Accuracy of GPS speed measurement

Instrumentation for measurements, and instrumentation for anything else; also control systems including self-steering gears
Post Reply
John Perry
Committee Member
Posts: 52
Joined: Tue Nov 08, 2016 6:39 pm

Accuracy of GPS speed measurement

Post by John Perry » Sun Sep 29, 2019 2:40 pm

Can anyone tell me about the accuracy of GPS speed measurement - I ask because I am wondering if it is worth including GPS speed measurement in the unit I built to measure the drag of towed hulls at the recent AYRS event at the Bassingstoke Canal Centre - see posts here under 'Events'.

The American government provides information about the positional accuracy of GPS at the website gps.com, but there is nothing there about speed accuracy. I understand that gps determines speed by a doppler shift method, its not just a matter of timing between two positions.

A couple of writings I have found on the internet sugest gps speed error limits around 0.2 knots for a straight line course (higher on bends and wiggles but that is not relevant to our current application). Presumably that means +/-0.2 knots. The timing at Weymouth Speed Week is by use of GT31 GPS data loggers. The manufacture's (Locosys Technology Inc.) specification for this unit mentions a figure of 0.1m/s, again, about 0.2 knots.

Somewhere on the internet I have seen mention of a Garmin specification of 0.1knots for GPS speed accuracy but I havent been able to find this in a document from Garmin.

For the tow testing we are considering, for the time being at least, speeds will probably not exceed 5 knots, so +/0.2 knots is +/- 4% or more. We should be able to improve on that by refining the current timing method, so if the gps errors are really that great there may not be much point having the gps module in our equipment. Would be nice to know.

Fredthecharlie
AYRS Chairman
Posts: 75
Joined: Sun Oct 11, 2015 5:31 pm

Re: Accuracy of GPS speed measurement

Post by Fredthecharlie » Mon Sep 30, 2019 6:09 pm

I believe that over a reasonable distance(?) it is accurate but the best GPS units are I think differential and accurate in position to cms where as older (and ? cheap chips ) accurate to 15 metres so the speed may be in error but it may be worthwhile as then on has automatic speed recording at known moments to compare with other data

John Perry
Committee Member
Posts: 52
Joined: Tue Nov 08, 2016 6:39 pm

Re: Accuracy of GPS speed measurement

Post by John Perry » Tue Oct 01, 2019 1:14 pm

Hi Fred

My understanding is that GPS determines speed by a method that is fundamentally different to the method by which it determines position. Hence knowledge of the positional accuracy does not necessarily tell us much, if anything, about speed accuracy.

The American government provides information about the accuracy of GPS here:
https://www.gps.gov/systems/gps/performance/accuracy/

Looking at the above web page a second time I see that there are a few paragraphs about speed, I somehow missed that when I looked at it before. Quoting from the above web page:

The government provides the GPS signal in space with a global average user range rate error (URRE) of ≤0.006 m/sec ( over any 3-second interval, with 95% probability.
This measure must be combined with other factors outside the government's control, including satellite geometry, signal blockage, atmospheric conditions, and receiver design features/quality, to calculate a particular receiver's speed accuracy.


For the GT31 GPS unit used at Weymouth Speedweek, the manufacturer (Locosys) gives an accuracy of 0.1m/s (0.19knots). The GT31 has now been replaced by a new model, the GT-60, that is in the form of a watch worn on the wrist. According to the webpage gps-speedsurfing.com, the GT-60 has 99.7% probability of accuracy over a 10s averaging period to within 3cm/s (0.056knots)

There are plenty of scientific papers that consider GPS positional accuracy but these generally do not address speed accuracy. One paper that does discuss gps speed accuracy is here:
https://scialert.net/fulltextmobile/?do ... .1518.1522
This seems rather a simplistic study compared with some studies of gps, the authors simply drove around in a car recording speed from a gps - what could be easier than that? They used a sensor measuring the angular speed of one of the car wheels as a reference measurement but they don't give any indication of the accuracy of that reference measurement. Their main conclusion was that 50% of the data points were in error by less than 0.51km/hr (0.27knots) and 95% by less than 5.54km/hr (2.99knots). This was for speeds ranging from 5km/hr to 50km/hr. It is a pity that they didnt average groups of readings over three second periods to allow direct comparison with the US government specified URRE for the 'in space' signal. Also a pity they didnt try different gps units to see if that makes any odds, maybe it does not. Occasional readings were obtained with very large speed errors, which is worrying.

So from the above I see widely varying indications for the speed accuracy of gps. I note that the new GT-60 appears to have about 3.4 times better accuracy than the earlier GT31 and both these have better speed accuracy than is claimed for gps units in general. I wonder why these big differences since they all use the same basic system? The GT31 was advertised as a 'doppler' system and the GT-60 as a 'full doppler' system, but I read that all GPS units measure speed by a doppler method. It may be that the difference between the GT-60 and the GT-31 is mainly, or entirely, due to the accuracy for the GT-60 being quoted for a 10 second averaging period, whereas that for the G-30 is for an unspecified averaging period.

I am not sure how far I wish to take this since I have other yacht research projects in mind. It would depend to some extent on whether the results might be of interest to other AYRS members or whether members would like to do more tow testing at future meetings. I think it is potentially a good 'team building' exercise for an event, with different members providing their craft for testing.

If we do continue with this kind of testing I think we could do much to improve the accuracy of speed measurement by timing and this may be preferable to using GPS. For low speed work we could probably improve on the accuracy quoted for the GT-60 - +/- 0.056 knots is a bit over +/-1 %. at 5 knots. I am wondering about infra red light beams to mark start and stop lines, this could give a timing accuracy of perhaps a few milli-seconds or better, so accuracy would then depend mainly on how well we can measure the distance between start and stop lines, presumably using a surveyor’s tape. An infra red transmitter and receiver as used for detecting the presence of a vehicle under a traffic barrier is not expensive. These typically have a specified range of 20m, but somewhat degraded in rain or mist. I think this is just about adequate for working on a canal with units set up on opposite canal banks or on poles driven into the canal bed. Two transmitter/receiver sets would be needed and each would be connected to a microcontroller with a radio system transmitting to a third microcontroller which would probably be on the towed craft and would also be recording the towline tension. 5 battery packs needed. Quite complicated to set up on site (good team building exercise!), but doable and not all that expensive?

I know that some of the Weymouth Speedweek people have put a lot of thought into speed measurement over the years, perhaps I will be able to ask them about all this at Weymouth next week.

Post Reply