+julianh Posted May 21, 2008 Share Posted May 21, 2008 While I have repeatedly given assurances that I have conducted tests, unfortunately, these forums do not allow attaching of images or other files, so I have not been able to provide the documentary evidence. Well, I have now taken the plunge, and subscribed to a public file hosting service (hope I don’t get deluged with spam, etc!), so I can now provide the evidence which can be perused at your leisure! In my first test, I set my GPSr up at the same fixed location on two consecutive nights (at my home), so the visible GPS satellite constellation is essentially the same on both traces. I am using a Garmin Summit HC, with latest software. On the first night, I recorded GPS elevation, and on the second night, I recorded the auto-calibrated barometric elevation. The location has "reasonable to good" satellite visibility - partially blocked to the East by my house, and some partial tree cover to the North and West, but mainly close to the horizon. Overhead sky visibility is generally good. (I was attempting to simulate "real world" observations, not "best possible" conditions.) The true elevation is 50 metres above sea level. The comparison of elevation plots may be seen here: http://www.mediafire.com/imageview.php?qui...gix&thumb=4 GPS elevation on the first night varied from -10 m to 80.4 m (90.4 m range!), with an average of 47.4 m (2.6 m average error), and a standard deviation of 9.2 m. On the subsequent night, auto-calibrated barometric elevation varied from 24.2 m to 58.3 m (34.1 m range), with an average of 48.1 m (1.9 m average error), and a standard deviation of 7.7 m. The auto-calibrated barometric elevation is clearly superior to the raw GPS elevation. The GPS elevation trace shows sudden spikes from time to time, presumably as satellites drop in and out of visibility. The barometric elevation trace is much more stable, and responds only slowly to the effect of poor satellite visibility giving a poor GPS elevation. There is a period of very poor GPS elevation at about 4:00 am to 4:30 am - presumably, I had a very poor visible satellite constellation for this period – but, hey – that happens out in the real world too! This also affects the auto-calibrated barometer elevation, but there is a significant time lag, and the maximum error in the barometric elevation is a lot less. In "typical real world" conditions, I generally get significantly better performance than this, because even though the instantaneous satellite visibility can vary a lot as you move in and out of tree cover etc, the GPSr can typically track all satellites currently visible, even though individual satellites may drift in and out of view periodically, whereas at my fixed location, some satellites were totally blocked for significant periods (or subject to significant multi-path errors) due to my fixed location with partial blockage. In my second set of tests, I repeated the same sort of exercise, but this time on three consecutive days at my office, which is in a window seat in a mid-level floor in the “high-rise canyon”. This test can be considered to be a “worst-case” scenario – unless you are travelling deep in canyons or very heavy forest cover, you would rarely see such extended periods of poor satellite coverage. The first day was a full office day of recording GPS elevations, while on the following two days, I recorded auto-calibrated barometric elevation. Again, these tests were conducted on consecutive days, so the satellite constellation can be considered to be essentially identical in each case. I don’t know the exact elevation of my office window sill, but it is about 23 m above sea level as near as I can tell (ground level is 13 m; 3 stories is approximately an additional 10 metres). Comparative plots are here: http://www.mediafire.com/imageview.php?qui...3pd&thumb=4 For this test, GPS elevation varied from -21.5 m to 141 m (162.5 m range!!!!), with an average of 36.0 m (13 m estimated average error), and a standard deviation of 34.8 m. The first day of auto-calibrated barometric elevation varied from 1.6 m to 31.4 m (29.8 m range), with an average of 13.6 m (9.4 m estimated average error), and a standard deviation of 8.7 m. The second day of auto-calibrated barometric elevation varied from 6.4 m to 78.0 m (71.6 m range), with an average of 42.7 m (19.7 m estimated average error), and a standard deviation of 16.5 m. Once again, the auto-calibrated barometric elevation is clearly more stable and reliable than the GPS elevation. Should you want to conduct your own statistic analysis of my logs, you can play with the data in this spreadsheet: http://www.mediafire.com/?glcsxdq1lcg In summary, the auto-calibrated barometric elevation is clearly more stable and more accurate than the raw GPS elevation. When your 2D and / or 3D GPS position accuracy is less than ideal (limited satellite visibility, multi-path errors, etc), the GPS elevation accuracy tends to suffer particularly badly. However, because the Garmin algorithm seems to only use longer term trends in GPS elevation to auto-calibrate the barometer, this short-term effect does not seem to carry over to the same extent to the auto-calibrated barometric elevation accuracy. I suspect the latest Garmin software revisions improve the auto-calibration algorithm. In particular, I think my old B&W Vista used to appear to over-respond to fluctuations in the GPS elevation, whereas my Summit HC with current software appears to be quite heavily "damped" in this respect, so it seems to respond to significant trends in GPS elevation, but not to short term changes (e.g. when you walk into a location with poor satellite visibility, so you get sudden apparent GPS elevation shifts due to blockage and / or multi-path errors). That is, by a combination of using barometric pressure for short-term elevation changes, underpinned by longer-term auto-calibration using "damped" GPS elevation, the unit gives me excellent behaviour all day. Hope this helps anyone who has been wondering whether auto-calibrated barometric altimeters can really work! Quote Link to comment
Forkeye Posted May 21, 2008 Share Posted May 21, 2008 Thanks ,I like that kind of research.I find Garmin products have alot of gadgets that don't really work properly.If you want a good altimeter / barometer,get a Thommen ,they are temperature compensated,and used by mountaineers all over the world,and if you need a compass that works get a Suunto,Silva or Bruton,not a Garmin.I wonder if all of Garmins high end marine products have as many issues as their hand held line does? Quote Link to comment
roybassist Posted May 21, 2008 Share Posted May 21, 2008 Thanks ,I like that kind of research.I find Garmin products have alot of gadgets that don't really work properly.That is a rather odd comment, considering that Julian’s experiment shows that a Garmin auto calibrated barometric altimeter provides more stable and accurate elevation data than elevation determined by GPS alone. That is one of the purposes of auto calibration. Julian’s experiment shows that it does work properly. If you want a good altimeter / barometer,get a Thommen ,they are temperature compensated,and used by mountaineers all over the worldBut isn’t a Thommen just an ordinary barometric altimeter? Even typical changes in barometric pressure due to weather can cause such an altimeter to be out of calibration by many tens of meters in few days. The other purpose of auto calibration of Garmin’s barometric altimeters is to compensate for changes in barometric pressure due to weather. In that situation, a Garmin auto calibrated barometric altimeter will give more accurate results than an ordinary barometric altimeter. The fact that Julian’s barometric altimeter varied from the actual elevation during his test is not indicative of a problem with it. An ordinary barometric altimeter might have varied much more during that time, due to changes in weather. if you need a compass that works get a Suunto,Silva or Bruton,not a Garmin. The electronic magnetic compass in any GPS is a convenience feature, not a substitute for a mechanical compass with a magnetic needle. Anyone in a situation where a compass might be critical should of course carry a mechanical compass. I always carry one as a backup. But with the built-in magnetic compass in my GPS, I can stand in one spot and have the unit point toward a waypoint that I’ve set. Without a built-in compass, I have to read from the GPS the bearing to the waypoint, pull out my mechanical compass, set the bearing, sight it in the field, then put the compass away again. It isn’t a big deal, but the built-in magnetic compass in the GPS makes the operation faster and less bother. Those who have units without a built-in magnetic compass often say that they can just walk a few steps and get the direction, and that’s true; but there are some situations where walking a few steps can be anything from difficult to hazardous. And even if you can walk those few steps, it still isn’t as easy and convenient as just being able to point the receiver. That’s the advantage of the built-in magnetic compass. I wonder if all of Garmins high end marine products have as many issues as their hand held line does? Mostly the issues with handheld units are with newly introduced products. The ones that have been around a while, like the 60 and 76 series are generally very reliable. I suspect that a lot of the problems people attribute to the magnetic compass needing to be calibrated are actually either misunderstanding or operator error. Quote Link to comment
PerryB2 Posted May 21, 2008 Share Posted May 21, 2008 I’ll add a little to julianh’s experiments. I’ve compared the elevation profile of a gps without auto-calibrated barometric elevation to a 60CSx with. Particularly when reception is not great the gps only elevation fluctuates a lot more. To get the best elevation profile from the 60CSx it helps to recalibrate the altimeter if the gps has been turned off for sometime…because barometric pressure may well have changed due to weather. The 60CSx will adjust for the change but it takes quite while. Quote Link to comment
+julianh Posted May 21, 2008 Author Share Posted May 21, 2008 I’ll add a little to julianh’s experiments. I’ve compared the elevation profile of a gps without auto-calibrated barometric elevation to a 60CSx with. Particularly when reception is not great the gps only elevation fluctuates a lot more. To get the best elevation profile from the 60CSx it helps to recalibrate the altimeter if the gps has been turned off for sometime…because barometric pressure may well have changed due to weather. The 60CSx will adjust for the change but it takes quite while. I probably should have mentioned (as I have done repeatedly in other threads in these forums) that I ALWAYS manually calibrate my barometric altimeter at the start of each day - then I generally maintain 5-metre vertical accuracy for the whole day. Best option for manual calibration is if you know your altitude at the start of the day, but if not, get a good 3D lock, and calibrate against the GPS elevation. Even though that can be out by 20 metres (or more, as these tests show!), it is probably better than no initial calibration at all, and the unit will generally correct for a 20-metre initial mis-calibration in 15 minutes or so anyway. My tests were conducted with the barometric altimeter calibrated manually at the start of the session in each case. Quote Link to comment
