How accurate is a GPS-based speedometer?


Many GPS devices display your current speed, based on a simple calculation of how much distance you are covering in a given time period. However, drivers are often confused with these readings since it generally varies significantly from the vehicle’s speedometer.

So which speedometer is more accurate? Which one should be used to determine if you are driving within the posted speed limits? How does it relate to using speeding alerts in your telematics program?

Speeding is widely regarded as a significant contributor to fatal road accidents, and fleet owners find GPS-based speed calculations very useful in maintaining fleet safety. It’s important then that these speed calculations can be trusted to avoid drivers being falsely accused of speeding, as well as providing a reliable guide to determining a legally-safe driving speed.

How does a GPS device calculate speed?

Using GPS, a device is able to calculate a lot of information about a moving object. Using even basic time and location data, a GPS unit can quickly calculate the relative speed of the object, based on how much distance it covered in a given time.

GPS devices are positional speedometers, based on how far the device has moved since the last measurement. The algorithm also uses the doppler shift in the pseudo range signals from the satellites. It should also be noted that the speed reading is normalized, and is not an instant speed.

Speeds are updated at short intervals to maintain accuracy at all times. It uses frequent calculations to determine the vehicle’s speed. For example, using a standard movement per time calculation, if you have covered 80 feet in one second, the GPS device works out and converts that to MPH, which in this case is 55MPH.

Why does GPS speed differ from the vehicle’s own speedometer?

GPS speed calculations are more accurate than a vehicle’s speedometer since it is not affected by the same inaccuracies, including the vehicle’s wheel size or drive ratios. It is dependent however on GPS satellite signal quality but with the use of moving average calculations any errors are minimal.

Vehicle speedometers also require calibration to maintain perfect accuracy as general wear and tear, changing wheel sizes, and the manufacturers own ‘erring on the side of caution’ and setting the speedometer to read higher than the actual speed, all contribute to inaccuracies.

Generally most manufacturers claim a 0.1MPH tolerance with speed readings, however will go as far to say a 0.5MPH inaccuracy is possible, to make allowances for any satellite signal variations beyond their control.

Which speedometer should a driver use?

It really comes down to personal preference since both gauges are reliable for determining a safe, legal traveling speed. In most cases the vehicle’s speedometer will show a higher speed due to the aforementioned inaccuracies but this at least ensures the driver stays lower than the posted speed limit.

And while logged GPS data has been used to overturn speeding tickets, the cautious driver, who is keen to avoid infringements, fines, trouble from their employer or the possibility of an accident, will tend to err on the side of driving slower rather than pushing it too close to the limit.

Using GPS to record company speeding incidents

Any business that has company vehicles available for employee use is no doubt keen to avoid dangerous speeding, both from a public liability point of view, the possibility of accidents as well as infringement notices or demerit points. Using fleet management software such as Telogis Fleet, fleet owners are easily able to specify speeding notifications that can alert managers to speeding incidents in real time, or provide regular reports of this sort of activity.

Speeding alerts can be setup to record either speeds in excess of the posted speed limit, or simply if a certain speed is reached (ideal for vehicles that become unsafe above certain speeds e.g. forklifts).

How closely the speeding alerts are monitored can be set as well, such as how long the vehicle is speeding before the alert is triggered. This can be used to avoid false positive reports due to overtaking or temporary distraction.

Since employers know that speeds calculated by GPS devices are not only more accurate than a vehicle’s speedometer but also lower, it gives employees no excuse to be caught driving at unsafe speeds. Through a combination of regular reporting and driver coaching, fleet owners can make sure they are doing everything in their power to improve the safety of their staff and other road users.

Reduce fleet miles

21 Responses to “How accurate is a GPS-based speedometer?”

  1. Michael Barrett Says:

    I ride a motorcycle and when I use my GPS the speed differential between speedometer and GPS is quite significant. I decided to calculate the speed according to the RPM of the engine adjusted by all the different gearing ratios to determine the rear wheel speed based on the actual circumference of the wheel. According to these calculations my speedometer is accurate making the GPS wrong. If the mathematical calculations say the speedometer is accurate why does the GPS differ so significantly? TYre sizes are standard according to the manufacturer’s specifications as well.

  2. Eric Says:

    Seed is proportional to the distance. Since that, car manufactures intentionally calibrate higher speedometers to void earlier car warranty. Warranty is dependable of two factors: time or mileage.

