*5.5. Elevation, Cadence and Speed*

It is interesting to understand whether runner's speed or cadence change as a result in elevation changes during the run. That is, we ask whether cadence or speed drop as a result of running uphill and whether they increase as a result of running downhill, relative to the cadence and speed during running on a flat surface. This type of analysis can be beneficial in analyzing the performance of each runner for each run. In order to perform this analysis, we split each run into segments of one minute each. Each segment was defined as "running uphill" if the net change in elevation is at least +2 m. Each segment was defined as "running downhill" if the net change in elevation is at most −2 m. All other segments were marked as running on flat surface. In order to estimate elevation for each data point, we used the Digital Terrain Elevation Data (DTED) maps from the United States Geological Survey (USGS) [48] to estimate the elevation of each GPS data point during the run.

During each run and for each surface type (uphill, flat and downhill), we explored the distribution of cadence and speed. It is ideal if the speed and cadence remained unaffected due to terrain elevation changes. The data indicate that, for most runners, cadence is less affected by terrain elevation changes than speed. Figure 12 displays this analysis for two runs by two different runners. In run 50, there seems to be a noticeable change in speed due to terrain elevation changes; however, cadence is less affected. In run 52, there is a smaller effect of terrain elevation changes on both speed and cadence, indicating a more consistent pace during the run (which is desired). This type of running performance analysis is possible with the type of data collected by RunningCoach. One can envision this analysis as a source of intervention to help the runner minimize injuries, although further longitudinal studies are necessary to validate this claim.

**Figure 12.** The relationship between elevation changes (ascending/descending), cadence and speed for runs 50 and 52 by subjects i989kje and b01k1o, respectively.