+julianh Posted May 22, 2008 Author Share Posted May 22, 2008 Some additional barometric pressure data for those who might question the impact of changes in barometric pressure over the duration of my tests: My tests at home were conducted on the nights of 31 March to 1 April 2008 (GPS elevation), and 1 April to 2 April 2008 (barometric elevation). My tests at work were conducted on 31 March 2008 (GPS elevation), and 1 April and 2 April 2008 (barometric elevation). I have posted Bureau of Meteorology records (9:00 am and 3:00 pm summary data) here http://www.mediafire.com/?emwzfdnwnmr and here http://www.mediafire.com/?jgmjg20lg2m . As yet, I have note been able to access more detailed hour-by-hour records of barometric pressure, but I will post it if I can locate it. Weather throughout the test period was fine, with no rain (this is shown on the BoM records), and no warm or cold fronts passed through (as can be seen from the temperature data), so I would not expect there to have been any sudden departures of pressure from the overall trends. The total pressure range over the test period was about 10 hPa (10 millibar), which would result in an apparent barometric elevation shift of about 90 metres on an uncorrected barometric altimeter. When used properly, Garmin units with auto-calibrated barometric altimeters do MUCH better than this, as my tests show. Quote Link to comment
+julianh Posted May 22, 2008 Author Share Posted May 22, 2008 For those who might be interested in how I recorded the GPS elevation and barometric elevation data, it's pretty straightforward with my Garmin Summit HC. I first set the track-log to record every 30 seconds (instead of using "automatic" mode, which won't record multiple trackpoints while you are stationary). To get GPS elevation, just set the altimeter to "fixed elevation" mode. This effectively switches the barometric altimeter off, so the tracklog records GPS elevation instead of barometric elevation. For barometric elevation, you switch the altimeter to "variable elevation" mode (which is how you normally use it). This means that the unit records barometric elevation into the track-log, instead of GPS elevation. At the end of each test, I then downloaded the active tracklog to my computer, which gives me a set of 3D trackpoints, all at the same location, at 30 second intervals, all time and date stamped. I don't think there is any easy way of recording both GPS elevation and barometric elevation simultaneously, but it might be possible using real-time logging of the output NMEA data (or something like that) - but this is beyond my technical capability at the moment. Quote Link to comment
Phoenix2001 Posted May 22, 2008 Share Posted May 22, 2008 Normally the GPS calibrated barometric altimeter smooths out the elevation data as your tests demonstrate. However, gusty winds can cause rapid pressure changes on the barometric sensor resulting in fluctuations in the elevation data. Changing the direction of the receiver within the airflow (sensor opening into the wind then rotate 180 deg. so that opening is away from the wind) can exacerbate the fluctuations. On a day with gusty winds, the GPS elevation can be more stable than the output from the barometric altimeter especially if WAAS (or equivalent - EGNOS) is available. Quote Link to comment
+julianh Posted May 22, 2008 Author Share Posted May 22, 2008 Normally the GPS calibrated barometric altimeter smooths out the elevation data as your tests demonstrate. However, gusty winds can cause rapid pressure changes on the barometric sensor resulting in fluctuations in the elevation data. Changing the direction of the receiver within the airflow (sensor opening into the wind then rotate 180 deg. so that opening is away from the wind) can exacerbate the fluctuations. On a day with gusty winds, the GPS elevation can be more stable than the output from the barometric altimeter especially if WAAS (or equivalent - EGNOS) is available. I am fully aware of this effect - but my observations indicate it is nowhere near as drastic as suggested in normal outdoor conditions (at least, the sort of conditions that I consider "normal"). I have reported elsewhere in these forums that when using the altimeter in a moving car (60 km/hr to 100 km/hr), I have seen spurious elevation changes of the order of 5 metres, maybe 10 metres at 100 km/hr. As my experiments show, and user's observations will confirm, GPS elevation accuracy is only good to about 20 metres under even ideal conditions. (Your mileage may vary - it has a lot to do with the car's cabin configuration etc, but it gives an idea of the sort of relationship between wind speed and elevation shift that you might expect.) In "normal" handheld outdoor use, with "normal" wind speeds, I have not seen the effect at all on my Summit HC. I am sure you would see something if you are in an exposed conditions with high sustained wind speeds. I guess it depends on what you mean by "gusty" - might be unreliable to take your elevation during a passing cyclone! (But then again, if I happened to be "lucky" enough to have my GPSr with me during a cyclone strike, I would probably switch it to barometer mode to watch the massive air pressure drop!) Quote Link to comment
+julianh Posted May 23, 2008 Author Share Posted May 23, 2008 (edited) A bit more on the possible impact of wind speed on barometric elevation: The free stream dynamic pressure for moving air is given by: q = 1/2 rho v^2 where rho is the density of air (generally taken as 1.2 kg/m^3 near sea level), v is the velocity in m/s, and the pressure is calculated in pascals. The pressure actually felt at a given location is given by: p = Cp q where Cp is the Pressure Coefficient, and depends on the shape of the object (i.e. you, holding your GPSr), and which face we are considering (i.e. windward, side on, or leeward). Cp can typically range in the order of plus or minus 0.8, say. Near sea level, the ambient pressure drops by approximately 12 Pa per metre rise (or conversely, each 1 hPa of change in air pressure is equivalent to about 8.3 metres of elevation change). Using the above relationships, we can deduce the likely potential for elevation error when the barometric altimeter is exposed to significant winds. Fear not! I have done this for you! A plot of the relationship between wind speed and potential barometric elevation error is here: http://www.mediafire.com/imageview.php?qui...mxh&thumb=4 You can see that the potential error due to wind speed is less than 10 metres for wind speeds of less than 50 km/hr or so. (And my observations in moving cars back this up.) My calculations are here (Excel spreadsheet) so you can check the methodology for yourself, if you want: http://www.mediafire.com/?nzvyby2m4t0 Note that these calculations are not trying to predict the ACTUAL error, just to give you some sense of the potential for error. If you face into the wind,the GPSr (windward side) is likely to be in an area of high pressure, so the barometric elevation reading may drop. If you stand side on or back to the wind, the GPSr is likely to be in a zone of reduced pressure, and the barometric elevation reading is likely to rise. Edited May 23, 2008 by julianh Quote Link to comment
+julianh Posted May 25, 2008 Author Share Posted May 25, 2008 (edited) Okay – this is provided for Team CowboyPapa (and other sceptics). Here is some real-world test data. On Sunday 25 May 2008, I conducted the following tests driving around a circuit around Mount Coot-Tha (Brisbane), being a convenient location to test GPS elevation vs. barometric elevation. I ran three loops around the mountain, the first using auto-calibrated barometric altimeter, the second using GPS elevation, and the third again using auto-calibrated barometric elevation. I am using a Garmin Summit HC (Software Version 2.70, GPS SW Version 2.60). I had WAAS turned on, and was picking up a WAAS satellite and getting differential corrections reported throughout the tests, although the benefit of this is debatable, as Brisbane is outside the range of the ground station network, so no ionosphere corrections are available. However, recent observations suggest that late-model Garmins MAY correct for ephemeris and clock errors, while ignoring ionosphere corrections when outside the ground station coverage, so some benefit MAY be achieved nonetheless. 1. Turned on GPSr at 12:50 pm at my home, waited for 3D GPS fix, then calibrated barometric altimeter at 50 m above sea level (being the true elevation of my home “calibration point”), altimeter set to “auto-calibrate” mode. GPS elevation was showing as 53 m; i.e. 3 m error (8 satellites, 5 of which showed differential corrections, plus satellite 50 – WAAS). 2. Drove to public park at J.C. Slaughter Falls (near base of Mt Coot-Tha. (N.B. this section of the published track has been deleted for privacy reasons.) 3. Marked waypoint “Bench” at a picnic table at 1:00 pm, as a convenient reference point for the tests. Barometric elevation found to be 68 m. This elevation is consistent with SunMap 1:25,000 topographic maps (5 metre contour interval), and also with my previous visits to this location. GPS elevation reported as 69 m (9 satellites, 5 of which showed differential corrections, plus satellite 50 – WAAS). I.e. at this point, GPS elevation and barometric elevation agreed to within 1 metre – so far so good! 4. Headed off in anti-clockwise direction around the mountain circuit. Stopped at the main public lookout, and marked waypoint “Lookout” at 1:24 pm. Barometric altimeter showed 228 m; GPS elevation showed 225 m (9 satellites, 5 of which showed differential corrections, plus satellite 50 – WAAS). Deviation now 3 m. 5. Continued anticlockwise circuit until I got back to “Bench” at 1:41 pm. Barometric altimeter again showed 68 m (identical to first visit), but GPS elevation now showed 54 m – i.e. 15 m variation from previous visit (9 satellites, 7 of which showed differential corrections, plus satellite 50 – WAAS). 6. At 1:43 pm, set altimeter to “fixed elevation” mode, so altimeter would track GPS elevation instead of barometric elevation. Set off on a repeat of the circuit. 7. At 1:51 pm, returned to “Lookout”. GPS elevation showed 230 m – i.e. 5 m variation from previous GPS elevation at same location, but only 2 m from barometric elevation (which I consider to be more reliable anyway). Barometric elevation not available. (9 satellites, 8 of which showed differential corrections, plus satellite 42 instead of 50 – WAAS). 8. At 2:06 pm, returned to “Bench” once more. GPS elevation now showing 66 m; barometric elevation not available. (9 satellites, 7 of which showed differential corrections, plus satellite 42 – WAAS). 9. Turned barometric altimeter back on, and re-calibrated back to 68 m before proceeding for my third and final loop. 10. At 2:20 pm, returned to “Lookout”. Barometric elevation showed as 226 m – i.e. 2 m variation since first visit at 1:24 pm. GPS elevation showed 225 m (10 satellites, 8 of which showed differential corrections, plus satellite 42 – WAAS). 11. At 2:30 pm, returned to “Bench” for the last time. Barometric elevation was 67 m; GPS elevation was 58 m (9 satellites, 7 of which showed differential corrections, plus satellite 42 – WAAS). 12. Returned home at 2:40 pm. (NB: this leg of my trip deleted from the published log for privacy reasons.) Barometric altimeter showed 48 m (2 m deviation for start of trip); GPS elevation showed 47 metres. (10 satellites, 8 of which showed differential corrections, plus satellite 42 – WAAS). Summary: At Home: Barometric Elevation: 50 m / 48 m (2 m variation in 2 visits over 2 hours) GPS Elevation: 53 m / 47 m (6 m variation in 2 visits over 2 hours) At “Bench”: Barometric Elevation: 68 m / 68 m / N.A. / 67 m (1 m variation in 3 visits over 1 1/2 hours) GPS Elevation: 69 m / 54 m / 66 m / 58 m (15 m variation in 4 visits over 1 1/2 hours) At “Lookout”: Barometric Elevation: 228 m / N.A. / 226 m (2 m variation in 2 visits over 1 hour) GPS Elevation: 225 m / 230 m / 225 m (5 m variation in 3 visits over 1 hour) I know which record of elevations I find more consistent and reliable! For most users, GPS elevation may be fine (most users may not care about their elevation at all!), but if you need a reliable elevation track, with approximately 5 metres accuracy – get a Garmin with auto-calibrating barometric altimeter! Record of my tracks and waypoints is here, for anyone who wants to see the evidence: Map Source database format: http://www.mediafire.com/?x9anzjwzxjq (You may need Garmin-compatible maps of Australia to make sense of this in Map Source.) Google Earth format: http://www.mediafire.com/?jojkwo1tjxi Copy of Bureau of Meteorology data (including barometric pressure – in case you want to check for actual changes in barometric pressure) is here: http://www.mediafire.com/?gj5xmumwc0o Barometric pressure varied by a range of approximately 0.7 hPa over the duration of my tests, which corresponds to about 6 metres of barometric elevation variation, if not compensated for. The fact that I did not get this much variation of barometric elevation at ANY of my test sites indicates that the auto-calibration algorithm IS correcting for changes in barometric pressure. Hope this provides the last nail in the coffin of the argument that auto-calibrated barometric altimeters can’t work! Edited May 25, 2008 by julianh Quote Link to comment
