Vlog 2: I love my Garmin, but…

Sure it is

Back in August I wrote about my frustrations with running performance. At that time I thought I could improve my pace by pushing my heart rate beyond my typical 75% of Max. After seeing some very weird numbers coming from my FR35, I figured I was either experiencing a medically concerning pulse or there was something wrong with the device that captured my heart rate. To get the details, see below.

My Garmin lied and the truth hurts

My Garmin made up most of this route

Today's run (street): 3.75 miles

Halfway through today's run, I glanced at my Garmin and saw that I had covered two miles in about 17 minutes. That would have put me on track for my best training run in memory. I was puzzled because my perceived effort was nowhere near what I'd expect for that pace. I tried to rationalize the reasons for such a dramatic improvement in my performance compared to yesterday. It was 10 degrees cooler and cloudy, but could that account for running over a minute per mile faster?

When I got home and saw my time and mileage on the Garmin, I thought I'd rocked it. Maybe I was turning a corner with my training. After all, I used to regularly average 8:45 paces on my daily runs. Breaking 9:00 minutes on a run these days is a notable achievement for me. I hoped it was accurate and not some weird Garmin fail.

Despite that wishful thinking, it did turn out to be a badly confused GPS. For some odd reason, the Garmin put my starting point 3 miles north of where I began my run. Looking at the run data through Garmin Connect, my course appeared to have frequent 50-foot elevation changes. That's definitely not the case for my relatively flat route. I'm guessing that the low cloud cover may have interfered with the GPS signal and caused it to skip.

I Gmapped my route and was disappointed to see that I'd only covered 3.75 miles, rather than the 4.35 that the Garmin said I'd run. I wanted to run 4-5 miles today and thought I'd met my objective. The good news is that I beat yesterday's pace by 42 second per mile. The other good news is that it's a long weekend and tomorrow we're hosting a Runsketeer pool party. If the weather reports are accurate, the skies will be clear and I'll be able to get in a longer run in the morning.

Cadence increasing, visibility high

Artist's rendering using Google Maps street view

Today's run (street): 3.3 miles

I missed Tuesday's run and needed to make it up today, although the Higdon plan called for a rest day. I felt ambivalent about today's workout and whether to go easier this morning after yesterday's semi-fast run. My motivation for speed was low, so I decided to let my level of intensity play out as I ran. Once I got the door I noticed the cold, but the winds that made Wednesday's conditions feel like 24°, were no longer a factor.

My Garmin hasn't performed well lately in terms signal accuracy, making my real-time pace data suspect. I've been running by feel and heart rate which seem to provide more useful feedback. The type of workouts that I've been doing over the last five weeks have helped me increase my cadence. As a result, I'm finding it easier to hold a decent pace on shorter runs. I kept my heart rate around 85% today and ended up running my distance in the high 9's. That's exactly what I wanted to do.

Today's run was unremarkable, except at one point when I was running down a long road near my house. I spotted my wife's car heading in my direction. She and the kids were going out to do errands and when they passed me, the windows were down and they were cheering for "The Emerging Runner." They saw me from a distance because I was wearing a bright orange running shirt. High visibility isn't just for safety!

Lies, damn lies and Garmins

Look how fast I didn't run!

Today's run (street): 3.4 miles

I was in meetings all day on Tuesday and didn't get a chance to do a run. I did cover a lot of ground on foot, so that should count for something. Along the way I noticed many marathon tourists (after all these years I can easily spot them) enjoying the sights prior to running the race on Sunday. A couple of my meetings were held near Time Square, where sports demonstrations publicizing the Sochi Olympics were going on. It was a mob scene, but fun to watch.

I had no city meetings today, so I resumed my run schedule this morning. I've been running with my foot pod so that I can capture my cadence, but I still use the Garmin's GPS to measure my mileage. Due to that, I haven't bothered to calibrate the foot pod for distance. When I fire up the Garmin, it detects the foot pod and asks whether I'm running indoors. If I say yes, it will turn off the GPS radio and use the foot pod for measurement instead.

