I didn't notice this for quite some time...my reports matched...I had 24.8 for the total ..which seemed short but possible: I was not surprised that the distance was not 26.2...that happens way more in trail marathons than road marathons.
The garmin often UNDER reads in hilly terrain with switch-backing...This is because it thinks you jumped across the switchbacks from one sample to the next. Only if it samples you at the exact turnback point does it get it right. So all the switchbacks get shortened.
Eventually I noticed the discrepancy ..and I knew what caused it right away.
When I finally started looking at my watch in the last few miles, I accidentally hit the 'return to start' option in the gps menu. This is the feature that tries to guide you home if you get lost by having you retrace your steps.
When I hit that, the watch went south and basically displayed 'Calculating track' for ages. The 'back' function didn't work..it was stuck. It did wake back up after about 15 minutes..just about a mile from the finish. I was thinking the entire run was going to be toast so I was very happy to see the watch go back to the normal display.
And it even guided me back down the trail..it told me before every switchback that there was "Turn Ahead" ..kind of cool actually.
Anyway, what happened was that the watch did not do the auto split for the mile that ended when it was busy crunching. It was still taking ground track data though...that's the highest priority task for the thing after all.
My software that parses the tcx files was using the lap data only for calculating the distances and times so my summary line matched the sum of the splits: therefore I was missing a mile but didn't notice anything wrong.
The Garmin software is smarter and always shows you the summary line from the trackpoints data.
I fixed my software to do something better than Garmin, IMHO, I take the extra segment and lump it into the previous split:
2012-06-23 Sat, San Lorenzo River Marathon
| ||||||||
|---|---|---|---|---|---|---|---|---|
Split
|
Dist
|
Pace
|
HR
|
% WHR
|
Cad
|
Elev+
|
Elev-
|
YPB
|
1
|
1.04
|
15:38
|
125.0
|
64.75
|
68.0
|
374.2
|
43.8
|
1.087
|
2
|
1.04
|
9:47
|
129.0
|
68.03
|
86.0
|
87.2
|
42.8
|
1.447
|
3
|
1.04
|
10:41
|
124.0
|
63.93
|
83.0
|
72.6
|
198.1
|
1.294
|
4
|
1.04
|
13:06
|
124.0
|
63.93
|
75.0
|
184.8
|
251.0
|
1.103
|
5
|
1.04
|
16:40
|
129.0
|
68.03
|
61.0
|
506.2
|
108.0
|
1.030
|
6
|
1.04
|
11:35
|
122.0
|
62.30
|
85.0
|
25.6
|
476.8
|
1.064
|
7
|
1.04
|
10:35
|
128.0
|
67.21
|
84.0
|
23.7
|
15.5
|
1.311
|
8
|
1.04
|
14:16
|
127.0
|
66.39
|
69.0
|
358.7
|
12.5
|
1.177
|
9
|
1.04
|
12:23
|
125.0
|
64.75
|
78.0
|
151.6
|
427.4
|
1.064
|
10
|
1.04
|
13:39
|
130.0
|
68.85
|
73.0
|
194.7
|
243.6
|
1.018
|
11
|
1.04
|
12:34
|
133.0
|
71.31
|
75.0
|
290.1
|
75.9
|
1.206
|
12
|
1.04
|
10:28
|
129.0
|
68.03
|
84.0
|
51.4
|
78.1
|
1.308
|
13
|
1.04
|
10:36
|
130.0
|
68.85
|
86.0
|
46.5
|
333.3
|
1.168
|
14
|
1.04
|
15:57
|
135.0
|
72.95
|
64.0
|
370.9
|
86.7
|
0.971
|
15
|
1.04
|
11:19
|
130.0
|
68.85
|
81.0
|
92.4
|
69.9
|
1.233
|
16
|
1.04
|
12:26
|
125.0
|
64.75
|
77.0
|
68.0
|
198.9
|
1.100
|
17
|
1.04
|
17:47
|
126.0
|
65.57
|
67.0
|
269.3
|
306.2
|
0.820
|
18
|
1.04
|
19:06
|
130.0
|
68.85
|
59.0
|
515.2
|
167.5
|
0.878
|
19
|
1.04
|
13:24
|
121.0
|
61.48
|
82.0
|
9.3
|
423.8
|
0.938
|
20
|
1.04
|
14:41
|
121.0
|
61.48
|
75.0
|
29.0
|
26.3
|
0.998
|
21
|
1.04
|
18:36
|
125.0
|
64.75
|
59.0
|
463.3
|
22.0
|
0.961
|
22
|
1.04
|
14:36
|
116.0
|
57.38
|
75.0
|
105.7
|
487.4
|
0.926
|
23
|
1.04
|
16:09
|
121.0
|
61.48
|
68.0
|
266.1
|
225.9
|
0.967
|
24
|
1.96
|
13:38
|
123.0
|
63.11
|
68.0
|
274.4
|
195.0
|
1.102
|
25
|
0.91
|
17:05
|
118.0
|
59.02
|
78.0
|
45.1
|
356.8
|
0.778
|
Sum
|
26.7
|
13:51
|
125.7
|
65.29
|
74.2
|
4876.1
|
4873.2
|
1.056
|
A nice feature I also have in my software is the ability to scale the miles and the times to match the actual race reported values.
I found that to match the recorded pace the RD has for me (13:51 m/m) I had to scale the race to 26.7 miles. (nowhere on his site does he report the actual distance). After scaling the mile splits become 1.04 miles. I have to admit that in a trail race the thought of running the tangents is very low down on my priority list 8). And there are a lot of curves.
Not only was it a good catch by yourself (and Arthur) that the total was off, however, you get additional props for figuring out that the switchbacks were the problem.
ReplyDeleteSuper interesting, I didn't know that about switch backs (I don't run a lot of trail I guess so I wouldn't.)
ReplyDeleteVery interesting about the switchbacks - I did a 10K trail race at Woodside in 2010 that had a ton of switchbacks and noticed that my garmin watch only read 6.0 miles instead of 6.2, this probably explains why my watch was short! I took my average pace anyways for the race, so I'm glad this confirms that my pace actually was the same then!
ReplyDelete