Does "corrected" GPS elevation ignore bridges?

Elevation data on the typical GPS trainers runners use is notoriously bad. You can take a look at the raw GPS data from a run through Kansas and it will look like the Himalayas. To get around the problem, many running websites allow you to correct the elevations by cross-referencing against known databases maintained by NASA and the USGS. The Garmin Connect website I use to track my runs can do that. Take a look at the elevation profile from today's run, for example:

On top is the uncorrected elevation chart, and below is the corrected version. Not only does the uncorrected version appear more "jagged," it also reports a hillier overall course -- a cumulative elevation gain of 472 feet, versus 245 feet in the corrected version. The corrected version definitely feels more like what I actually experience on a run. But is it possible that even "corrected" elevation charts can be wrong?

My standard hill workout ends with a tough quarter-mile climb. But when I use the corrected elevation chart, it simply doesn't show up. I know it's there: My lungs tell me that, but Garmin doesn't see it. I guess there's just a discrepancy in whatever database Garmin uses to make its corrections. But a recent comment on the Big Sur Marathon facebook page made me wonder if there was a consistent bias to elevation corrections.

The commenter said he had noticed that corrected elevations seem to always get bridges wrong. It makes some sense: The US Geological Survey isn't going to measure the elevation of the man-made bridge; it's going to measure the elevation of the earth (or water) below the bridge. I searched around the web to see if I could find any independent confirmation of this, and didn't find much, except for this forum post: "I shut off my elevation correction because it tells me that I only traveled + or -4 feet on each run and it's more like several hundred feet because I cross over a lot of very high bridges that are high enough to allow cruise ships pass underneath."

While it's just a couple of anecdotal reports, it makes sense, and it sure sounds like the corrected data is ignoring bridges! What would that mean for Big Sur? The original course had eight high bridges; I haven't been able to get a good number for the new course, but I think it's around five. Since the course is now an out-and-back, that's around ten bridge crossings. If a corrected GPS reading measures the distance to the ground below the bridge, when a bridge spans a canyon the way the bridges at Big Sur do, the corrected elevation change will be overestimated: The corrected version assumes you ran to the bottom of the canyon and up the other side, when in fact you crossed over a level bridge. Supposing these bridges are an average of 50 feet high, that means the corrected elevation gain for the course would be overestimated by 10 X 50 or 500 feet -- not a trivial amount! That would bring the cumulative elevation gain from the official course map down from 2,406 feet to 1,906 feet!

Details of today's workout are below.

Today the plan was to run a marathon preview: Get up about the same time I would for the actual race, eat the same food, wear the same clothes, and so on. Well, I didn't want to get up at 3:15, and anyways, with the time shift, that's more like 6:15 local time, so I compromised and awoke at 5. I ate a bowl of cereal at 5:10 and consumed a gel 15 minutes before the run, even though for a 6-miler I normally wouldn't eat at all beforehand. I also wanted to see if I want to carry a camera, so I loaded my new small camera in my Spi Belt, and drove into town to meet the group for the DART run at 6.

Surprisingly, Jeff was the only one there. I told him my plan was to run two easy miles, then two at race pace, and then a two-mile cool-down, and that sounded good to him. Our "easy" miles ended up being pretty quick, around 8:30 pace. Then I sped up and pulled ahead of Jeff. I wanted to see if I could take pictures without losing speed, so I pulled out my camera and snapped this self-portrait:

Hello, blurry world!

Uh, that's pretty bad. Maybe a camera isn't a great idea. On the positive side, I was able to snap it without slowing down. About a half mile later, Jeff had caught up to me and I tried to take a picture of the two of us:

Or, rather, the one of us

Hmmm... that's slightly better. Maybe I need a little more practice. It was also still fairly dark, so perhaps I'll do better in full daylight. On the plus side, carrying the camera didn't bother me much, so I think I will probably give this a shot, so to speak.

My two race-pace miles felt good, and the food I ate stayed down fine. On race day there will be a lot more time between when I get up at 3:15 and the actual race start at 6:45. I will probably bring a granola bar or two with me to eat while I'm waiting around for the start. It's just two days before I'll be heading to California for the race. I can't wait!