Today I went through the routine and when it looked like the signal had locked in, I was on my way. I hadn't gone half a mile before the Garmin chirped saying I'd reached my first mile. I figured that the GPS signal must not have actually acquired before I started and the watch was working off the (uncalibrated) foot pod. I didn't care much, because I always Gmap my run to get exact distance.

The watch did switch to GPS mode shortly after that, and my remaining splits were in line with my normal pacing. While I would have liked to meet the performance that the Garmin recorded for today's run, I must admit to a far less impressive pace in the mid-9:00 range. So the Garmin lied, but I'll forgive it. If I could run five minute first miles for real, I might actually break an 8:00 pace on my training runs.

One of these routes is not like the other

Post run relaxing

Today's run (street): 3.36 or 3.41 miles

It was another mid-high intensity run this morning. My heart rate (the only performance metric I'll view while on a run) was close to target, and I knew I'd be happy with my finish time. A quick look at the Garmin confirmed that I'd managed another good run. I cooled down by the pool when I finished and tried mapping my route using the Gmaps web app on my phone

I was pleased to see that the Android OS was finally supporting Gmap's functionality. That allowed me to record my distance and calculate my true pace. I liked what I saw: 9:05. Although I worked for it, getting there didn't feel as hard as I thought it would be. Looking a gift horse in the mouth, I mapped the run on my laptop after I'd finish showering and discovered a surprising disparity in distance.

Distance: Android version

Web version, same route

For some reason, the route on my laptop registered 5/100ths of a mile shorter than the exact same map on my phone. Weird and disappointing, because that small disparity meant my average pace was actually 9:12. Not bad, but not what I wanted to see. I remapped the route to see if I'd somehow missed a street, but the distance came up the same each time.

So what if the phone mapping is actually the correct measurement? That would be nice. But I use the web version as the standard, so I'll have to go with that. In the end, the difference is measurement is minimal. Unfortunately, the difference in pace is not.

My Garmin speaks the ugly truth

It's a match

Today's run (street): 3.25 miles

I'm incredibly annoyed with my Garmin today. I did this morning's run and, as always, recorded my distance using GPS. I always assume the watch will under-count my distance, so I wasn't upset to see the indicated time and pace. Usually, after Gmapping my route, I'll need to add about 3% more mileage to the calculation. That often makes the difference between a good and mediocre pace. Today, both the Garmin and Gmaps said exactly the same thing, and what they said wasn't good. I missed my targeted pace by 23 seconds per mile.

There's no reason why I ran this slowly today. I'd tried to focus on form and turnover. The weather was perfect, so heat and humidity were not a factor. Perceived effort was on par with my better runs and I even ran the last quarter in a semi-sprint. But in the end, my performance did not match up to expectations. Tomorrow is my last run before Sunday's race, so I need to make it count.

Zeotrope concept (left), example (right)

I took most of the day to be with the family and we spent it at the Museum of the Moving Image in Astoria, Queens. The experience was great, and I even got to play with a nineteenth century zeotrope of a man running. But instead of marveling at this 100 year old wonder of technology, I was watching his form and thinking how he should shorten his stride.

Rethinking GPS versus foot pod

Today's run (treadmill): 2.5 miles

My first method of capturing running data was with the Nike+ chip that fit into a concave spot located under my shoe's sock liner. The accuracy of the this system was surprisingly high, but the software was buggy and the wristband that displayed metrics like pace, time and distance had serious corrosion issues. After going through three of these wristbands in less than a year, I got my money back and bought a Garmin FR50.


FR60



The Garmin 50 (and after that the FR60) uses a foot pod that works in a similar way to that Nike+ chip and I got used to tracking my distance and pace that way. The foot pod needed to be calibrated each time I switched running shoes (in my case, frequently) but the accuracy was very high. I started running with the Saucony Hattoris that have no laces to hold a foot pod, and made the switch to the Garmin FR210 GPS watch thinking I'd be upgrading my experience.

