Elevation Confusion
A few years ago my friend Mike and I wrote a facetious Monterey Herald article that was a somewhat fictionalized account of our frustrations with using GPS devices in everyday training. Last month, we had occasion to write more seriously on the topic, but the outcome was no less frustrating than the first time around - in fact, it was probably worse.
What started as a simple inquiry about the modified Big Sur Marathon course turned into a trip down the analytical rabbit hole, where we saw and learned more information than we possibly imagined – not to mention about 100 times more than we wanted. Fortunately we had a pretty reliable tour guide to help us make partial sense of this strange new world, but despite his best efforts, the whole process was something that Mike and I couldn’t escape quickly enough.
We stuck around long enough to make an article out of it, though, which follows below. One more note about this post which has nothing to do with the topic at hand: Mike wrote the first draft of this one, and took a page from my playbook by opening with a song quote. I don’t know if he purposely picked a tune that accompanies one of my favorite music videos of all time, but that’s the way it turned out. The video is after the post.
**
Running Life 5/5/11 “Elevation Confusion”
“You don’t need a weatherman to know which way the wind blows.”
- Bob Dylan, “Subterranean Homesick Blues” (video after post)
Any runner will tell you that it’s easy to figure out when you are running into a headwind, or to determine whether you are headed uphill or downhill. You can also be relatively certain about the distance you’ve traveled, especially if there are mile markers on the road or if you’re wearing a GPS device.
But how high are those hills you climbed? And how much climbing have you done over all the miles you’ve measured? Answering those questions sometimes requires an advanced degree – and even then, you’re probably not certain of your accuracy.
Case in point: prior to our Big Sur Marathon preview article, we attempted to quantify the differences in climbing between the traditional course that features Hurricane Point and the out-and-back course that includes the rolling hills of Carmel Highlands twice. Researching the subject was one of the most mind-boggling ordeals we’ve experienced lately.
We started with the elevation profile on the race website (admin note: this page has now been taken down), which indicated that the total climbing over the new course was 2400 feet. The data came from a Naval Postgraduate School scientist who used a USA Track and Field website mapping tool. He explained to us that the elevation information is stored in US Geological Survey NED database tables, and then extrapolated over a known distance (in this case, 26.2 miles). He also said that this could differ from GPS measurements, as hand-held GPS units update position and elevation on a scheduled interval which might impact the fidelity of the data (seriously – those are his words).
Another tech-savvy (our polite way of saying geeky) friend of ours used his Garmin GPS watch for two previous Big Sur Marathons as well as this year’s modified course, and shared with us his GPX files, which plot latitude and longitude points alongside data from the USGS database. His readings showed approximately 1750’ of climbing for the standard course, and 1630’ on the new course. He disputed the idea of inaccuracy due to delayed reporting, and noted that his dataset contained 3425 points in it – approximately one measurement every three seconds, or every 40 feet of road.
 |
| 3425 data points plotted over a 30m NED dataset ... or something like that |
Then he shifted into high gear, and explained that position readings typically aren’t updated on regular intervals, but are dependent on the rate of change. In other words, the faster your position or elevation is changing, the shorter the recording interval becomes in order to enhance accuracy. He also advised us that handheld devices rely on triangulation of satellites in the “GPS constellation” for accurate position reporting, as well as the visibility of a 4th satellite to add the elevation component. Finally, the fact that the dataset was taken on the side of a cliff (the race course on coastal Highway 1) might be problematic, since positions just a few meters away on either side of the road could have statistically significant differences in elevation.
It was right around here that our heads started to spin.
The discussion wasn't finished, however … because our friend’s data from his two Big Sur Marathons on the standard route also deviated by about 100’ from each other. When we asked him to account for this, we inadvertently opened the floodgates to a whole world of fractals, interpolations, calibrated barometric variables and fluctuating weather permutations, smoothing algorithms, and numerous other scientific conditions that we couldn’t begin to comprehend. Suffice it to say that any elevation data you see in course profiles is going to have a degree of uncertainty – in some cases, quite a significant amount.
Fortunately, none of this distracted from the task at hand on race day for our tech-savvy friend, who ran 2:53:05 for 15th place overall at this year’s Big Sur Marathon, using his Garmin GPS watch as usual. It’s worth noting that his time this year was virtually identical to his personal record from the standard course, even though there was (according to his watch) slightly less climbing this year. External conditions such as wind and air temperature might have made an impact as well, but honestly, who the heck really knows?
All that the two of us learned for sure in this process is that we prefer to stick to simple considerations like knowing which way the wind is blowing. From now on we’ll just refer to our local marathon as a hilly, challenging course, and leave the elevation claims for the scientists to figure out.
**
As for the video: this one is often imitated, but never duplicated. I’ve always considered both the song and video clip to be two of the best snapshots of Bob Dylan’s overwhelming hipster genius in the prime of his career.
Bob Dylan, "Subterranean Homesick Blues" (click to play):
Get updates as soon as they're posted! Click here to subscribe to Running and Rambling.
Check out the Running Life book for a collection of our most popular columns.
5 comments:
I am no expert on much of anything but have often read that Garmin data can be faulty. After each run I dump my Garmin data into WKO+ and then use the correction option to fix the elevation numbers. Usually that fix is substantial.
When I eyeball the results, they do seem more in line with the actual topography of my running route.
Once the data is analyzed, processed, manipulated and adjusted, expectations govern everything.
BTW: a 10-second PR is not "virtually identical" to the mathematically inclined, unless your expectations were greater by a factor of n to the 10th.
As one who is relatively obsessed by vertical gain info, I thought I'd put my 2 cents in.
I've used an Avocet altimeter watch for nearly 20 years. I put otgether a spreadsheet running log that allows me to track vertical gain along with the other pertinent info. It's allowed me to keep track of total vertical miles for each year.
My initial goal for a given year is a mile per week (52 miles) of vertical ascent. I've found that I always start off the year slow, running on roads in the winter. Once the trails open up, I usually catch up sometime over the summer, then try to hang on once the snow hits again in the fall.
Once I achieve the initial goal, my second goal is a mile per year of my age (currently 59). Yes, it gets more challenging every year. But 2010 was my best year ever... 66 miles. Not bad for an old guy!
I think the altimeter watch is pretty accurate. I always set the initial elevation (I live at 4100 ft). The watch reading changes from day to day due to atmospheric changes, with high/low pressure systems moving thru. If I really wanted to be accurate, I would check my finishing elevation watch reading. It may have drifted a bit as the pressure changes, so I should correct accordingly. I've never done that, preferring, conservatively, instead to chop the number to the nearest 100 ft reading (e.g., 1290 ft of ascent on the watch = 1200 ft in my log).
I've used GPS a bit, too. The problem with GPS is the way they drop satellites. It can cause major errors in both distance an ascent. The best quality control I've found is being able to download the GPS data to mapping software (National Geographic TOPO!). It will display the track and errors will be apparent. If the track looks good, then I give the numbers a lot of credence.
The problems show up when too many satellites are dropped. The GPS is recording waypoints, then it will stop recording waypoints until a sufficient number of satellites are recovered. The recorded track display on the mapping software will show a straight line between those two waypoints. This will often cut off part of the course, making the distance short. If (for example) the trail was contouring around a hill, connecting the two points will cross several contour lines that the trail does not, adding a "hill" to the elevation profile, with erroneously more ascent.
Martin: Thanks! Very informative.
My head would be spinning too. I did not realize how much it could all fluctuate like that. Thanks for sharing the great information! And Bob Dylan!
Liz Lembeck-CoolProducts
Post a Comment