There are also elevation effects which aren’t corrected by car GPS and also the GPS assumes movement on a flat plane when the earth is in fact round.
There are also others as well:
In my last job we used a Trimble differential with Omnistar signal to get accuracy down to about 0.10m. Without the Omnistar signal, the accuracy was submetre…but only when multiple points were taken for averaging.
A car GPS has the following accuracy:
The United States government currently claims 4 meter RMS (7.8 meter 95% Confidence Interval) horizontal accuracy for civilian (SPS) GPS. Vertical accuracy is worse. Mind you, that’s the minimum. Some devices/locations reliably (95% of the time or better) can get 3 meter accuracy.
Even using say the 4 metre accuracy, if a car is travelling along a road, each sample point taken by a GPS can be 4m out. If there is a shorter sampling duration, the accuracy of a car GPS providing accurate speeds is low and could be out by >10% or more. Reliability of the speed measurement would only occur when speed is high and sample duation is long (sample time say corresponds to say 100m or more and then the accuracy could be <8% (4m + 4m over 100m).
This is why car GPS readings aren’t admissible in court when challenging a speeding fine as they are not accurate.
Unfortunately, there are many who think that GPS are more accurate than a car speedo and use the GPS speed when driving…and why some receive fines for speeding thinking that have being obeying the speed limit.
Maybe the manufacturers of the GPSes should state clearly the accuracy of the speed detected by their devices.