I get questions about inconsistencies in Strava data, probably because I’m not only the gear maven but also the most versed in trigonometry in my circle of riding friends. And some of the biggest inconsistencies involve elevation data.
Elevation data is rubbish. The base data is rubbish: unless an elevation has been confirmed by human surveyors, you can’t trust it. And then elevation as determined by GPS — is rubbish. Your GPS device can determine your location with signals from just two or three satellites, but for elevation to be meaningful it needs a minimum of four satellites. And it’s still wildly inaccurate.
PR and KOM
After my morning commute, I had a couple of very surprising results which drove home the above points.
When I arrived at the office (and convinced Garmie to sync) I was stunned to find a KOM and a PR for essentially the same segment — the first climb of the day, a gradual rise along the western edge of Chuo Koen (Central Park). The top shows my custom segment (which explains why I have a KOM — no competition) and the bottom a public segment.
Close inspection shows that my custom segment gives the elevation gain as 9m, for a 2.8% average, while the public segment at bottom has an elevation gain of 15m, for 4.7%.
Now here’s the point — and try to stay with me here — the custom segment at top wholly contains the public segment below. Yet the the elevation data shows a 66.7% greater climb in the wholly contained segment.
As I was saying: rubbish.
(And far from being KOM, in the public rankings I’m just barely in the top 50%. Where I belong. Depending on the segment, I tend to be in the top 10%-50% … )
Baling wire and chewing gum
In (perhaps) totally unrelated news, the weather has been very good so far this week — excellent commuting weather. I didn’t ride Monday and Tuesday as I was coming off a 135km weekend ride, but I couldn’t resist the good conditions today.
Apart from the PR on the first little rise this morning (which I totally didn’t expect — I just felt like I was cruising, neither fast nor slow), I had another amusing result after my ride home.
Strava refused to match my ride with any of the 4-5 segments that are usually included in this route. When I prompted Strava to match one — a short but steep climb at the start of my commute home — it showed me the overlap between the public segment and my “as ridden” data, and refused to accept they were overlapping.
All the above is not really the fault of the GPS unit I’m using. Similar results (and discrepancies) can be found with any of the major brands. We’re triangulating* from an altitude of 20,200km to a base of a handful of meters using a device that weighs a few hundred grams, fits in the hand and has a battery life of 20 hours. When you’ve got a singing pig, you don’t complain that it’s occasionally off-key.
* A lot of people far more intelligent (and possibly less drunk) than I am have put an awful lot of effort into making GPS work, and my hat’s off to them. And I freely and humbly offer my apologies for the horrible over-simplified hash I’ve made of how the damn thing really works.
And the living is easy
Meanwhile, it’s almost summer. The weather looks good for a ride this weekend, and rainy season is on the horizon after that. I hope to get the riding in while it’s good. Following the 135km effort last weekend, I’ve surpassed 400km for the month for the first time since October 2021, and I’m on track to go more than 500km