As it turned out, after almost two years, I've discovered I've given up more than I've gained by switching to GPS. The accuracy of GPS (~ 3%) is far worse than with the foot pod (~ 1%). The foot pod also captures cadence, an important metric, but the FR210 does not.



FR210



I had an amusing experience on the treadmill with the FR210 this morning. I wore the watch to capture my heart rate but, even indoors, it had locked in on satellite. When I finished my run I saw that the watch had recorded my distance at .14 miles. I've been considering using the FR60 again with the foot pod for treadmill runs. But for outdoor runs, I have to say the one big advantage of using the GPS watch is that there's no fussing with calibration or switching foot pods. Nothing's perfect, but at least I have a choice.

The declining accuracy of my Garmin FR210

I'd add 3% just to be safe

Today's run (street): 2.53 miles (Gmaps measured)

I'm not sure why, but my Garmin FR 210 has been under-performing lately. It was especially bad this morning. I generally run the same route every day at 4:00 AM and, after careful measurement, I know that the distance is exactly 2.53 miles. The Garmin's margin of error is generally -3%, which means that it under-counts to that degree fairly consistently. Since it is consistant, I accept that variance and correct for it in my pace calculation.

It's no mystery why this happens. Looking at my run captured in Garmin Connect, I can see that the watch will vectorize corners and straighten out curves. This is due to the capture time between GPS signals (approximately one sample per second). If the signal was captured continuously, the course would be displayed accurately, with no corners cut, etc. When it's cloudy, the signal can get interrupted. When that happens, the watch interpolates the distance between signals as a straight line. Enough of those and your accuracy really suffers.

Lately my FR210 has been under-reporting by about 5% and this morning it came in at -7%. It was cloudy, but clear enough to see some stars. My run could only be described as slow, and with the under-counted distance, my watch recorded my pace as glacial. Correcting for the true distance, it was still the slowest 4:00 AM run in memory.

I'm puzzled as to why my Garmin's GPS has become less accurate of late. Perhaps it's just going through a bad spell. The GPS accuracy tends to be better when I run a mostly straight course like at Bethpage. I'll see how it does tomorrow.

In treadmill I trust. Or do I?

Today's run (treadmill): 25 minutes

This is the key

A rainy morning kept me inside today so I did yet another treadmill run. I'm curious to know how close the readout on the treadmill's display matches my actual progress and performance. I replaced the treadmill's console a few months ago but wasn't able to run the calibration sequence when I installed it. Every time I run on the machine I question the accuracy of its recorded distance and speed.

I've often felt that running on this treadmill (a Sole F63) and our prior Pro-Form unit felt harder at a given speed than when running on the road. A 9:00 pace on the Sole feels like about 8:50 on pavement. It may be that the treadmill forces a shorter stride and higher cadence that feels more challenging. I would think that the treadmill would feel easier and street running because there's no wind resistance. Plus, you don't experience hills unless you select an incline.

I still have my foot pod that links to my Garmin and I could probably answer my accuracy questions by comparing the Garmin performance with the Sole's on the same run. Before I do that, I need to replace the foot pod battery and do some calibration runs outside in order to establish an accurate baseline. The other issue is that the foot pod won't attach easily to my Hattori's so I'll need to run in a different shoe. I guess I can make excuses why it will be hard to test or I can just go ahead and try it.

News flash - GPS watches aren't accurate



Graphic from 12/19 NY Times article
Photo credit: Martin Strauss, via Garmin and Microsoft

 Today's run (street): 2.5 miles
There was an article in Tuesday's New York Times about the accuracy (or should I say, inaccuracy?) of GPS watches. The writer made the same point that I've been making for years, that GPS technology does not provide exact measurement. The same can be said for smartphone GPS apps that show lots of metrics and graphics, but also under-count distance. The thing is, if your distance is off, so is everything else, including pace.

I'm a bit of a belt and suspenders guy when it comes to tracking my runs. I use a Garmin FR210 GPS watch to capture my run metrics and route. But later, I usually measure where I had run using Gmaps to get the most accurate distance. This way I can calculate my true pace. Why use the Garmin if I know it's inaccurate? Good question. But just try taking my Garmin away from me...

