Fidelity of GPS derived speed/distance measurements

[BdaGhisallo]

Can anyone educate me on how accurate GPS speed and distance based measurements tend to be compared to traditional wheel magnet/sensor setups?

My wheel transmitter seems to have given up and I have an SRM PC8 incoming in the next month. It has GPS and can, therefore, seemingly do without the wheel speed sensor. If I can do without getting another sensor I will.

I am not that fussed about distance and speed being exactly reported so if GPS gets me within a few percentage points, I'll go that way.

I think I have my answer as you can probably tell but some corroboration would help.

Cheers,
Geoff

[Ken Robb]

I think GPS calculates speed/distance "as the crow flies" so the numbers might be compromised on a curving road. This just a guess based on my limited understanding of how satellites see travel.

[ergott]

I used to compare my garmin with a speed/cadence sensor to my phone and just the garmin without the sensor. The mileage results were real close (within a couple tenths of a mile) using the same route each time.

That's for my commute. For significantly longer rides it was still within a percent or two.

[11.4]

GPS measures over your actual path of travel. Better units also factor in the extra distance involved in climbing or descending.

Where GPS is weak is in adjusting for instantaneous speed (e.g., you suddenly go up a steep hill and want to know if you're really going only 3.5 mph, and it doesn't update until you are already over the top, and dropped).

Also, just a piece of sage advice from a trackie, it doesn't work at a velodrome. The GPS sometimes realizes you are moving, sometimes can't discern the distance, so it is wildly inaccurate. I've seen this happen on small cross courses as well. For whatever reason, an iPhone does a lot better at this situation than a handlebar-mount dedicated cycling device.

[BdaGhisallo]

Thanks guys. It sounds like the GPS will do the trick.

[shovelhd]

Tree cover can also affect speed accuracy. I have a speed only sensor on my chainstay and use the SRM for cadence. The Garmin is a speed backup.

[Mark McM]

I think GPS calculates speed/distance "as the crow flies" so the numbers might be compromised on a curving road. This just a guess based on my limited understanding of how satellites see travel.

Well, this is essentially true - but because it takes a reading every second or so, it adds up a large number of straight lines.

Much like you can approximate a curve with a large number of straight lines, if the GPS adds together a large number of small straight lines, it becomes a very close approximation of the true distance.

To answer the original question:

I can't say which is more accurate than the other, but my Garmin 800 GPS reports distances and average speeds about 2% shorter than my Sigma ROX 8.0 with a wheel sensor. As far as instantaneous speed displays, the variation in reported speeds varies a bit with speed. At the slowest speeds (as when going up steep climbs) the variation in reported speed can be high (more than 1 mph). That is likely to do with the error in successive GPS readings being a larger proportion to actual distance traveled between readings.

[nm87710]

I have an SRM PC8 incoming in the next month. It has GPS and can, therefore, seemingly do without the wheel speed sensor. If I can do without getting another sensor I will.


Good Luck

[Jaq]

GPS measures over your actual path of travel. Better units also factor in the extra distance involved in climbing or descending.

Where GPS is weak is in adjusting for instantaneous speed (e.g., you suddenly go up a steep hill and want to know if you're really going only 3.5 mph, and it doesn't update until you are already over the top, and dropped).

Also, just a piece of sage advice from a trackie, it doesn't work at a velodrome. The GPS sometimes realizes you are moving, sometimes can't discern the distance, so it is wildly inaccurate. I've seen this happen on small cross courses as well. For whatever reason, an iPhone does a lot better at this situation than a handlebar-mount dedicated cycling device.

Went out once a while back with Strava running on my phone. As I understand it, the system not only uses GPS, but local wi-fi to really dial in the accuracy. Anyway, at about half-way thru the ride, I stopped at a Ralph's to grab a bite.

Got home and looked at the results and saw that my average speed was much lower than normal. I forgot to pause the app at Ralphs, so it kept tracking. I called up a ride map, zoomed in, and saw my path all over the the store, from deli (hm, maybe a sandwich? no, too heavy) to deserts (oh, a cookie sounds good!) to liqueur (a cold beer would be awesome) back to fruit (yeah, maybe just a couple bananas) and so on. The whole thing looked like a worker bee's honey dance.

Good news: I got the KOM for that Ralphs.

[DfCas]

I've had 4 Garmins and current speed is NOT reliable. It will bounce around the correct speed but rarely display the correrct current speed you are traveling. It will average out just fine, but cannot display current speed accurately, even with a speed sensor on the wheel.

[wasfast]

What are you defining as reliable and accurate? 10mph, 1mph, .1mph?

[1centaur]

Agree with that (DfCas). When I first got GPS with the speed sensor I also had a Flight Deck and compared the two as I rode. The Flight Deck (uses a magnet) was intuitively accurate and the GPS speed sensor was fairly close most of the time but lagged, in part because the algorithm seemed programmed to try for the satellite feed first and the sensor second, so it took time to go to the second choice. I gave up on the speed sensor and therefore simultaneous speed as I moved on to new bikes, which can be a little annoying but I learned to live with it.

