I finally figured out the mystery of the inaccuracies between the AllSport GPS and the Garmin Forerunner. If you are not sure what I am referring to, please read my previous post on the topic.
I have been running with two devices--the AllSport GPS software on my Blackberry, and the Garmin Forerunner 205 watch. I noticed when I started comparing the information collected by the two devices that they did not agree on distance--sometimes up to a half mile off on a six mile run. So I set out to understand why and to figure out how I could fix the problem.
I kinda stumbled on the cause and the solution. Let me explain. I was reading an article on improving GPS accuracy when I decided I needed a simple test to see which device was recording accurately. I mounted both devices on my bike and then rode around the neighborhood, staying on the sidewalks, so that when I uploaded the data I could see it overlayed on the satellite images. I thought this would give me a pretty good idea about which device was more accurately recording the data.
After riding around for a few miles I went back home and reviewed the recordings from both devices. The AllSport GPS was VERY accurate. It was almost spooky how close it recorded my ride. However, I had a very different reaction when I looked at the Garmin software. In spots it was WAY off... like by 10 feet or more. At some spots it looked like I drove my bike out onto the road--but I didn't. This seemed really strange to me. If you read my last post on this topic you will remember the results were the exact opposite on my last test... the AllSport GPS was way off in some areas and the Garmin appeared to be very accurate.
I thought about it for almost a full day when the solution finally came to me. Why would one device work so well one time and so poorly the next time. I thought about what was different in the two scenarios. Yes, I was in the woods one time and on my neighborhood streets the next. But I ran the test with both devices at the same time in both locations, so location should not have made a big different. What else was different? That's when it came to me... I was running in the woods and riding my bike in my neighborhood. Now wait... before you call me crazy and stop reading, let me explain why this difference is significant.
In the article I read, the author went on at length about the importance of being able to see the sky and lock into satellites. Makes sense... the more satellites you can lock into the more accurate the information would be. If one device was more accurate than the other, than you could conclude that device had a better lock on more satellites. So why did one have a better lock than the other? The answer is so simple it's almost silly... the device with the better lock had a better view of the sky!!! Yes, I told you it was so simple. Let me continue to explain.
When I run, I attach my Blackberry to my water belt inside one of the pockets. In this position, the Blackberry has a limited view of the sky. But my Garmin, attached to my arm, has a clear shot at the sky. In contrast, when I attach both devices to my bike, the face of my Blackberry points upward, giving it a clear shot at the sky. However, in order to see the face of my Garmin I rotate it towards me, which in turn points the GPS receiver toward the ground. So while running my Garmin was getting a better lock, and while riding my bike the Blackberry had the better lock.
To test this theory I decided to run with both devices pointing up at the sky (yes, it sucked holding my Blackberry and Garmin out in front of me while I ran, but hey...). I ran for about 2 miles. This time the distance difference between the two was minimal. The AllSport recorded 2.03 and the Garmin recorded 2.07. Now we're talking. But this created a new problem. No way am I going to continue to run with my devices held out in front of me. Not only would my arms get tired, but I think I would look pretty stupid too!!!
So on the way back I returned my Blackberry to its home in my water belt, strapped my Garmin onto my arm, and prepared for another quick test. I reset both devices and then ran back to the start, following the exact same course I had taken on the first test. My assumption was that if the Garmin could see the sky clearly then it should record the same distance on the way back as it did on the way out. I also assumed that the AllSport would record a different distance because of its inhibited view of the sky. Sure enough, when I arrived back at the start the AllSport was off. It recorded 2.03 out and 1.85 back. That's a big difference over two miles. However, the Garmin recorded 2.07 out and 2.08 back... pretty darn close.
OK, mystery solved.
Well, since I didn't feel like going home yet, and one test really isn't enough to be certain about my assumption, I decided to do a quick 1 mile out and back on a different part of the trail. Same setup... Blackberry in my water belt pocket and Garmin on my arm. Similar results. The Garmin recorded exactly 1.00 out and 1.00 back. The AllSport, however, recorded 1.09 out and 1.61 back--an obviously unacceptable variance.
My conclusion... I think both devices are very accurate IF the GPS receiver has a clear shot at the sky. However, because I do not plan to run holding my Blackberry out in front of my, the Garmin Forerunner 205 will be my GPS tracking device of choice as I continue to train to run my first ultramarathon.
Hey, I just realized that I ran 6 miles during this test. Cool!
August 12, 2010
Subscribe to:
Post Comments (Atom)
All rights reserved.
Contact The Trail Jogger
Very interesting observation Thom...and one that I probably would have never figured out. And the mental image of you running down the trail with both arms extended, a device in each, is just downright hilarious. All in the name of science and technology though, right?
ReplyDeleteGreat analysis and one that I'll consider if I ever splurge for GPS. It clearly makes a difference where the device is.
ReplyDeleteGreat analysis, however as you get more into this stuff dead on balls accuracy simply won't matter.
ReplyDeleteI also found that if the Garmin does not complete it's 'locating satellites' function, it will give you an inaccurate reading.
ReplyDeleteIf in the next 20 years I buy a Garmin or Use a gps device on a run, I will remember this post. Who, knows I might buy one next year when I save up enough running bucks (bucks earned according to the mileage I run put into savings....26.2 cents per mile) Thanks for sharing man.
ReplyDeleteThomas: Kris from AllSport GPS/Trimble Outdoors. Thanks for the post. The distance recorded depends on several things: the view to the sky, if the GPS signal gets dropped, and the accuracy of the GPS chip. Distance also depends on how the GPS device or GPS phone software creates and measures its tracklog--the digital route it draws as you run. If you look closely at a tracklog in map software, it is basically a series of dots connected by a line. In theory, the more dots, the more accurate the distance will be. Some devices record by distance, so if you move say 52 feet it marks another dot, move another 52 feet it lays down another dot, etc. Over the course of a mile it would draw 100 dots. If the device collects at intervals of ever 250 feet it would only draw 21 dots over a mile. So in the case of 250 feet, you might miss details like a switchback on a trail; it will draw a straight line instead of nice curve. Other devices record tracklogs by time (usually every second or so). AllSport draws tracklogs based on distance and you can change the setting under Options>Settings>Update Distance. Anyway, different devices use different distance settings based off of many factors: what kind of activity is being recorded, what kind of memory capacity is required for the device needs to run, and how much battery power is needed to draw tracklogs. Lots of details. Hopefully I provided more insight than confusion.
ReplyDeleteAs far as using your BlackBerry on runs, many AllSport GPS users use armbands. Personally, I slide my phone in my bike jersey side pocket. Hope this helps and good luck with training.