This morning I got back outside for my early run and was rewarded with dry roads and mild temperatures. There was wind coming from the north that, when traveling head-on, made things chilly. Most of my route went either south or east so it wasn't a problem. I managed an average pace, though my PE (perceived effort) was higher than that result. Then again, when you get your performance information from a GPS watch, you never know what to believe!

Garmin 210 - trouble keeping track on the track

A shot of the track from this morning

Today's workout (track): 8 x 400's plus 3 mile run

Since I'm racing again next Sunday, I wanted to get in a little speed work before I taper my training next week. This morning I headed over to the local high school track to run some intervals. There were a handful of runners and walkers already on the track when I arrived but the lanes were fairly clear. I brought a stopwatch as well as my Garmin and I ended up using the FR 210 to record my runs and used the stopwatch to time my rest periods between intervals.

I didn't run the intervals hard. My goal was to do mile-equivalents a little faster than my 5K race PR and I managed to average 8:06 across eight quarter miles. I hoped that would recruit enough fast twitch fibers to give me some speed when I ran the following three miles (5K actually) and it did. I averaged 8:45 for that run.

A clear margin of error

The Garmin did not do a good job capturing my route. The picture above is a grab from Google Earth using the imported KMZ file. I stayed primarily in a middle lane but the GPS interpreted that much differently. Even though the 210 is always a little off I still like it a lot. It's great to be able to run without swapping the foot pod every time I change shoes. Especially in the case of the Hattori's that don't even have a place to attach a foot pod.

I'm really happy with today's workout. While I'm still not speedy, I've proven to myself that I can run sub-9:00 paces when I need to. The Hattori's were interesting to use on the track and they responded well when I took off in a sprint to begin each interval. I haven't decided what I'll do for tomorrow's Memorial Day workout but I am hoping to fold in a bike ride or two before I return to the office.

Garmin FR210 - your mileage may vary

Example of 210 cutting corners, under counting distance

Today's run (street): 3.92 miles

As expected, the rain started last night and continued into the morning. At 7:00 AM I was considering doing an extended elliptical session rather than suffering a soaking run. By 7:30 the rain had slowed to a drizzle and I went outside in hopes of running 30-40 minutes before the next deluge. Yesterday's Stillwell workout was rough, in the best of ways, but I felt no residual effects of it this morning.

It didn't take too long to acquire a signal on the Garmin and I took off after a couple of minutes of dynamic stretching. I still had some pain in my left knee but it wasn't too bad. At this point it's an irritant more than an injury and it usually goes away after a few minutes.  I felt that I was moving fairly well at the beginning but according to the Garmin I was running slightly over 10 min/mile. By now I know the difference between a 9:30 and 10 minute pace and I also know that the Garmin, at best, under-counts my speed by 3%. I didn't worry too much about my pace but I began to pick things up by mile 2.

After verifying my true distance on Gmaps I saw that the Garmin had under-recorded my run by 6.8%. An examination of my route recorded on the 210, using Garmin Training Center, showed the many variances from my actual route that added up to that margin of error. I'm guessing the heavy cloud cover may have interfered with the GPS sampling frequency. I was hoping that this watch would provide a closer margin than -3%, which seems to be the average variance. I'll just expect to have an even greater variance on cloudy days.

Garmin 210 -- upload troubles but consistent inaccuracy

Today's run (street): 2.5 miles

Yesterday's run, as recorded by the Garmin 210, came up short when compared to route mapping on both Gmaps and Google Earth. The variance in distance was about -2.5%. That seems close when viewed reciprocally (being 97.5% accurate) but I commonly saw less than +/-1% variance with a calibrated foot pod. I tried to upload my run to Garmin Connect so I could view the GPX and KMZ files that would show the recorded vectors against a Google map and, perhaps, reveal where the GPS capture cut corners. Unfortunately that attempt didn't work within the time I had to try it so I'll fight that battle this weekend.