GPS is really a series of pings with an algorithm to correct for what is missing. From moment to moment (i.e., current speed) there is no chance for an algorithm to respond as quickly as reality, so your speed will be pretty accurate with a wide open sky and good atmospheric conditions and no sudden changes. But over the course of an entire ride the algorithm will link point A to point B to point C and thereby get your distance about right, and it knows your time, so it will calculate your average speed correctly.

[DfCas]

What are you defining as reliable and accurate? 10mph, 1mph, .1mph?

I ride a rail trail a lot and we have a long ~12 mile section that climbs slightly, maybe 1 -2 %. It is my benchmark ride and how I guage my fitness, so I want to know how fast I am going.

If I am traveling at 10 mph, in a 10 second period I may get the following speed readouts on a garmin.

9.2, 11.1, 8.9, 8.7, 10.7.

All are close, but none are actually correct.

With my basic $20 computer, It will sit on 10 or very close to it, so a given glance at the speed readout is accurate. It is not with any of the 4 Garmins I've had.

Garmins are great for a lot of things, but current speed is not their forte.

[christian]

It samples once per second.

Broadly, it's very good at distance and average speed. It's less good at instantaneous speed.

[shovelhd]

It's not accurate enough for racing, and I do use current speed while racing.

[unterhausen]

I don't really care much about speed, and the navigation function doesn't really depend on distance, so these don't really matter to me much. It's good enough that I feel no need to install a speed sensor. I went for years without without a computer. However, I needed help navigating at night, so I got the GPS.

[Mark McM]

GPS measures over your actual path of travel. Better units also factor in the extra distance involved in climbing or descending.

Do they really do this? Even up steep hills, the error is quite small. For example, riding up a 10% grade the error in horizontal measurement is only about 0.5%. Given the other errors in trying to measure an undulating path with periodic GPS measurements, this would get lost in the noise.

[Mark McM]

As I understand it, the system not only uses GPS, but local wi-fi to really dial in the accuracy.

Actually, the Wi-Fi isn't to increase GPS accuracy, it is to speed up the initial location fix (a freshly started GPS can take some time to establish the initial fix). After the initial fix is established, it uses the GPS signal for subsequent motion.

[ptourkin]

Last week I was told February or March for the PC-8. I'm putting the SRM Suunto speed sensor on the chain stay vice front wheel so it will work on the trainer and moving the Garmin over to my track bike for cadence. The GPS speed data is full of errors, if you're really interested in it.

[ergott]

Speed and mileage on a trainer aren't as important as time and watts.

[cderalow]

don't most Garmins allow you to choose sensor input vs gps for speed?

I would think the sensor would be pretty accurate for instant speed. GPS not so much

[ergott]

I haven't dug through the settings in a while, but I thought the Garmin will automatically favor the information from the sensor over gps for speed/distance etc.

[DfCas]

don't most Garmins allow you to choose sensor input vs gps for speed?

I would think the sensor would be pretty accurate for instant speed. GPS not so much

No, they favor GPS over the sensor and sometimes try to reconcile the 2, leading to wildly innacurate speed numbers.

[ergott]

No, they favor GPS over the sensor and sometimes try to reconcile the 2, leading to wildly innacurate speed numbers.

Where did you get this information?

The manual and other reviews of the Garmin point to the sensor data being recorded and used first. At the slow speeds you are talking about the wheel isn't really rotating that fast. Do you really find the data that inaccurate at road speeds and what are you comparing it to?

[DfCas]

I had a 510 with a sensor kit. I tried all the settings and could not get stable,reliable speed readings. Maybe I was using it wrong, but I RTFM repeatedly and could not get it to work well.

[shovelhd]

Where did you get this information?

The manual and other reviews of the Garmin point to the sensor data being recorded and used first. At the slow speeds you are talking about the wheel isn't really rotating that fast. Do you really find the data that inaccurate at road speeds and what are you comparing it to?

That's the way I understand it. The sensor overrides the GPS if present. You can verify this by shutting the sensor off during a ride.

[ergott]

I had a 510 with a sensor kit. I tried all the settings and could not get stable,reliable speed readings. Maybe I was using it wrong, but I RTFM repeatedly and could not get it to work well.

I've read about dead sensors even new. Did you ever get it checked out? Was it that inaccurate at road speeds?

[DfCas]

I would not say that, I would say erratic. I would be going 15-20 and suddenly it would say 3 or 6mph or something stupid and then go back to correct speed. I got tired of it and bailed.

The original poster asked if GPS speed was accurate and I vote no.

[unterhausen]

that's interesting, my 800 has been rock steady. I really couldn't tell it from a speedometer with a sensor

[roydyates]

I haven't dug through the settings in a while, but I thought the Garmin will automatically favor the information from the sensor over gps for speed/distance etc.

True for the 705. Dunno if it's true for other models.

[roydyates]

No, they favor GPS over the sensor and sometimes try to reconcile the 2, leading to wildly innacurate speed numbers.

set your wheel size to something really wrong and see what the garmin reports. My 705 reported speed from the sensor, not the gps. Of course this is something garmin could cary with firmware revisions.