Hertzog Posted May 25, 2008 Share Posted May 25, 2008 I don't think there is any easy way of recording both GPS elevation and barometric elevation simultaneously, but it might be possible using real-time logging of the output NMEA data (or something like that) - but this is beyond my technical capability at the moment. In the fixed elevation mode the elevation data field will still display the barometric elevation, so you could manually record it during a test. As you say, not an "easy" way to do it. However, another approach you can use is to connect to nRoute during your tests. nRoute always logs the GPS elevation, so you can operate normally and end up with simultaneous track files of both elevations. Should work great at home or office; but not too good for field trials. (Sorry I haven't mentioned this before; I've been following your posts with interest, but been too preoccupied with other matters to really get in and digest them. These are the types of tests I've liked to do myself, but haven't done them with your rigor.) Quote Link to comment
jmundinger Posted May 25, 2008 Share Posted May 25, 2008 (edited) While I have repeatedly given assurances that I have conducted tests, unfortunately, these forums do not allow attaching of images or other files, so I have not been able to provide the documentary evidence. Are you sure about that? In order to post an image, it first has to be hosted some place, like imagecave, photobucket etc. or on your ISP's server. Download the image from your gpsr to your computer - if it's a garmin, you can use the ximage utility, downloaded from from Garmin's website. Then upload the image to a hosting site and then copy the link to the image into your post. For it to work, you also have to use the appropriate have use html. For posting, that would be one set of brackets, [], with the letters img typed between that set of brackets. Then, paste the link to the image. Then type a second set of brackets, [], with the letters /img typed between that set. You can see what it looks like by hitting the reply button to my post. Edited May 25, 2008 by jmundinger Quote Link to comment
+julianh Posted May 25, 2008 Author Share Posted May 25, 2008 (edited) While I have repeatedly given assurances that I have conducted tests, unfortunately, these forums do not allow attaching of images or other files, so I have not been able to provide the documentary evidence. Are you sure about that? OK - I should have said "I don't know how to post images "! Getting them up on a public hosting service was a necessary first step, it seems, as you can't embed an image directly in a post. In future, I should be able to manually add the HTML code so as to embed the image as you have done - thanks for the tip! (P.S. I tried to go back to my earliest posts in this thread to make the changes so the images would display - but editing is disabled for me on the first few posts for some reason - maybe there is a time limit since first posting? Anyway, next time ... Have added a couple of embedded images to my latest post, which still permits me to edit it. Thanks again!) Edited May 25, 2008 by julianh Quote Link to comment
jmundinger Posted May 25, 2008 Share Posted May 25, 2008 Getting them up on a public hosting service was a necessary first step, it seems, as you can't embed an image directly in a post. In future, I should be able to manually add the HTML code so as to embed the image as you have done - thanks for the tip! You're welcome. Correct. The picture has to be hosted someplace that this site can access - which you tell it to do with the html and the url to the site. In the case of your graph, all I did was right click on the image, copied the image location and then pasted that between the ] and the [ in the image html. By the way, thanks for sharing your analysis. Quote Link to comment
+julianh Posted May 25, 2008 Author Share Posted May 25, 2008 I don't think there is any easy way of recording both GPS elevation and barometric elevation simultaneously, but it might be possible using real-time logging of the output NMEA data (or something like that) - but this is beyond my technical capability at the moment. In the fixed elevation mode the elevation data field will still display the barometric elevation, so you could manually record it during a test. As you say, not an "easy" way to do it. However, another approach you can use is to connect to nRoute during your tests. nRoute always logs the GPS elevation, so you can operate normally and end up with simultaneous track files of both elevations. Should work great at home or office; but not too good for field trials. (Sorry I haven't mentioned this before; I've been following your posts with interest, but been too preoccupied with other matters to really get in and digest them. These are the types of tests I've liked to do myself, but haven't done them with your rigor.) Hertzog, Good comments. 1. I'm not sure what happens with auto-calibration of the displayed barometric elevation if the altimeter is in "fixed elevation" mode - does it still auto-calibrate, or does it effectively lock barometric pressure to a fixed value? It's making my head spin just thinking about it! The only reasons I have ever turned the altimeter to "fixed elevation" mode is to track barometric pressure during a storm event or similar (which is really only for personal interest, not for my navigation purposes, but might be important for someone hiking overnight in the mountains), and to conduct these tests. Otherwise, my altimeter is always in variable elevation mode, with auto-calibration turned on. 2. I wasn't aware of the fact that the real-time USB output alway logs GPS elevation instead of barometric elevation, but if so, yes, that would work, as you would then get two concurrent elevation records - GPS elevation with nRoute, barometric elevation on the GPSr itself. (I guess the NMEA output sentences or whatever they are called must specifically output GPS data rather than any other data that the GPSr has on-board?) Anyway, this wouldn't really be an option for the tests I have been conducting. Firstly, my laptop is not waterproof or even condensation proof, so I can't leave it outdoors overnight. (Overnight tests were conducted with GPSr outdoors for good satellite visibility, to model real-world use.) Secondly, I was a worried enough when conducting my circuit tests on Mt Coot-Tha - returning to the same couple of spots every half an hour or so, GPSr in hand, waiting for people to move so I could get to my exact waypoints, then taking scribbled notes, then coming back 30 minutes later to repeat it. I suspect in these days of heightened security, if I tried this armed with a laptop computer as well, the Tactical Response Group would probably have me pinned to the ground in seconds! Quote Link to comment
Hertzog Posted May 26, 2008 Share Posted May 26, 2008 (edited) 1. I'm not sure what happens with auto-calibration of the displayed barometric elevation if the altimeter is in "fixed elevation" mode - does it still auto-calibrate, or does it effectively lock barometric pressure to a fixed value? I've had a few headaches thinking about this as well. Here's my take: 1. The fixed/variable elevation setting doesn't affect the barometric elevation calculations at all; they go on independent of this setting. 2. The fixed/variable setting affects how the barometric pressure gets calculated from the ambient pressure and the barometric elevation. In the variable elevation setting it seems pretty evident that it is taking the ambient pressure and and changing it by the current barometric elevation reading, based on the standard pressure altitude curve (or maybe some digital approximation to it) to get the barometric pressure. However, in the fixed elevation mode it looks to me like it is reverting to the last manually calibrated elevation, or possibly the elevation you would have had at the time of the switch if you hadn't been in autocal; in any case, it doesn't look like it is using either the autocaled barometric elevation or the GPS elevation at the time of switch for it's barometric pressure calculations in the fixed elevation mode. OK, now MY head is spinning Edited May 26, 2008 by Hertzog Quote Link to comment
+julianh Posted May 26, 2008 Author Share Posted May 26, 2008 (edited) Some additional barometric pressure data for those who might question the impact of changes in barometric pressure over the duration of my tests: My tests at home were conducted on the nights of 31 March to 1 April 2008 (GPS elevation), and 1 April to 2 April 2008 (barometric elevation). My tests at work were conducted on 31 March 2008 (GPS elevation), and 1 April and 2 April 2008 (barometric elevation). I have posted Bureau of Meteorology records (9:00 am and 3:00 pm summary data) here http://www.mediafire.com/?emwzfdnwnmr and here http://www.mediafire.com/?jgmjg20lg2m . As yet, I have note been able to access more detailed hour-by-hour records of barometric pressure, but I will post it if I can locate it. Weather throughout the test period was fine, with no rain (this is shown on the BoM records), and no warm or cold fronts passed through (as can be seen from the temperature data), so I would not expect there to have been any sudden departures of pressure from the overall trends. The total pressure range over the test period was about 10 hPa (10 millibar), which would result in an apparent barometric elevation shift of about 90 metres on an uncorrected barometric altimeter. When used properly, Garmin units with auto-calibrated barometric altimeters do MUCH better than this, as my tests show. The Bureau of Meteorology has been kind enough to send me 15-minute records of barometric pressure for Brisbane at the time of my tests: Pressure range on the night of 1 April / 2 April was approximately 1020 hPa (low - around 3:00 am) to 1022.4 hPa (high - around 8:00 pm). 2.4 hPa range would correspond to about 20 metres of variation on an uncalibrated barometric altimeter. Pressure range on the day of 1 April was approximately 1020 hPa (low - around 2:00 pm) to 1022 hPa (high - around 10:00 am). 2 hPa range corresponds to about 17 metres of error on an uncalibrated barometric altimeter. Pressure range on the day of 2 April was approximately 1017 hPa (low - around 5:00 pm) to 1021 hPa (high - around 9:00 am). That is, on this test session, there was a significant drop of 5 hPa range, corresponding to about 40 metres of error on an uncalibrated barometric altimeter. Edited May 26, 2008 by julianh Quote Link to comment
