What is wrong with our GPS data?

Recently, I have been surprised by the differences between the distance reported by my GPS watch (a Garmin Fenix 2) and what is reported when importing the data into website such as Mapmyrun or Strava.

Data from Garmin Connect
Data from Garmin Connect

It happened this week-end at Rock’n The Knob, a 20 miles run on trails. I went off course, making an additional 1.5-ish mile. At the end of the run, my watch reported 21.27 miles, which made sense (20 miles for the race plus an additional mile or so for being off course). On the other hand, both mapmyrun and strava report 18.9 miles.

Same GPS file imported on mapmyrun
Same GPS file imported on mapmyrun

 

So, who can you trust?

I tried to look at the strava website to see if they modify the GPS data. I found the following discussion and the technical support note to be interesting. It seems that website are doing some post-upload processing that mostly change the data you import and try to remove any bad GPS data. On the other hand, considering the race I did, it seems that this post-upload processing removes a lot of data, about 11%.

Same GPS file on Strava
Same GPS file on Strava

Many folks lately asked what is the most accurate data source between a watch and a phone. It turns out that this is not only the source of the data but also where you put the data and how it is processed. I am not a GPS specialist and strava or mapmyrun will not publish their own algorithms but for now, I rather trust my watch: so far, it has been consistent with other data sources (signs on the road, watches from other brands, etc.).

What is wrong with our GPS data?

Comparing GPS running systems (part1)

That’s a fact: everyday, I do not commute from home to work using the bus, the car or my bike. I run. Approximatively 6 miles, depending on my mood, my running shoes and if I am late. No matter if the sun is here, if this is pouring or if roads are icy: I run, no matter what. I do not always take the same route, it varies according to how I feel, traffic, congestion, etc. But when conditions are stable, I often take the same route for several days.

This motivates me to compare the GPS systems for running. So far I have the following devices:

So, as I take the same road since a couple of days, I mapped the same route with different devices and import the results on the same website. Please find the link to the data below.

Using the Android with Mapmyrun

Using the Garmin Forerunner 10

 

Analysis

The data from the Garmin are quite consistent. the variation can be explained easily by how I run (taking the tangent, etc.). One would expect a slight different before two runners taking the same route. On a 6 miles route, it is reasonable to think we can have a different of 0.05 mile or even 0.1 mile. On the other hand, the difference with the data aken with the phone is really different, the variation being more like 0.2 mile.

 

Understanding the GPS accuracy

The difference can be also due to the GPS system itself. The accuracy of the positioning system depends on several factors that may vary according to independent conditions (atmospheric condition, sky, etc.). But on a $100 phone, the quality of the GPS chip is mostly poor but enough for most usages (i.e. get direction when driving) but may be not good enough for having a very good accuracy. But it can have an impact of the data being transmitted to the phone. Another important factor is work load on the device that acquires and aggregates all the data. On a watch, there is a single application that acquires the GPS signal and record the time. But your phone might execute several applications with many background tasks (hey Facebook or GMail updates!). These tasks may preempt the execution of your main application (i.e mapmyrun). Also, the application might be designed in a manner that GPS data from the phone and the watch are acquired/processed differently.

The take-away

There are clearly some differences between the same route acquired by the GPS watch or the phone. Who is right or wrong? I will try to investigate by timing the same route with the Forerunner 310XT and see the difference. In the meantime, for most runners, the difference is not really significant and any application would be more than enough for you. After all, running is about having fun and even if we like to time and log our course, we have to keep in mind that the most important is to enjoy and have fun.

 

 

Comparing GPS running systems (part1)