  3. Jay Mueller Says:

    This is almost all wrong. The GPS speed calculations based on distance between positions is quite inaccurate. Most GPS positions are accurate only to within ~5m. This means if you actually travel in a straight line of 50m in 1s, the GPS could see that as a distance of up to sqrt(50^2 + 5^2) = 50.24m. There is an accurate form of GPS speed measurement but this is using Doppler shift measurement. Have a look:

  4. S Burns Says:

    Does a GPS speedometer just use horizontal movement. If you went down a steep hill then up the other side wouldn’t the gps only see the movement as the distance between the two peaks rather than that travelled down to the bottom and back up the other side.

  5. Tony S Says:

    @4 Whether the GPS speed reading takes into account the movement in 3D is dependant on the software, one would hope that designers had taken this into account.

    @1 Motor cycle – You do not know the accuracy of your speedo or rpm guages. They have been set with assumptions on wheel diameter and general accuracy and legal considerations. GPS IMO gives a better result.

    @3 – GPS accuracy. It is well known that GPS has an absolute accuracy of +/-5m, this is mainly dependant on timing accuracy which is fairly fixed over short period, therefore relative accuracy is higher. Furthermore, even considering GPS using absolute errors as a worst case, you can see that it will get more accurate as the speed increases as the position error becomes less significant over the distance between samples increases.

    Generally I find my speedo and GPS differential fairly linear as a %. Since I know the GPS is more accurate higher up, I consider that both are fairly close lower down and the absolute differential is small in both cases. The speedo error seems mainly down to the fact that the designer has built in a ‘always read too high worst case’ into the car so they do not get sued for speed infringements/injury by owners rather than choosing the most accurate reading over the expect range of rolling radius’. ie if the system can manage +/-2.5% they design the speedo to read +5%.

  6. LBoden Says:

    I think SBurns question deserves more specifics. Your statement “@4 Whether the GPS speed reading takes into account the movement in 3D is dependant on the software, one would hope that designers had taken this into account.” Is not a good answer since you include the word “hope”. Can you name GPS manufacturers that include software/algorythms that truly calculate speeds in 3D? Other than time and distance, what is the third variable used by GPS satellites to accomplish 3D speed calculations?

    Also, what is the refresh rate of the speed calculation? It can’t be in real time regardless of how fast the refresh rate is since “distance traveled” used for calculations are always past tense.

  7. LBoden Says:

    One more comment. If you have a speedometer checked by a device that is calibrated to a NIST standard, the resulting readings will reveal the true and actual speedometer display while taking the total possible error variables into account (i.e., at a true 50mph the speedometer displays 48 mph, at a true 60 mph, the speedometer displays 59mph, etc.) After all, the wheels on the road travel the complete distance by marrying itself to the asphalt during the distance traveled. Obviously, there is no need for a third variable to calculate speed or distance….and it’s real time.

    If you know your speedometer certified error, by interpolation your speedometer reading is the most reliable instrument to keep you accurately informed of your speed.

  8. Michael Barrett Says:

    @5 TonyS – You and everyone else ignores my question which is “The arithmetical calculation of the speed of my motorcycle equals the speed shown on my speedometer which differs from the speed shown on my GPS. WHY???”. I have used the actual gear ratios as published by the manufacturer. I calculated the circumference of the driven rear wheel AND I actually measured that circumference. You later comment that your GPS and speedometer differential is a fairly constant %, therefore the GPS must be correct. I could just as easily say the speedometer must be correct. Your statement is merely your supposition as you have no evidence proving it’s veracity. No one here has given any real explanation to my question. Everybody just says the GPS must be correct. I just rode 315km from Sedgefield to Port Elizabeth based on my odometer. When my GPS calculated the distance it said it was 288km. I often find that the GPS distance also differs from the ditance shown on signboards along the road but my odometer always agrees with those signs. How do you explain that?