roybassist Posted May 27, 2008 Share Posted May 27, 2008 I've just returned from a 570-mile round trip (285 miles each way) on which I gathered some data to add to this thread. The trip was made in about 5 and a half hours each way, with one day in between. I carried with me a weather radio to receive broadcasts from the National Weather Service stations along my route. In general, I was able to receive one about every 30 miles or so. I had my 76CSx running with Auto Calibration turned on. My car windows were closed, but outside air was coming in via the ventilation system. I started the trip each way with my barometric altimeter calibrated to the reported barometric pressure at my departure point. Along the route, I received the latest hourly barometric pressure observations from the NWS stations. Most of them were between 25 and 35 miles from my route, about the maximum range of the weather radio. I compared the reported hourly barometric readings with the barometer reading from my 76CSx. The difference was never more than 0.02”. On the outbound trip the barometric pressure didn’t vary by much during the entire trip; but on the return leg, it started at 29.82” and rose gradually to 29.99 at the end of my trip. Even though I traveled 285 miles over 5 hours 21 minutes, and the barometric pressure reported by NWS changed by 0.17”, Auto Calibration kept the barometric pressure displayed on the unit within 0.02” of the hourly reported values. I have determined by repeated observations that up to 0.03” of difference is fairly common between reporting stations that are about 25 miles apart. Given that the stations I could receive were almost all more than 25 miles away from my route, a difference of 0.02” between the reported readings and the barometric pressure reported on my unit is very good agreement. At the end of my return trip, where I was much closer to the reporting station, my reading was 30.00” compared to the reported 29.99” Clearly, Auto Calibration kept the unit’s barometric pressure reading very close to the actual, varying barometric pressure. A difference of 0.02” corresponds to about 20 feet of elevation. For an altimeter without a means of calibration, the change of 0.17" in barometric pressure would have resulted in elevation readings being off by about 170 feet. Even though I was in a car with outside air coming in, this condition was relatively steady-state. The amount of air flow didn’t vary much over the trip. Given such a steady-state condition, Auto Calibration is able to compensate and keep the barometric sensor in calibration. Obviously, if I were to stop and get out of the vehicle, the relatively quick change in pressure resulting from that would be interpreted by the sensor as a change in elevation. My experience has been that if the ventilation fan is on its highest setting, going from the mildly pressurized interior of the vehicle to the ambient pressure outside typically results in a maximum variation of about 25 feet of elevation. Auto Calibration can adjust out a small difference like that in about 20 minutes. Quote Link to comment
+julianh Posted May 27, 2008 Author Share Posted May 27, 2008 I've just returned from a 570-mile round trip (285 miles each way) on which I gathered some data to add to this thread. The trip was made in about 5 and a half hours each way, with one day in between. Thanks, Roy - more good data. Your tests show pretty conclusively that the auto-calibrate algorithm can adjust the unit's estimate of sea level barometric pressure over long distances and long time intervals, and with significant actual barometric pressure changes. This auto-correction for sea level barometric pressure is the key - this enables the barometric altimeter to make an accurate calculation of your actual elevation. Your tests show that the unit can maintain an elevation accuracy of about 20 feet or better, and this is consistent with my own tests and observations of about 5 metres typically. This is a LOT better than can be achieved by consumer-grade GPS elevation alone (as my earlier tests show). Your observations on the potential error due to air pressure in a moving vehicle also agree with mine. This doesn't actually worry me, as these units are intended to be used for hand-held walking use, etc, so I can live with the potential error in a fast-moving car - it's a simple case of the physics, and there's not much we can do about that! As you say, any such error is corrected for in a few minutes anyway, once you get out of the car. (In any case, it's STILL likely to be more accurate than GPS elevation alone.) I wouldn't recommend using a handheld GPSr with auto-calibrated barometric altimeter to attempt an instrument landing in an aircraft, however! Thanks again! Quote Link to comment
SiliconFiend Posted May 27, 2008 Share Posted May 27, 2008 I've just returned from a 570-mile round trip (285 miles each way) on which I gathered some data to add to this thread. The trip was made in about 5 and a half hours each way, with one day in between. Thanks, Roy - more good data. Not to puncture anyone's balloon, but Roy's test didn't exercise the auto-calibration of anything. If I'm reading correctly, he was simply monitoring the pressure reported by his GPS and comparing it to the pressure broadcast by the weather service. All he proved is that the pressure sensor has good accuracy. That's (very) important and we do appreciate the data collection, but that's not where the auto-calibration comes in. The auto-calibration takes into account the elevation calculated from the GPS signals in order to correct for changes in pressure due to weather which would otherwise show up as elevation changes. Julianh's test is a good exercise of the auto-calibration function. Quote Link to comment
+julianh Posted May 27, 2008 Author Share Posted May 27, 2008 I've just returned from a 570-mile round trip (285 miles each way) on which I gathered some data to add to this thread. The trip was made in about 5 and a half hours each way, with one day in between. Thanks, Roy - more good data. Not to puncture anyone's balloon, but Roy's test didn't exercise the auto-calibration of anything. If I'm reading correctly, he was simply monitoring the pressure reported by his GPS and comparing it to the pressure broadcast by the weather service. All he proved is that the pressure sensor has good accuracy. That's (very) important and we do appreciate the data collection, but that's not where the auto-calibration comes in. The auto-calibration takes into account the elevation calculated from the GPS signals in order to correct for changes in pressure due to weather which would otherwise show up as elevation changes. Julianh's test is a good exercise of the auto-calibration function. With all due respect, SiliconFiend - I think you're wrong. The actual instrument inside the GPSr measures the Ambient Pressure - i.e. the absolute pressure at your current location. An altimeter has no way of directly knowing the sea level barometric pressure, only the ambient pressure. However, you need to know the sea level barometer pressure in order to be able to determine an accurate elevation. On my Summit HC, you can view two separate pressure figures - "Ambient Pressure" (which the instrument measures directly - really only of academic interest to the average user), and "Barometer Pressure" (which you can enter when you are doing a manual calibration, and which the instrument periodically deduces and updates when applying the auto-correction). "Barometer Pressure" means the sea level pressure, and is the figure you can check against your local Met Bureau data, so this number has some relevance to a casual user. If auto-calibration isn't working, then the Barometer Pressure figure (i.e. sea level pressure) won't stay in sync with the true barometer pressure - it will be affected by changes in your elevation AND changes in barometer pressure. The fact that Roy's GPSr gave a consistently accurate figure for Barometer Pressure (i.e. sea level pressure) as he travelled for several hours, while the barometer pressure was changing, and presumably as his elevation was changing (I assume this trip was not across the plains of Kansas! *) proves that the Garmin auto-calibration routine DOES correct the sea-level Barometer Pressure, which is a necessary step to be able to use the directly measured ambient pressure to determine your current elevation. By using auto-calibration to maintain an accurate real-time estimate of Barometer Pressure, the unit then simply takes the difference of the Barometer Pressure and the Ambient Pressure to get your elevation. (Without an accurate real-time figure for sea-level Barometer Pressure, the calculated elevation would also be inaccurate - which is precisely the problem with any barometric altimeter which must be manually re-calibrated.) Hope this makes sense! * By the way, scientists have proved that it is incorrect to describe Kansas as being "as flat as a pancake" - it is actually FLATTER than a pancake: http://www2.ljworld.com/news/2003/jul/27/h...hotcakes_study/ Quote Link to comment
Hertzog Posted May 27, 2008 Share Posted May 27, 2008 (edited) Not to puncture anyone's balloon, but Roy's test didn't exercise the auto-calibration of anything. If I'm reading correctly, he was simply monitoring the pressure reported by his GPS and comparing it to the pressure broadcast by the weather service. All he proved is that the pressure sensor has good accuracy. That's (very) important and we do appreciate the data collection, but that's not where the auto-calibration comes in. The auto-calibration takes into account the elevation calculated from the GPS signals in order to correct for changes in pressure due to weather which would otherwise show up as elevation changes. Julianh's test is a good exercise of the auto-calibration function. No, he mentioned that on his return trip the reported barometric pressure changed by 0.17 in Hg, but with autocal the 76 calculated barometric pressures to within 0.02 in of the reported values; without autocal the GPS would have assumed that the 0.17 in change was due to altitude changes, and the barometric pressures calculated by the 76 would have changed little if any. Oops, sorry julian, when I wrote this your more detailed response had not shown up on my computer; anyway, consider it a seconding of your response. Edited May 27, 2008 by Hertzog Quote Link to comment