I went out today and followed the same route as Wednesday, using yesterday's distance as a benchmark. Both runs felt about the same but I ran about 39 seconds longer this morning. The Garmin recorded the route .01 miles less than yesterday's distance so at least the 210 is consistent in its inaccuracy. I'm disappointed with Garmin for having such poor documentation related to uploading runs and I'm also disappointed to discover that the GPS is under-recording my distance. But I do like the watch and I'm certain that I'll get it all figured out eventually.

First run with the Garmin 210

Now it's just me, the road and 10 satellites

Today's run (street): 2.5 miles

Last night UPS delivered my new Garmin 210 and I was excited by the prospect of doing my morning run free of the foot pod and calibration concerns. The foot pod wasn't necessary but after the watches' first use I do question the overall accuracy. I bought the watch online this weekend -- the basic 210 without the foot pod/HRM bundle. I got a great price and free delivery two days later. I un-boxed the watch and plugged it in to charge the lithium battery. Setup took seconds and the watch automatically acquired the date and time via satellite signal.

This morning I was excited to try the watch. My only experiences with GPS tracking were with various GPS apps on my iPhone (bad) the QStarz Sports Recorder (worse). I stepped outside and activated the satellite receiver and the 210 grabbed its connection in seconds. My iPhone usually takes minutes to get a GPS signal. I hit the start button and off I went. My knee is still bothering me but after getting the "okay" to run from the doctor yesterday I accepted the mild pain. Three minutes into the run the pain disappeared and I was able to focus on my form. I think my stride was a little unbalanced at first but I managed to pick up my pace without a problem.

I passed mile one a short distance after my known benchmark but it was close enough to think the GPS's accuracy was in the ballpark. I covered the first mile in 9:27, mile 2 at 9:01 and the last half mile at an 8:50 pace. Those were my calculations based on careful retracing of my route on Gmaps. The 210 recorded my distance .07 miles (2.8%) less than that. I'll be curious to see if the accuracy varies consistently. I'm hoping that an examination of the GPX file overlay on Google Earth will show where the 210's GPS cut corners to end up short. Unless of course, it's actually Gmaps that's off...

AllSport GPS thinks I'm a slacker

I'm in taper mode as I move closer to Sunday's race and I'm also experiencing some discomfort with the outside toes on both feet. I decided that I would get a run in today but take it easier than I normally would. I brought along my iPhone and selected the AllSport GPS app to track my run and took the Garmin 50 along as well. I now have five GPS apps on the iPhone: Allsport and MotionX (paid apps) plus iMapMyRun, Run Keeper Free and RunGypsy (free apps) and they all capture data about run speed and distance. What separates these apps is what else they track and how they report data. The free apps generally give you pace and distance information and some rudimentary resources to store that data. The paid apps have lots of features, MotionX really provides a lot of utilities for the running experience including compass, mapping and music integration. AllSport generates an excellent report that shows many metrics including elevation (as does MotionX) and both paid apps export KMZ and GPX files that overlay on Google Earth.

The thing that these apps all have in common (though I have not yet tried RunGypsy) is a dependence on the iPhone's GPS service which has proven lacking. Each time I have captured a run with one of these apps I've seen differences between what's reported and what was actually run. My baseline is Google Earth where I can trace my route to the inch and compare it to the route created by the GPS apps. This morning I verified on Google that I covered 2.15 miles and the Garmin confirmed that. The AllSport app reported that I ran 1.98 miles and I can understand why after reviewing the route it displayed. There seemed to be some weak links from the signal (that were shown as yellow lines on the map) and the lines veered off the road enough to explain the shorter distance. What puzzles me is AllSport's insistence that I rested for 7:17 of my run.

Now I know I wasn't pushing too hard today (my actual pace was 9:39) but I don't recall resting during the run. It's not clear if that was a result of a weak signal or if AllSport was judging me for not picking up the pace (yes, I'm kidding). I sent them a note asking about it and I expect that I'll hear back. To their credit they've been good about responding to my questions. I like this app a lot but I want to rely on it and at this point I don't think I can. I'll try to use it on Sunday so I can capture the route but I'll have my Garmin 50 along to keep things honest.