Duncan Ridge Trail 50k Course Profile

Trolling through UltraSignup's Southeast races, I found the Duncan Ridge Trail 50k/30k and it looks spectacular! Here's the description from UltraSignup:

The Duncan Ridge Trail 50K/30K will be one of the toughest 50Ks in the Southeast with more than 10,000 feet of elevation gain. The actual elevation profile is displayed in the logo, but that is just the “out” of an “out and back” race. This race uses the Coosa Backcountry Trail and the Duncan Ridge Trail, which is considered one of the toughest trails in Georgia. The race will start and finish at Vogel State Park's Lake Trahlyta in the beautiful North Georgia Mountains. There will be a 30K option as well. If you think you're tough enough, sign up April 1st!

They may have a few spots left for this year's November 23rd race, so head on over to UltraSignup if you're feeling brave.

The plot below compares the course profiles of the DRT, the StumpJump, and the UpChuck, along with notable ascent and descent sections colored in red and blue.

Click on it for a larger view.

Whoa! The DRT course profile puts the StumpJump and even the UpChuck into a new perspective! This Tennessee boy might have to mosey on over to Georgia next year. I'd go this year, but I've already started my StumpJump 2014 training (up to 20 miles a week, wohoo!).

Also, here's a few tables listing ascents and descents corresponding to those red and blue triangles on the plot. I've only included climbs that are longer that 2000 feet containing at least one section of greather than 10% grade.

DRT

Start	End	Distance	Average Grade
0.59	1.30	0.71	0.11
1.30	1.96	0.65	-0.05
2.39	3.23	0.84	-0.11
6.00	6.52	0.53	0.19
6.68	7.18	0.50	0.23
7.64	8.51	0.87	-0.16
8.76	9.23	0.47	-0.10
10.04	10.87	0.84	-0.08
12.37	12.83	0.47	-0.23
13.05	13.45	0.40	0.20
14.51	14.94	0.43	0.12
15.35	15.81	0.47	-0.12
16.65	17.24	0.59	-0.17
17.46	17.93	0.47	0.24
21.81	22.62	0.81	0.18
23.12	23.61	0.50	-0.25
23.77	24.30	0.53	-0.19
26.19	26.59	0.40	-0.16
27.28	27.84	0.56	0.11
28.92	29.70	0.78	-0.10

StumpJump

Start	End	Distance	Average Grade
4.54	5.10	0.56	0.14
10.25	10.78	0.53	-0.06
15.16	15.66	0.50	-0.03
22.59	22.99	0.40	-0.11
23.83	24.26	0.43	-0.12

UpChuck

Start	End	Distance	Average Grade
12.49	12.92	0.43	0.23
12.92	13.55	0.62	-0.10
24.70	25.26	0.56	-0.08
25.51	26.25	0.75	-0.11

Predicting Your UpChuck Time with Your StumpJump Time

Over on Upchuck's Facebook page, the organizers commented that a good rule of thumb for average runners is to add 30-40 minutes to their StumpJump time. So, using the finishing times of both races (thanks again to UltraSignup) I created a very simple prediction model with a table and plot below. I found all runners who have finished both races, averaged their times, and ran them through an OLS model. If you want to see what a real statistician can do with this data, check out Dr. Chris Fonnesbeck's analysis for something a little more stringent (much thanks to him as he's mentoring me). I am showing you the results of my prediction model below as it's a little less intimidating, and it bears very nearly the same results as Fonnesbeck's.

I wish I had more time to spend on this, but I've run out and I wanted to get it out there before the race tomorrow. Head over to the Facebook page if you want to add to the discussion.

Prediction Table

So the table below has two recommendations. I call them DirectorRecommendation and ModelRecommendation. I've simplified the Director's recommendation to add 40 minutes as it gives you the more conservative estimate rather than the 30-40 minute range. So for example, if you've run the StumpJump in a time of 4 hours, the Director expects you to finish in 4:40 while the Model expects you to finish in 5:00. Interesting, a difference of 20 minutes. As you continue reading down the table, you'll see that for a StumpJump time of 5:30, both the Director and the Model agree that you will finish in about 6:10, no difference. Reading further if you finished StumpJump in 8:00, the Director expects 8:40 while the model expects 8:05.

StumpJumpTime	DirectorRecommendation	ModelRecommendation
04:00:00	04:40:00	05:00:46
04:30:00	05:10:00	05:23:51
05:00:00	05:40:00	05:46:57.1
05:30:00	06:10:00	06:10:02.3
06:00:00	06:40:00	06:33:07.5
06:30:00	07:10:00	06:56:12.6
07:00:00	07:40:00	07:19:17.8
07:30:00	08:10:00	07:42:22.9
08:00:00	08:40:00	08:05:27.11
08:30:00	09:10:00	08:28:32.12
09:00:00	09:40:00	08:51:37.14

Plot

The DirectorRecommendation is the red line, while the ModelRecommendation is the blue line. Start by finding your StumpJump time on the X axis with your computer mouse, scroll up to either one of the lines, then scroll to the left to get your predicted UpChuck Time.

Rock/Creek's StumpJump50k Vs. UpChuck 50k: Elevation Wars

It's very hard to get an idea of how difficult an ultramarathon course might be just by reading an elevation profile. Hopefully, it will show the cumulative elevation gain/loss, and it better show the altitude at which you'll be running. But it's really hard to judge how steep certain sections are. For example, the StumpJump50k map looks like shark teeth. Many of the hills look straight up and down, but this is due to the map's aspect ratio. Courses that are really long get squeezed so much that their hills can look incredibly daunting when in fact they may be gently rolling.

Using my course data from RunningAHEAD, I've created the following elevation maps comparing the StumpJump 50k to the UpChuck 50 (yet another excellent Rock/Creek race, with a storied beginning). The first shows the entire course for each race, while the subsequent maps show 5 mile sections providing a more accurate portrait of each climb. I've also developed a crude algorithm to detect notable sections with at least 100 meter ascents or descents at 10% grade. The algorithm is a work in progress, but it does a nice job characterizing the StumpJump so I thought I'd see what it would do for the UpChuck.

Also check back next week as I'll provide a summary of UpChuck's historical finishing times in anticipation of it's next running on November 9th.

DISCLAIMER: Elevation data are notoriously inaccurate so don't read too much into these results. Feel free to download the code and data and explore yourself.