toddm Posted May 27, 2008 Share Posted May 27, 2008 I won't argue one way or another, however I can say that in in places that are prone to large pressure changes say in the mountains where weather and pressure changes quickly that I've found the auto calibrated barometric elevation profiles to be less accurate than GPS elevations in most cases. That said I've seen places where the GPS reported elevation is horribly from surveying benchmarks as well. That said, would not a better test of the auto calibration be to say start a day hike or drive at a survey benchmark and then return to the benchmark or hike to another benchmark and compare the reported elevations? Or set up a route with mapsource topo hike/drive the route and then compare elevation profiles to the topo profile for the route/track? Something that takes more than a hour or two to complete so larger changes in pressure can be encountered. We can easily see a 15-20hpa change in a day here, and that's without heading up into the mountains. It would be interesting to see how well the auto calibration could deal with that type of pressure change while moving and changing elevation as well. Is it not probable that part of the garmin's auto calibration takes into account if the unit is moving......if the unit is stationary it's easy to adjust out pressure changes because the unit is not going anywhere. I would guess that type of correction is much harder when the unit is changing in elevation and position, and pressure changes occur. That would seem to be a more real world test than leaving it on a desk all day recording data. Quote Link to comment
+julianh Posted May 27, 2008 Author Share Posted May 27, 2008 (edited) I won't argue one way or another, however I can say that in in places that are prone to large pressure changes say in the mountains where weather and pressure changes quickly that I've found the auto calibrated barometric elevation profiles to be less accurate than GPS elevations in most cases. That said I've seen places where the GPS reported elevation is horribly from surveying benchmarks as well. As noted above, the instantaneous reading of GPS elevation on a consumer GPSr is only reliable to about 20 metres at best; 50 metres or more is not uncommon in sub-optimal reception conditions. (Long-term averages over several hours at a fixed location will generally be reasonably close to true elevation, but I am talking about the real-time on-the-spot reading that you get from your hand-held GPSr as you move around.) That said, would not a better test of the auto calibration be to say start a day hike or drive at a survey benchmark and then return to the benchmark or hike to another benchmark and compare the reported elevations? Or set up a route with mapsource topo hike/drive the route and then compare elevation profiles to the topo profile for the route/track? Something that takes more than a hour or two to complete so larger changes in pressure can be encountered. We can easily see a 15-20hpa change in a day here, and that's without heading up into the mountains. It would be interesting to see how well the auto calibration could deal with that type of pressure change while moving and changing elevation as well. I would gladly do a test with a benchmark - if I could find one in my area (Brisbane, Australia). However, I don't know how to find published benchmarks with elevation in my areas. My Mt-Coot-Tha test was an attempt to overcome this limitation by doing a circuit repeatedly, logging both GPS and barometric elevation. I have 1:25,000 topo maps of the area, with 5 m contour intervals. My track logs conform well with the topo map data; however, the variation I experienced between the accuracy of the GPS elevation trace and the barometric elevation trace isn't really enough to prove that my barometric elevation trace "fits" the topo maps "better" than the GPS elevation. However, the auto-calibrated barometric elevation is noticeably less variable than GPS elevation when returning to the same point - and I think that really proves the point. I have had enough experience with my Summit HC to know that this behaviour is also typical for longer records / greater elevation changes / greater barometer pressure changes. Yes, a controlled benchmark would be a better test, though. Is it not probable that part of the Garmin's auto calibration takes into account if the unit is moving......if the unit is stationary it's easy to adjust out pressure changes because the unit is not going anywhere. I would guess that type of correction is much harder when the unit is changing in elevation and position, and pressure changes occur. That would seem to be a more real world test than leaving it on a desk all day recording data. I don't know the details of Garmin's algorithm - it is trade secret, and Garmin aren't saying. However, my guess is that the auto-calibration algorithm is optimised for auto-calibration of a moving GPSr, because that is what the unit is basically designed for - and my Mt Coot-Tha test, and roybassist's test show that auto-calibration IS effective for a moving GPSr. I don't know whether the algorithm specifically changes when stationary, but my guess is that it does not - because if it did, I would expect the auto-calibrated barometric elevation traces from my static tests to be even less variable than they were. Hope this helps! Edited May 27, 2008 by julianh Quote Link to comment
Bradtholomew Posted May 27, 2008 Share Posted May 27, 2008 I was very happy with the accuracy of my Garmin's altimeter. I only have the Legend HCx, but it did a good job on this hike a few weeks ago in the Adirondacks. This is the profile of my first day hike - you can see the plateau where I set up camp around 800m, and the top of the mountain I climbed. That was Mt. Colvin - 1235m height according to the topo map. The actual height shown by the Garmin was 1236-1237m - basically dead on with the actual. I feel a little bad about posting about that trip on here, since I did no geocaching whatsoever heh. Quote Link to comment
+julianh Posted May 27, 2008 Author Share Posted May 27, 2008 The actual height shown by the Garmin was 1236-1237m - basically dead on with the actual. I feel a little bad about posting about that trip on here, since I did no geocaching whatsoever heh. Nice real-world test - I'm guessing that this log represents several hours of data, with over 800 m of elevation change, and in all probability, significant wind speeds and barometric pressure changes. 2 m accuracy should satisfy the needs of most recreational users! Is it too late to go back and place your own geocache at the top of the mountain? Quote Link to comment
roybassist Posted May 27, 2008 Share Posted May 27, 2008 The fact that Roy's GPSr gave a consistently accurate figure for Barometer Pressure (i.e. sea level pressure) as he travelled for several hours, while the barometer pressure was changing, and presumably as his elevation was changing (I assume this trip was not across the plains of Kansas! *) Well, actually… My trip was from Denver, Colorado to Norton, Kansas. Norton is located in western Kansas, and the trip is largely across the plains of eastern Colorado and western Kansas. However, the elevation at my starting and ending point in Denver is about 5500 feet, and the elevation in Norton is about 2300 feet. So there is an elevation difference of about 3200 feet between them. And as for the area being flat: there are sections that are flat for miles; but there are other portions where, if you were riding it on a bicycle, you would probably have harsh words for anyone who had told you to expect it to be flat! There are some very hilly areas along the way. Quote Link to comment
SiliconFiend Posted May 27, 2008 Share Posted May 27, 2008 I won't argue one way or another, however I can say that in in places that are prone to large pressure changes say in the mountains where weather and pressure changes quickly that I've found the auto calibrated barometric elevation profiles to be less accurate than GPS elevations in most cases. That said I've seen places where the GPS reported elevation is horribly from surveying benchmarks as well. As noted above, the instantaneous reading of GPS elevation on a consumer GPSr is only reliable to about 20 metres at best; 50 metres or more is not uncommon in sub-optimal reception conditions. (Long-term averages over several hours at a fixed location will generally be reasonably close to true elevation, but I am talking about the real-time on-the-spot reading that you get from your hand-held GPSr as you move around.) That said, would not a better test of the auto calibration be to say start a day hike or drive at a survey benchmark and then return to the benchmark or hike to another benchmark and compare the reported elevations? Or set up a route with mapsource topo hike/drive the route and then compare elevation profiles to the topo profile for the route/track? Something that takes more than a hour or two to complete so larger changes in pressure can be encountered. We can easily see a 15-20hpa change in a day here, and that's without heading up into the mountains. It would be interesting to see how well the auto calibration could deal with that type of pressure change while moving and changing elevation as well. I would gladly do a test with a benchmark - if I could find one in my area (Brisbane, Australia). However, I don't know how to find published benchmarks with elevation in my areas. My Mt-Coot-Tha test was an attempt to overcome this limitation by doing a circuit repeatedly, logging both GPS and barometric elevation. I have 1:25,000 topo maps of the area, with 5 m contour intervals. My track logs conform well with the topo map data; however, the variation I experienced between the accuracy of the GPS elevation trace and the barometric elevation trace isn't really enough to prove that my barometric elevation trace "fits" the topo maps "better" than the GPS elevation. However, the auto-calibrated barometric elevation is noticeably less variable than GPS elevation when returning to the same point - and I think that really proves the point. I have had enough experience with my Summit HC to know that this behaviour is also typical for longer records / greater elevation changes / greater barometer pressure changes. Yes, a controlled benchmark would be a better test, though. Is it not probable that part of the Garmin's auto calibration takes into account if the unit is moving......if the unit is stationary it's easy to adjust out pressure changes because the unit is not going anywhere. I would guess that type of correction is much harder when the unit is changing in elevation and position, and pressure changes occur. That would seem to be a more real world test than leaving it on a desk all day recording data. I don't know the details of Garmin's algorithm - it is trade secret, and Garmin aren't saying. However, my guess is that the auto-calibration algorithm is optimised for auto-calibration of a moving GPSr, because that is what the unit is basically designed for - and my Mt Coot-Tha test, and roybassist's test show that auto-calibration IS effective for a moving GPSr. I don't know whether the algorithm specifically changes when stationary, but my guess is that it does not - because if it did, I would expect the auto-calibrated barometric elevation traces from my static tests to be even less variable than they were. Hope this helps! Thanks. I was misunderstanding the barometric pressure as the absolute pressure. Quote Link to comment
SiliconFiend Posted May 27, 2008 Share Posted May 27, 2008 The actual height shown by the Garmin was 1236-1237m - basically dead on with the actual. I feel a little bad about posting about that trip on here, since I did no geocaching whatsoever heh. Nice real-world test - I'm guessing that this log represents several hours of data, with over 800 m of elevation change, and in all probability, significant wind speeds and barometric pressure changes. 2 m accuracy should satisfy the needs of most recreational users! Is it too late to go back and place your own geocache at the top of the mountain? Oops. Now it's your turn to be wrong. He has the Legend HCx, which does not have a barometric altimeter. His excellent accuracy was with only GPS elevation available. Quote Link to comment
