future finders. Quote Link to comment
+cerberus1 Posted October 26, 2020 Share Posted October 26, 2020 57 minutes ago, kunarion said: That's why I didn't go into the various averaging Apps. I'm GUESSING T. If you take a few points, and one is consistently real close, you have your cache coords already. I'm guessing the OP used a phone. If it's an iPhone with the glitch I have, there's the problem. But you may have to trust an averaging App in that case. If finders log some coord suggestions, maybe consider changing the cords. It's really popular around here to suggest "averaging" without the specifics of all the work that averaging requires. Yep. Unfortunately the OP never came back. Most of us remember how long-discontinued, and even some later phones used in this hobby kept the forums busy. The problem with clicking "averaging" for a location is if one's receiving poor signal (for any number of reasons), they're just averaging bad data. We still go back to a spot a few times, rather than press "average" while there with new folks, who ask for help with theirs. I agree that for their first hide, they might have to rely on others to get an idea what their issue is. - There's enough who believe their gizmo is the most accurate on Earth mentioning how far everyone's off in their Found It log. Quote Link to comment
+kunarion Posted October 26, 2020 Share Posted October 26, 2020 35 minutes ago, cerberus1 said: Yep. Unfortunately the OP never came back. Most of us remember how long-discontinued, and even some later phones used in this hobby kept the forums busy. The problem with clicking "averaging" for a location is if one's receiving poor signal (for any number of reasons), they're just averaging bad data. We still go back to a spot a few times, rather than press "average" while there with new folks, who ask for help with theirs. I agree that for their first hide, they might have to rely on others to get an idea what their issue is. - There's enough who believe their gizmo is the most accurate on Earth mentioning how far everyone's off in their Found It log. It also occurs to me that if you’re doing the full-blown averaging process, you must trust the output when you get done. If you’re comparing results as the OP did, the whole averaging thing was not necessary. At best, it’s the same as if you’d simply picked the closest saved waypoint snapshots and tested to see which one’s the most consistently close. But without the averaging part. Quote Link to comment
+kunarion Posted October 26, 2020 Share Posted October 26, 2020 (edited) 48 minutes ago, cerberus1 said: - There's enough who believe their gizmo is the most accurate on Earth mentioning how far everyone's off in their Found It log. Worst case scenario, logs will mention a pretty consistent offset, and even some coords to try. I rarely see COs update it. But I can sometimes find it anyway, by using someone’s suggested coords. I have a phone that is in fact not accurate enough for Geocaching (for walking guidance to a cache via the compass). So I know they aren’t all suitable. In my case, I think it’s just a relatively recent bug in the official App that affects certain iPhones. But it’s a thing for sure. Edited October 26, 2020 by kunarion Quote Link to comment
+ecanderson Posted October 26, 2020 Share Posted October 26, 2020 1 hour ago, kunarion said: It also occurs to me that if you’re doing the full-blown averaging process, you must trust the output when you get done. If you’re comparing results as the OP did, the whole averaging thing was not necessary. Yes, because of course, NOBODY ever fat fingers a set of coordinates. I always compare what I'm entering into a new cache page with a GE just to be sure I didn't do anything stupid. If for no other reason at all, it's a good double check. Quote Link to comment
+ecanderson Posted October 26, 2020 Share Posted October 26, 2020 1 hour ago, kunarion said: Worst case scenario, logs will mention a pretty consistent offset, and even some coords to try. I rarely see COs update it. But I can sometimes find it anyway, by using someone’s suggested coords. We have to do that frequently here for certain CO's caches. One of our locals who is great at being VERY persistent for an FTF has provided us coordinates in his logs as much as 100' from GZ (yes, GZ is where something blows up, not where it's aimed), and for that, we are always very appreciative. That said, it would nice for our FTF hounds if they weren't searching half a planet for certain CO's caches all the time. Until someone does post alternates, everyone is being treated to the un-fun. Somewhere over in the "irks" thread, I know we've had people talk about COs who don't take a hint from the logs and recheck and update coordinates when they see this sort of thing. I was surprised when GCHQ removed the tick box and input field to readily note alternate coordinates, the location in the log entry was always up top and predictable that way -- but most of us are still pretty diligent about noting them somewhere in our log entries if they're far enough out, or the site requires it to avoid needle/haystack situations. Quote Link to comment