  9. TonyS Says:

    Michael, you need to calm down and stop acting like a spoil child. I didn’t come on to give a perfect answer to your question, I came on to add my opinion and experience. I am not able to answer without all the detail and really analysing your method in detail, but I doubt you are capable of reporting it sufficiently. You did not mention on your first post that you have measured the wheel which is commendible. Are you able to measure the wheel diameter while traveling at speed? Perhaps the calculated speed is similar to the speedo because it is the same method used by the manufacturer. If you want a more accurate calibration I suggest you measure distance between as many mile markers as you can find and see which device is more accurate. You will find less debate about method and you might actually add something. If your GPS turns out to be inaccurate then please post the model.
    To Others- Again I am not here to give good answers to your questions. I am certainly unable to know how software engineers have calculated speed or distance without seeing the source code or doing some very clever analysis. FYI GPS does calculate position in 3D coodinates. It calculates distance to each satellite and then works out position in 3D space using 3 or more satellites, it would be very bad practice to not include the direct A-B distance travelled between samples in 3D.

  10. Stephen Says:

    @8 – I am curious about this because I am considering getting a GPS based speedo. What type of GPS are you using? Are you using an actual gps-based speedometer which is meant to replace a standard speedometer, or a map-route finding gps which is not meant to replace a standard speedometer? It sounds like your gps might not be taking the ‘z’ distance into account which would explain the shorter distance traveled. Presumably a map-route based GPS might not factor in the ‘z’ distance as it isn’t as important for its purpose.

  11. Doug N. Says:

    Michael, did you measure the circumference of your driven tire with you on the bike? An “unloaded” radius will be larger than a “loaded” radius, making the true radius that you are riding vary with load, and be smaller than what you measured in the garage. This will change the accuracy of the speedo, as it has no way to adjust to a changing rolling diameter

  12. Michael McDermott Says:

    just wanted to add that in GERMANY (lived ther long time as car designer)
    the LAW requires the OEMs (i.e. car manufacturers) to make the speedometers
    show too much – reason being that then no one can claim in court his or hers
    speedometer was showing a speed below the limit ….
    Don´t know about the LAW in your country – but I would assume something
    similar IF you have car manufacturers in your country …
    By the way – the inaccuracy limit is 7% – of the MAX value on your speedometer – so regarding above picture – 8.4 km/h
    AND that is for every speed indicated !!
    Another thing to remeber – ever noticed after buying new tires that the car is suddenly slower (topspeed)? The radius to the axle of a properly inflated NEW TIRE is used for designing a speedometer – remember a case where a guy went to court claiming not to have exceeded the speed limit – only that they found out he had mounted bigger tires than allowed – LOL

    P.S.: @Michael Barrett – Your odometer measures the distance with the same inaccuracy as the guys who put up the signs – they used something similar because they didn´t have GPS back then ….

  13. Michael Barrett Says:

    What Michael McDermott is missing is that I am using standard tyres on my vehicle, and, that I not only calculated the arithmetical circumference of that standard tyre, but I also measured the circumference with a tape measure. Everybody just keeps saying that the GPS correct is because it cannot be incorrect. Nobody has come up with any scientific evidence that shows why my arithmetical calculations show my speedometer to be accurate, while the GPS shows a different speed!!!

  14. Michael McDermott Says:

    @ Michael Barrett.
    I have to point out that the circumference is NOT valid (read previous post).
    Your speedometer will show a higher speed than in reality (by law).
    Your rpms (even if digital) will also be inaccurate.
    Every device has a tolerance – only that of a GPS device is smaller
    than your speedometer and rev counter.

  15. Michael Barrett Says:

    In answer to all the comments regarding the change in circumference of the driven wheel either through adding a load to the vehicle or expansion due to higher speed, I did some further calculations. At 120 km/h as per the speedometer the GPS indicates 112 km/h. The change required in the tyre size to achieve this is to reduce the diameter by 42.9mm or 1.7″. The tyres were inflated to the correct pressure (2.5 bar) and I weigh 78kg. There is no ways that the tyre could be squashed to that extent. If it was the motorcycle would have been impossible to handle anyway.

  16. Michael McDermott Says:

    P.S.: I looked up speedometer tolerances.
    The german law has been replaced by a european one (should have known).

  17. TonyS Says:

    MichaelB, you cannot take the difference between the readings and say that must be due to tyre squashing.
    You also cannot take the tyre circumference and use that in calculations because it is not the rolling radius. Measuring the distance travelled with you on it, will give a better result, but there will be a small error from the centrafugal force as you go faster.

    You are using several figures in your calculation all of which you do not know accuracy (RPMs + gear ratio + rolling radius). GPS is single measurement by a single pice of equipment. For that reason alone I suggest it is more likley to be accurate, but cannot be absolutely sure. I have also offered my own comparision which is consistent with more accurate GPS. Additionally it is much easier to measure timing accurately than it is to measure amplitude (or size). GPS works entirley on timing.