+fourbeer Posted May 27, 2008 Share Posted May 27, 2008 I did a three hour mountain bike ride this weekend. I have auto calibration turned on. The course was out and back with about 4,000 feet of elevation gain/loss. The finish elevation was 112 feet higher than the starting elevation. Did this happen because I did not calibrate my starting elevation and the auto calibration "fixed" this as I was riding or is this expected error for this amount of elevation change? You can see the data here: http://trail.motionbased.com/trail/activity/5763206# Quote Link to comment
+julianh Posted May 27, 2008 Author Share Posted May 27, 2008 Oops. Now it's your turn to be wrong. He has the Legend HCx, which does not have a barometric altimeter. His excellent accuracy was with only GPS elevation available. Oops - yes, I missed that - mea culpa! Anyway, it is true that GPS elevation can be spot on sometimes. It is also true that it can be out by twenty metres or more, even under ideal satellite reception conditions. The problem is that a casual GPSr user has know way of knowing whether their GPS is reading high, low, or spot on, nor even what the potential error is, until they happen to reach a location with known elevation. In general, the better the reported EPE displayed by the GPSr (Estimated Position Error), the better the GPS elevation accuracy is likely to be - but even this is not always the case. You will get the best possible 2D and 3D accuracy with the best satellite coverage - and the top of a mountain is likely to be as good as it gets! Optimal 2D accuracy is provided when you have a good spread of satellites across the sky, including a few close to the horizon (but not TOO close - otherwise atmospheric errors can balloon out!), and a few overhead. A satellite constellation that looks like this will give almost 360 degrees coverage for a 2D fix. However, GPS elevation is at best only about half as accurate as the 2D fix, because all of the satellites that are below you are obscured by the Earth. That is, the best possible spread of satellites for an elevation fix approaches only 180 degrees, not 360 degrees. When you are on top of a mountain, or on a flat treeless plain, you may approach this potential accuracy IF the current satellite constellation actually covers the whole sky, horizon to horizon and overhead. When the visible satellites are clustered into a smaller part of the sky (either because that is where they happen to be right now, or more particularly, whenever you are in a location with sub-optimal visibility, such as in hilly terrain, under dense foliage, etc), then the accuracy will fall off, and the elevation accuracy is more sensitive to such factors than the 2D fix. Survey-quality and professional GPSr units will provide more information to an expert user to enable them to know more about their real-time accuracy (including VDOP etc), but this information is not provided on consumer GPSrs, and most users wouldn't know what to do with this information if it was. Quote Link to comment
+julianh Posted May 27, 2008 Author Share Posted May 27, 2008 I did a three hour mountain bike ride this weekend. I have auto calibration turned on. The course was out and back with about 4,000 feet of elevation gain/loss. The finish elevation was 112 feet higher than the starting elevation. Did this happen because I did not calibrate my starting elevation and the auto calibration "fixed" this as I was riding or is this expected error for this amount of elevation change? You can see the data here: http://trail.motionbased.com/trail/activity/5763206# You didn't say what GPSr you are using. My comments relate to current model eTrex range in particular, with current software. I am not sure how other units perform, but I would certainly recommend that you make sure you have current software loaded - my Summit HC is a LOT better in tracking barometric elevation than my older B&W Vista (not the HCx). Yes, if you don't perform an initial calibration, the unit starts off assuming that the barometric pressure is whatever it last saw. Daily air pressure changes can certainly account for errors of the order of 50 metres or more, especially if a major weather event has passed through in the interim. The unit will self-correct for such initial calibration errors within half an hour or so, once you have a decent 3D GPS fix. Your elevation from that point on is generally pretty good. One other point - I had a look at your log, and followed the link to the Wiki page on elevation accuracy. http://wiki.motionbased.com/mb/GPS_Unit_Elevation MotionBased Gravity uses the horizontal position reported from a GPS and cross-references the position to an elevation database that has been surveyed by professional organizations. The known position is accurate to as little at 3 meters which can be a good source of elevation. It seems the "MotionBased" site is doing some post-processing of its own, so I am not sure how this interacts with what the GPSr is doing "on board". Again, my comments all relate to the performance of current eTrex units, using the built-in auto-calibration, and no post-processing. Hope this helps! Quote Link to comment
roybassist Posted May 27, 2008 Share Posted May 27, 2008 I did a three hour mountain bike ride this weekend. I have auto calibration turned on. The course was out and back with about 4,000 feet of elevation gain/loss. The finish elevation was 112 feet higher than the starting elevation. Did this happen because I did not calibrate my starting elevation and the auto calibration "fixed" this as I was riding or is this expected error for this amount of elevation change? The elevation data on the linked site shows Start Elevation 2,536 ft and Finish Elevation 2,648 ft. National Geo Topo! Backroads Explorer gives the elevation of your start/finish location as about 2,613 ft. I don’t know how accurate the elevation data in Backroads Explorer is, but your finish elevation is obviously a lot closer to the map elevation than your start elevation. Assuming that the map elevation is at least close to correct, this would indicate that your uncalibrated starting elevation was considerably in error, and that Auto Calibration brought it much closer to the correct value during your ride. Quote Link to comment
roybassist Posted May 28, 2008 Share Posted May 28, 2008 I can say that in in places that are prone to large pressure changes say in the mountains where weather and pressure changes quickly that I've found the auto calibrated barometric elevation profiles to be less accurate than GPS elevations in most cases.I would be interested in knowing more details about this. How much data did you collect and how? Under what conditions? How did you determine the correct elevation? Is it not probable that part of the garmin's auto calibration takes into account if the unit is moving On the contrary, that’s exactly what it does; which is why it seems like voodoo to some. if the unit is stationary it's easy to adjust out pressure changes because the unit is not going anywhere. I would guess that type of correction is much harder when the unit is changing in elevation and position, and pressure changes occur. That would seem to be a more real world test than leaving it on a desk all day recording data.That’s why I gathered the data that I did. Over 5 and a half hours, 285 miles, a 3,200-ft elevation gain, and an increase in barometric pressure of 0.17”, auto calibration kept the unit calibrated to within 0.02”. Of course it would have been better to compare the readings with known elevations, but that wasn’t feasible on this trip. However, since the unit can be calibrated by inputting the barometric pressure, the pressure can also be used as an indicator of how well calibrated the unit is. Quote Link to comment
+Team CowboyPapa Posted May 30, 2008 Share Posted May 30, 2008 OK, all this evidence is somewhat overwhelming and difficult for me to assess. So, please humor me by entertaining a simple example. For the example, let's make the assumption that the ambient pressure is invariant with respect to horizontal position. With this: 1. The initial ambient pressure (with zero temperature offset) at sea level is defined by this standard: http://www.digitaldutch.com/atmoscalc/ 2. During a climb to 5,000 feet, the ambient pressure at sea level falls due to weather the equivalent of 600 feet. 3. One would expect a raw barometric altitude to be 600 feet at sea level and about 5,600 feet at 5,000 feet true elevation after this weather event. What is the auto calibrated altitude at a true altitude of 5,000 feet? And, if that is too simple, what would be the result for the reverse case whereby the ambient pressure at sea level rose by the equivalent of 600 feet during a climb to 5,000 feet? Quote Link to comment
Phoenix2001 Posted May 31, 2008 Share Posted May 31, 2008 Here’s an example of the effect of gusty (gust: a sudden, strong rush of air or wind) winds on the barometric altimeter. I don’t mean breezy (lower wind speeds) conditions or steady winds. The peak speeds are above 30 km/hr (19 mph). The graphs below are from track logs; recording at once per second, while standing at two different points for almost three minutes during a bike ride. The GPSmap 76Cx and eTrex Vista HCx were mounted on the bike handlebars. The GPSmap 76CSx was connected to an external antenna on top of the helmet and the receiver was in a holster on my belt. The ride was in the South Shore area of Lake Pueblo, Colorado. It could be described as on top of a mesa since it is mostly grassland and flat on top with a clear view in all directions (except for me in the case of the two receivers on the handlebars). Cliffs drop down to the reservoir. At the time of the recordings, the wind was gusty. For brief moments, the wind could be fairly still. Strong gusts were very noticeable, perhaps in the 50 km/hr (31 mph) vicinity, give or take around 10 km/hr (6 mph). Here's some data that shows that the barometric altimeter stabilizes the elevation under calmer conditions: Results for a eTrex Vista HCx (barometric altimeter slowly calibrated by GPS elevation) Test runs are indoors so there is no wind but satellite reception has interference: Points are recorded at one per second. Meters (Feet) Run Number of Pts, Mean Elevation, Max. Elev., Min. Elev., Range, Standard Deviation 8/31/2007 10049 1891 (6204) 1894 (6215) 1886 (6188) 8 (27) 1.73 (5.66) 8/30/2007 10063 1887 (6191) 1891 (6204) 1883 (6177) 8 (27) 2.10 (6.88) 8/31/2007 13053 1890 (6202) 1893 (6212) 1888 (6193) 6 (19) 0.90 (2.96) 9/30/2007 9999 1890 (6202) 1897 (6223) 1886 (6188) 11 (35) 2.15 (7.07) Number of Times Elevation Changed from one second to the next: 8/31/2007 No Change: 4013 By 1 foot: 4774 By 2 feet: 1120 By 3 feet: 127 By 4 feet: 11 By 5 feet: 3 8/30/2007 No Change: 4150 By 1 foot: 4647 By 2 feet: 1124 By 3 feet: 134 By 4 feet: 7 By 5 feet: 0 8/31/2007 No Change: 5465 By 1 foot: 5946 By 2 feet: 1451 By 3 feet: 174 By 4 feet: 16 By 5 feet: 0 9/30/2007 No Change: 5582 By 1 foot: 1795 By 2 feet: 2339 By 3 feet: 238 By 4 feet: 37 By 5 feet: 7 ----------------------------------------------------------------------- Results for a GPSmap 76Cx (no barometric altimeter) First three below were run at the same time as the first three above. Test runs are indoors so satellite reception has interference: Run Number of Pts, Mean Elevation, Max. Elev., Min. Elev., Range, Standard Deviation 8/31/2007 10048 1893 (6211) 1915 (6284) 1858 (6096) 57 (188) 6.96 (22.83) 8/30/2007 10060 1891 (6204) 1914 (6278) 1877 (6159) 36 (119) 5.40 (17.71) 8/31/2007 13055 1891 (6205) 1925 (6317) 1872 (6141) 54 (176) 6.04 (19.82) 8/30/2007 8711 1892 (6208) 1907 (6256) 1882 (6174) 25 (82) 4.75 (15.58) 8/31/2007 No Change: 7110 By 1 foot: 2487 By 2 feet: 246 By 3 feet: 99 By 4 feet: 50 By 5 feet: 32 By 6 feet: 11 By 7 feet: 6 By 8 feet: 3 By 9 feet: 1 10 or more: 2 8/30/2007 No Change: 7685 By 1 foot: 2217 By 2 feet: 102 By 3 feet: 33 By 4 feet: 12 By 5 feet: 6 By 6 feet: 3 By 7 feet: 1 8/31/2007 No Change: 9976 By 1 foot: 2872 By 2 feet: 122 By 3 feet: 31 By 4 feet: 23 By 5 feet: 14 By 6 feet: 4 By 7 feet: 6 By 8 feet: 4 By 9 feet: 1 10 or more: 1 8/30/2007 No Change: 6663 By 1 foot: 1931 By 2 feet: 88 By 3 feet: 21 By 4 feet: 5 By 5 feet: 1 By 6 feet: 0 By 7 feet: 1 Quote Link to comment