+Harry Dolphin Posted October 26, 2020 Share Posted October 26, 2020 25 minutes ago, ecanderson said: We have to do that frequently here for certain CO's caches. One of our locals who is great at being VERY persistent for an FTF has provided us coordinates in his logs as much as 100' from GZ (yes, GZ is where something blows up, not where it's aimed), and for that, we are always very appreciative. A local cacher used his IPhone for coords. They were always off. 30' or more. I had FTF on one that was 210' off! (Great hint...) He used my coords. And cachers are finding it. He has since used his GPSr, and new caches have good coords. 1 1 Quote Link to comment
Blue Square Thing Posted October 27, 2020 Share Posted October 27, 2020 12 hours ago, Harry Dolphin said: A local cacher used his IPhone for coords. They were always off. 30' or more. I had FTF on one that was 210' off! (Great hint...) He used my coords. And cachers are finding it. He has since used his GPSr, and new caches have good coords. Unless it's a very old phone (pre-iPhone 4), my gut feeling is that this is a human being error rather than one caused by using a phone per se. As ecanderson suggests, it always seems to be the same people who get the major errors. Quote Link to comment
+kunarion Posted October 27, 2020 Share Posted October 27, 2020 (edited) 4 hours ago, Blue Square Thing said: Unless it's a very old phone (pre-iPhone 4), my gut feeling is that this is a human being error rather than one caused by using a phone per se. As ecanderson suggests, it always seems to be the same people who get the major errors. A problem I see often with cachers and coordinates, is trying to shoe-horn one format into another. The best method for getting great coords is to set the phone down for a couple minutes and let it settle, and then make a note of the displayed GPS location. Unless you're in a ravine or a cave or something, you're fine. But it's also a great idea to see if your coords consistently return you to the same spot. But my iPhone 8 developed a very loose caching ability (there are several unaddressed bug threads about the issues), it can be 250 feet off or more in any direction at random. I don't know for sure it's only on The Official Apps (GC and AL), but it affects those for sure. Anyway, it's not just me. I'd bet GC lost potential paying customers over this, we'll never know. Someday I might post videos that demonstrate these chronic issues. But I'm not personally all that interested in fixing it. Sue me. Edited October 27, 2020 by kunarion Quote Link to comment
+ecanderson Posted October 27, 2020 Share Posted October 27, 2020 53 minutes ago, kunarion said: But it's also a great idea to see if your coords consistently return you to the same spot. Within what time frame? When people mention the need for doing this at different times on different days, they're trying to explain the problems that can occur with satellite constellations. Feeling lucky? We are presently in a really odd situation in our area where for several weeks, varying between the hours of about 9am and noon, the DOP for GPS will be less than optimal. I don't recall seeing the constellation do something this odd and also consistent over such a long period of time. Use this tool https://satpredictor2.deere.com/ and punch in 80501 (mine) for location, assume a non-GLONASS (GPS only) device, and observe what happens when you move forward from today (10/27) all the way into November. Only if GLONASS is added do things look better. If you want a chuckle, select for GLONASS only, and you can see why it's best to have more than their service available! Looks awful by itself. Quote Link to comment
+kunarion Posted October 27, 2020 Share Posted October 27, 2020 10 minutes ago, ecanderson said: We are presently in a really odd situation in our area where for several weeks, varying between the hours of about 9am and noon, the DOP for GPS will be less than optimal. I don't recall seeing the constellation do something this odd and also consistent over such a long period of time. Use this tool https://satpredictor2.deere.com/ and punch in 80501 (mine) for location, assume a non-GLONASS (GPS only) device, and observe what happens when you move forward from today (10/27) all the way into November. Only if GLONASS is added do things look better. If you want a chuckle, select for GLONASS only, and you can see why it's best to have more than their service available! Looks awful by itself. Sounds good! I didn't have GLONASS turned on lately. So I'm trying that, and I turned on WAAS for now. I haven't seen any particular issues on my Oregon 750. But it makes sense that more satellites could help. Quote Link to comment
+ecanderson Posted October 27, 2020 Share Posted October 27, 2020 Depending upon your exact location and time, all kinds of weird things can happen. GLONASS is barely adequate by itself, but as you could see from that chart of my area - be sure to toss in your own location and try it out some more! - GLONASS can definitely fill in some gaps in a very helpful way. I still consult that site (my favorite ephemeris crystal ball) when placing a cache, avoiding any hours that look less than primo, and when possible, using times that look optimal. It saves time in the long run getting best possible coordinates. Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.