  18. Michael McDermott Says:

    @ TonyS
    Speed is in m/s or km/h or miles/h
    So distance and time.
    GPS works on time and distance – otherwise it wouldn´t be possible to calculate speed – for obvious reasons.
    BUT – you are RIGHT because it CALCULATES the distance traveled by the difference in TIME it takes to recieve a signal from a satellite in two different positions. Assuming the “clock” in the GPS is inaccurate – lets be generous and say 24 seconds a day (for obvious reasons). Thats 1 second per hour and 1/60th of a second per minute – thats less than .03%
    The positioning accuracy of my GPS is 2.5 meters. Lets say your driving at 108 km/h which is 30 m/s. Assuming you do that for 1 minute you traveled a distance of 1800 meters. Inaccuracy of distance worstcase 5/1800 – thats less than .3% – the longer you travel the accuracy only gets better.
    If I remember correctly ALONE the difference between new and old tire already gives you 2% inaccuracy.

  19. Michael Barrett Says:

    I realise that you are all trying to give the correct answers based on your own knowledge and experience and I apologise if I have occasionally thrown my toys out of the cot because I was getting frustrated.
    Firstly I have all the specifications of the engine in my motorcycle. From the crankshaft via the Primary Ratio to the individual gear ratios to the final drive Front & Rear sprockets. The rolling radius is half the diameter of the wheel plus the tyre. The tyre in question is 150/70 R17. Doing the calculations the rolling radius is 320.9mm. Then the circumference is Pi x D or Pi x 2R. This gives a rolling circumference of 2016.3mm.
    The formula to calculate the speed of the vehicle at a particular RPM is: (rpm x rear tyre circumference in inches x front sprocket) / (prim ratio x gear ratio x rear sprocket x 1059) = Speed mph x 1.60934 = Speed kmh. Is this formula incorrect???
    Then there is the question of the accuracy of the Rev Counter. At 120km/h on the speedo the GPS shows 112km/h, which is a difference of between 6% and 7%. This would mean that the Rev Counter would be showing 5500RPM when the actual RPM is between 5115 and 5170. I cannot believe that a manufacturer would have such large discrepancies. Also I cannot find any websites reporting such large errors in Rev Counters.
    A final point is the question of what the dynamic rolling diameter would be at a particular speed. I have not been able to find any definitive information on this point but if we accept that the Rev Counter is accurate then this would have to be 43mm (1.7”) smaller. With the sidewall height being 105mm this is 41%. I have never seen a correctly inflated tyre deform to that degree. One must also remember that while adding weight deforms the tyre travelling at speed makes it expand slightly as well.

  20. Michael McDermott Says:

    @Michael Barrett

    I am guessing if you have a rear and front sprocket your rev counter is driven by a rotating flexible cable – please correct me if I am wrong.
    They are per definition inaccurate – not digital.
    Coming from car design I would assume that the manufacturer ALWAYS wants the rev counter to show too much rather than less – for saftey reasons. So as not to over rev the engine. Modern engines don´t need that because the electronic cuts off the ignition to prevent over rev. Downshifting on an automatic is prevented electronically. Manual can be forced to over rev – especially diesel engines (no ignition) – but thats misuse and is logged in the vehicules compter to prevent unlawful claims. Bottom line is cable driven like mine – inaccurate.

  21. TonyS Says:

    I’d say the rev counter ‘could’ be accurate, as I said timing is easy to measure electronically to excellent accuracy, but if its a mechanical rev counter (dragging magnets) then all bets are off. A simple test is to get a digital rev counter and trigger it from the ignition. This will give a good idea how accurate it is. You can do this static on a engine tuner, or mechanics multimeter.

    2ndly you don’t know the accuracy of the engine specs, ie the gear ratio.

    Another way of getting a good comparision is to use a bicycle speedo. You normally enter the rolling circumference and it then measures timing electronically. The measurement mainly depends on the accuracy of your circumference measurement. Not sure how well the wheel will take to the weight of the trigger (magnet normally) and the max speed ability of the unit and trigger.
    Theoretically you could make an optical speed sensor that looks at the road and measures speed absolutley, works a bit like an optical mouse triggering from the surface texture. I wonder if anyone has made one yet.

Leave a Reply