+julianh Posted May 31, 2008 Author Share Posted May 31, 2008 (edited) OK, all this evidence is somewhat overwhelming and difficult for me to assess. So, please humor me by entertaining a simple example. For the example, let's make the assumption that the ambient pressure is invariant with respect to horizontal position. With this: 1. The initial ambient pressure (with zero temperature offset) at sea level is defined by this standard: http://www.digitaldutch.com/atmoscalc/ 2. During a climb to 5,000 feet, the ambient pressure at sea level falls due to weather the equivalent of 600 feet. 3. One would expect a raw barometric altitude to be 600 feet at sea level and about 5,600 feet at 5,000 feet true elevation after this weather event. What is the auto calibrated altitude at a true altitude of 5,000 feet? And, if that is too simple, what would be the result for the reverse case whereby the ambient pressure at sea level rose by the equivalent of 600 feet during a climb to 5,000 feet? TCP, Assuming you have set things up properly by first getting a 3D fix, then performing an initial altimeter calibration to sea level at the start of your trip, then keep the GPSr on for the duration of your trip (by foot, horseback or bicycle *) and have auto-calibration on the whole time (i.e. use the unit as intended!), then your Garmin auto-calibrated barometric elevation will read 5,000 feet plus or minus about 5 metres (15 feet) when you get to a true elevation of 5,000 feet. It's that simple! * If you have travelled by car or light aircraft, the error could be rather worse on arrival, due to velocity pressure effects - perhaps of the order of an additional 5 to 10 metres for 100 km/hr / 60 mph, say. However, auto-calibration should correct this error in 15 minutes or so. The altimeter reading will be meaningless if you have travelled in a pressurised aircraft. Your GPS elevation will show an elevation of about 5,000 feet plus or minus an indeterminate error which could be anywhere from plus or minus 10 metres (best practical case with ideal reception) to plus or minus 50 metres or worse (depending on satellite reception and constellation configuration at your location). It won't matter whether: 1) Barometer pressure is invariant with position, or varies with position - it doesn't matter - the unit will compensate either way. (Note that ambient pressure WILL vary with position if the elevation changes, because ambient pressure is the absolute pressure at your location, and this WILL change if you change elevation. Barometer pressure is the sea level pressure, and may or may not be invariant with location). 2) Barometer pressure is steady, or goes up, or goes down, and if it changes, it doesn't matter by how much - the whole point of auto-calibration is to compensate for changes in pressure due to both effects - changes in barometer pressure (due to weather) and changes due to elevation changes. Garmin's altimeters are not temperature-compensated, but I can honestly say I have never seen this to have any significance. (And playing with the link you gave suggests that the standard model doesn't have much temperature effect either.) Hope this helps! Edited May 31, 2008 by julianh Quote Link to comment
+julianh Posted May 31, 2008 Author Share Posted May 31, 2008 Here’s an example of the effect of gusty (gust: a sudden, strong rush of air or wind) winds on the barometric altimeter. I don’t mean breezy (lower wind speeds) conditions or steady winds. The peak speeds are above 30 km/hr (19 mph). The graphs below are from track logs; recording at once per second, while standing at two different points for almost three minutes during a bike ride. The GPSmap 76Cx and eTrex Vista HCx were mounted on the bike handlebars. The GPSmap 76CSx was connected to an external antenna on top of the helmet and the receiver was in a holster on my belt. The ride was in the South Shore area of Lake Pueblo, Colorado. It could be described as on top of a mesa since it is mostly grassland and flat on top with a clear view in all directions (except for me in the case of the two receivers on the handlebars). Cliffs drop down to the reservoir. At the time of the recordings, the wind was gusty. For brief moments, the wind could be fairly still. Strong gusts were very noticeable, perhaps in the 50 km/hr (31 mph) vicinity, give or take around 10 km/hr (6 mph). Here's some data that shows that the barometric altimeter stabilizes the elevation under calmer conditions: ... Phoenix2001, Thanks for the results - another nice set of tests. (Can I suggest that some people just have too many GPSrs? ) You didn't state whether you know the true elevation of your test sites - I would suggest something like 5,010 feet at TP19, and 5,000 feet at TP20 - doe that sound reasonable? At TP19, you get a typical variation of about 10 feet (3 metres) with occasional excursions greater than this, which corresponds to my own expectations. You saw a bit more variability at TP20, suggesting a significant increasing wind speed towards the end of the trace (higher wind speed gives lower pressure which gives higher apparent elevation). Wind speeds are naturally accentuated at sharp changes in terrain (e.g. top of cliffs, edge of a mesa, etc), so I guess this is an effect to look out for. Your static indoor tests show a clearly more stable barometer elevation than GPS elevation. One point to note - your reception indoors will generally be significantly inferior than outdoors, and this will typically impact the elevation accuracy even more than the 2D accuracy. This in turn will also affect the auto-calibrated barometric elevation, because the unit uses the GPS elevation to calibrate the altimeter, but the variation in the barometric altimeter is effectively damped compared to the GPS elevation (i.e. smaller amplitude of variation, and variation lags the changes in GPS elevation), so the accuracy of both elevation figures are affected. In normal use (outdoors), both GPS elevation and barometric elevation will typically be rather better than indoor tests might suggest. Quote Link to comment
Phoenix2001 Posted June 4, 2008 Share Posted June 4, 2008 (edited) You didn't state whether you know the true elevation of your test sites - I would suggest something like 5,010 feet at TP19, and 5,000 feet at TP20 - doe that sound reasonable? [snip] I generated 10 foot contour lines from the NED (Horz: NAD83, Vertical: NAVD88). The two locations are between the 4960 and 4970 foot contours. Since the 4970 line is on the "mesa top", which is fairly "flat", I'll give both locations an estimate of 4968 ft. (favoring the mesa top). The average of all the trackpoints for TP19 is 5,000 and is 4999 for TP20. Three feet was subtracted from the two receivers on the handlebars and six feet from the receiver with the antenna on my helmet to get the ground level elevation. The accuracy and datum of the table that Garmin uses to convert the ellipsoid height to the geoid height is unknown to me. Oh, and yes, I have too many GPSrs! Edited June 4, 2008 by Phoenix2001 Quote Link to comment
+julianh Posted June 4, 2008 Author Share Posted June 4, 2008 I generated 10 foot contour lines from the NED (Horz: NAD83, Vertical: NAVD88). The two locations are between the 4960 and 4970 foot contours. Since the 4970 line is on the "mesa top", which is fairly "flat", I'll give both locations an estimate of 4968 ft. (favoring the mesa top). Phoenix2001, Not doubting your methods (although coming from Australia, I don't know what the "NED" dataset is, although I can hazard a guess), but if you don't have good topo maps for your area, you can get some idea of true elevation using a couple of free tools on the internet: 1: Try Google Maps (maps.google.com), and click the "Terrain" button on the top right of the screen. This will turn on a topo contours layer. (I am guessing this data comes from the NASA SRTM dataset). 2. Try the Garmin MapSource Map Viewer at http://www8.garmin.com/cartography/ and check out "US Topo 2008". In either case, you will need to zoom in to maximum zoom to get the best contour details. Hope this helps! Quote Link to comment
Phoenix2001 Posted June 5, 2008 Share Posted June 5, 2008 Not doubting your methods (although coming from Australia, I don't know what the "NED" dataset is, although I can hazard a guess), but if you don't have good topo maps for your area, you can get some idea of true elevation using a couple of free tools on the internet: 1: Try Google Maps (maps.google.com), and click the "Terrain" button on the top right of the screen. This will turn on a topo contours layer. (I am guessing this data comes from the NASA SRTM dataset). "NED" stands for National Elevation Dataset and is the most accurate available here. It is better than the SRTM data. Quote Link to comment
+julianh Posted June 16, 2008 Author Share Posted June 16, 2008 (edited) As I have said before, the actual Garmin algorithm is a trade secret to which I have no access, but I believe it is possible to infer something about the probable basic goings-on by observing how Garmin sensor units actually respond to changes in barometric (sea level) pressure, changes in elevation, and changes in accuracy of the GPS elevation estimate. The following periodic calibration routine could be carried out using nothing more than a GPS and a manually calibrated barometric altimeter, if you have the time and the patience, but it would need to be done at least once every half hour or better in order to be reasonably accurate. Also, the determination of an appropriate “damping factor”, and calculating the real-time VDOP, may be beyond the average casual user. However, I am sure that it is theoretically possible to use a GPS to periodically re-calibrate a barometric altimeter. (Luckily, Garmin has done all the hard work, and has implemented something like this in their recent units with sensors!) Possible algorithm for auto-calibration: 1. Get current estimate of elevation and barometric (sea level) pressure from the barometric altimeter 2. Get current estimate of GPS elevation 3. Calculate difference between GPS elevation and barometric elevation 4. Get current estimate of VDOP (roughly speaking – estimate of reliability of the GPS elevation) 5. Calculate a “GPS Elevation Weighting Factor” based on the VDOP – use 1.0 for a very small VDOP (i.e. elevation accuracy of better than 5 metres, say); use 0.0 for a very high VDOP (i.e. elevation accuracy of worse than 50 metres, say); interpolate for intermediate values 6. Calculate the change in barometric pressure which would be required to make the barometric elevation agree with the GPS elevation (e.g. if current barometric elevation reads higher than GPS elevation, how much would you need to reduce the current sea level barometer pressure to get the two elevation estimates to agree) – call this “Delta B.P.” 7. Change the estimate of barometer (sea level) pressure by Damping Factor times Delta B.P. times GPS Elevation Weighting Factor 8. Wait one Time Increment 9. Repeat The Time Increment would be any reasonable increment for recalibrating the barometric altimeter. Since air pressure changes due to