## Accuracy In Estimating Elevation Gain

The primary factor in determining the amount of energy needed for a hike is the total elevation gain. While being off by a few hundred feet is usually not a big deal, being off by a 1000 ft or more is.

Of course, by looking at topo maps we can certainly get a good enough estimate of how much elevation gain there will be, but with GPS devices people have taken to assessing how much gain they have done on a hike based on what the device spits out.

Now, I have a Garman GPSMAP 60CSx, and probably like most others, Garmin has its own algorithms for taking data and trying to remove the error. However, when people upload to sites like MotionBased, they will spit out a report with extremely high numbers. For example, a hike up to Mt. Diablo in the Bay Area probably takes up to 4500 ft gain, but I was reading some people’s comments that they did over 7,000 ft gain! This is problem, what will this person think when someone says they are going to hike a real 7,000 ft hike?

Anyways it’s fun to have control over your own data. I looked at the data I collected for the Register Ridge hike from last weekend. I took samples at every second, the highest rate that the device would allow.

Here is an elevation plot of the hike:

Now, let’s zoom in on one area, I believe on the saddle between Mt Baldy and West Baldy:

The numbers in the legend indicate the number of points used in averaging for that color line. So the blue, with ’0′, is the raw data with no averaging. The purple ’60′ averages using the 60 points to each side of the current data point, and the others use a # of points in between.

Based on visual inspection, we can obviously say that the raw data needs to be smoothed. However, the purple line is too smooth. It looks like the red and cyan curves, with 5-20 points smoothing, gives results that are reasonably smooth without loosing too much information.

We would expect then for the red and cyan curves to give the most accurate estimates of total elevation gain. We would expect the blue (raw) to overestimate and the purple to underestimate. But by how much?

This figure plots the estimate total elevation vs. the # of points used to smooth the data. We can see clearly that as smoothing increases, our gain estimate decreases.

We can also see that as we decrease the smoothing, the estimate increases and at faster and faster rates. Without any smoothing/filtering, the calculated gain would have been some 9,000 ft, around 4,000 ft over the likely actual estimate!

The region is the box is probably the sweet spot where the ‘actual’ elevation gain lies. I mean ‘actual’ in the sense how being close to a calculated value based on topo calculation, maybe with a bit more accuracy. The left side of the box shows where the line looks to slow down. 6,000 ft is definitely too high, while 5,000 might be pretty good. Between 10 and 20 looks to give a stable & reasonable estimate.

To go back to the beginning, we need to at least make sure we aren’t overestimating or underestimating the gain by 1,000 ft. We are more safe by smoothing with more points than less, but we still want to analyze visually to make sure we aren’t smoothing unnecessarily high (60). I believe my gps estimated something like 5300, which in my opinion is a bit too high.

Most of this may only be of interest to a few of you, but the general take home message is: don’t automatically trust someone’s information on the elevation stats of a hike until you know how they calculated it. With new technology, it’s actually more likely that error in reported stats increases.

### Related posts:

Submit Article :- BlinkList + Blogmarks + Digg + Del.icio.us +

This entry was posted on Wednesday, December 17th, 2008 at 11:18 pm and is filed under Gear Review. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.