weather systems etc are relatively slow (say 1 hPa in an hour, or thereabouts), you only need to recalibrate every few minutes, so I would suggest something like 5 to 15 minutes would be reasonable. If the Time Increment is too long, your GPS elevation could jump around a lot between calibration points, and the “noise” in the GPS elevation could swamp the calibration process. (In a Garmin GPSr, the auto-calibration algorithm runs more or less continuously, so the unit must be doing this process every couple of minutes, I think.) The damping factor should be less than 1.0, so the unit responds in a damped fashion to changes in GPS elevation, rather than instantaneously responding to all changes in GPS elevation. If it was set equal to 1.0, the barometric altimeter would potentially reset itself every few minutes to exactly match the GPS elevation, and we know that this is not desirable, because we know the GPS elevation can vary by 20 metres or more, depending on current satellite observation conditions. Calculation of an ideal Damping Factor would be based on experiment, and would also be linked to the chosen Time Increment, but I would suggest a value of 0.5 would be a reasonable first guess. The shorter the Time Increment (i.e. the more auto-calibrations per hour), the smaller the damping factor should be, to ensure the barometric altimeter does not over-respond to fluctuations in GPS elevation which are due to poor accuracy of the GPS elevation rather than actual changes in elevation. I would be very interested to see how this works, if anyone wants to give it a go. Personally, I have neither the patience nor the inclination, because I am comfortable that my Summit HC already does something like this without needing any manual intervention! Note that you would not need to do this in “real-time”. You could set up your GPSr at a fixed location, and record the GPS elevation every 5 minutes or so. Get a value for your barometer (sea level) air pressure at the start of your test from your local Bureau of Meteorology. (Better still, get a complete record of barometric pressure over the course of the test – you will only need the initial figure to calibrate the start of your test, but it will be helpful to compare the Met Bureau records of varying barometer pressure against the predictions made by the algorithm when you post-process the results.) Then copy this record of varying GPS elevation into Excel, and you can do all of these calculations in Excel by simply copying the algorithm formula down the whole column of GPS elevations to predict a new value of barometer (sea level) pressure and barometric elevation for each data point. You might need to experiment with the Time Increment you use for the algorithm and the Damping Factor in order to get a “best fit” (I would bet this is what Garmin has done in refining their algorithm!) Also, I am nor sure of the best way to get predictions of VDOP for each time step in your test, if your GPSr doesn’t automatically record it in the track-log. One option would be to use a 3rd-party utility such as Trimble Planning http://www.trimble.com/planningsoftware_ts.asp to produce an estimate of VDOP for your location. The plot below shows that the GPS elevation accuracy (VDOP) at my location varies markedly over a 24-hour period, with very good readings currently possible at about 4:30 a.m. and 8:00 p.m., but poor elevation accuracy at 2:00 a.m. and 2:00 p.m. for example): Have fun! Edited June 16, 2008 by julianh Quote Link to comment
+bollar Posted July 22, 2008 Share Posted July 22, 2008 I don't know the details of Garmin's algorithm - it is trade secret, and Garmin aren't saying. However, my guess is that the auto-calibration algorithm is optimised for auto-calibration of a moving GPSr, because that is what the unit is basically designed for - and my Mt Coot-Tha test, and roybassist's test show that auto-calibration IS effective for a moving GPSr. I don't know whether the algorithm specifically changes when stationary, but my guess is that it does not - because if it did, I would expect the auto-calibrated barometric elevation traces from my static tests to be even less variable than they were. Sorry to resurrect this old thread, but Garmin's algorithm is patented, and is therefore accessible to us. You mathematicians can have a field-day with this link: http://www.patentstorm.us/patents/7142152/fulltext.html Quote Link to comment
+bollar Posted July 22, 2008 Share Posted July 22, 2008 And, while we're at it, I did these tests a few years ago when the Edge 305 was introduced. The Edge 305 and 705, as you probably know, have barometric altimeters, but they can't be calibrated. I posted a plot of four Garmin devices here: http://www.bollar.org/images/edge305elevations.pdf In short, the barometric Garmins started off with the same deviation from true altitude and converged on true altitude over time, which took about an hour. I never did a test to see how long this would take if the GPS was left on but motionless before the ride. Other thoughts I had about the Edge 305, which was cutting edge at the time with the SiRFstarIII chipset are here: http://www.bollar.org/edge305.htm Quote Link to comment
+Team CowboyPapa Posted July 22, 2008 Share Posted July 22, 2008 Perfect! This is that for which we have been waiting. However, I've had trouble with some of that symbolism and I need some help. Can someone who can read that text better than I give me some help to sketch out a trial scenario: 1. I live at a true elevation of Sea Level. 2. The local ambient pressure is now 29.20 in Hg. 3. I climb to a true elevation of 2,000 meters. 4. At this time the local ambient pressure at Sea Level has increased an amount equivalent to 200 meters. 5. Assume that the local ambient pressure at 2,000 meters increases an equivalent amount. 6. What altitude will be displayed as a result of the patented corrections/algorithm? Thanks! Quote Link to comment
+bollar Posted July 22, 2008 Share Posted July 22, 2008 6. What altitude will be displayed as a result of the patented corrections/algorithm? Assuming it was calibrated at the beginning of the test, or enough time has elapsed to allow auto-calibration to work, it will be 2000 meters or something very close to it, depending on how recent and significant the change in barometric pressure. Quote Link to comment
+julianh Posted July 22, 2008 Author Share Posted July 22, 2008 (edited) I don't know the details of Garmin's algorithm - it is trade secret, and Garmin aren't saying. However, my guess is that the auto-calibration algorithm is optimised for auto-calibration of a moving GPSr, because that is what the unit is basically designed for - and my Mt Coot-Tha test, and roybassist's test show that auto-calibration IS effective for a moving GPSr. I don't know whether the algorithm specifically changes when stationary, but my guess is that it does not - because if it did, I would expect the auto-calibrated barometric elevation traces from my static tests to be even less variable than they were. Sorry to resurrect this old thread, but Garmin's algorithm is patented, and is therefore accessible to us. You mathematicians can have a field-day with this link: http://www.patentstorm.us/patents/7142152/fulltext.html bollar, Thanks for the link. I have found these patent sites myself on previous searches to try to uncover the secret workings of the altimeter calibration routine, but unfortunately, I can only access plain text versions of the patent, which have all the words, but no equations, and no images, so sadly, we mathematicians can't do anything with this information. I would need a FULL copy of the patent (with figures and equations) to be able to "reverse engineer" the method. I am not prepared to pay for a copy of the full patent (and I suspect there are still some subtle bits of information which are not published in the patent - such as the actual values of the time increments between samples, etc) - I have no desire to go into business in opposition to Garmin, and I am simply happy that the auto-calibrating barometric altimeter on my Summit HC really works! And thanks for posting the links to your tests - more good evidence that the Garmin auto-calibration process really does work - especially in the hand-held units, which can be calibrated at the start of the day. Even when you don't / can't do an initial calibration, the auto-calibration routine will self-correct in half an hour to an hour, as your tests demonstrate. (By the way - has anyone ever told you that you have too many GPSrs? You must have received a few strange looks when conducting your tests! ) Edited July 22, 2008 by julianh Quote Link to comment
+bollar Posted July 22, 2008 Share Posted July 22, 2008 Thanks for the link. I have found these patent sites myself on previous searches to try to uncover the secret workings of the altimeter calibration routine, but unfortunately, I can only access plain text versions of the patent, which have all the words, but no equations, and no images, so sadly, we mathematicians can't do anything with this information. I would need a FULL copy of the patent (with figures and equations) to be able to "reverse engineer" the method. I am not prepared to pay for a copy of the full patent (and I suspect there are still some subtle bits of information which are not published in the patent - such as the actual values of the time increments between samples, etc) - I have no desire to go into business in opposition to Garmin, and I am simply happy that the auto-calibrating barometric altimeter on my Summit HC really works! This link should take you to all of the images included in the patent: http://patimg2.uspto.gov/.piw?Docid=071421...View+first+page Good luck pasting all of that into the browser... Quote Link to comment
+julianh Posted July 22, 2008 Author Share Posted July 22, 2008 Perfect! This is that for which we have been waiting. However, I've had trouble with some of that symbolism and I need some help. Can someone who can read that text better than I give me some help to sketch out a trial scenario: 1. I live at a true elevation of Sea Level. 2. The local ambient pressure is now 29.20 in Hg. 3. I climb to a true elevation of 2,000 meters. 4. At this time the local ambient pressure at Sea Level has increased an amount equivalent to 200 meters. 5. Assume that the local ambient pressure at 2,000 meters increases an equivalent amount. 6. What altitude will be displayed as a result of the patented corrections/algorithm? Thanks! tcp, See my reply #38 to your almost identical question in post #36 above: http://forums.Groundspeak.com/GC/index.php...t&p=3495618 Nothing has changed (apart from the specific elevations and barometric pressures you have used in your questions)! Assuming you are using a late-model Garmin unit with auto-calibrating barometric altimeter, and you give it an initial calibration at the start of your trip (at sea level), it will hold accurate elevation for the duration of your trip, and when you get to a true elevation of 2,000 metres, it will show an elevation of 2,000 metres, plus or minus about 5 metres, in my experience (personal Summit HC and company 76 CSx) - regardless of what the weather has been doing! If you forget to do the initial calibration at the start of your trip, the first half hour or so of your track-log MAY have elevation errors until the auto-calibration is completed, but by the time you get up to 2,000 metres, your unit will be showing 2,000 metres, plus or minus about 5 metres. I can't speak about the performance of older model Garmins or other brands. Hope this helps! 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.