Matt, All: I'm wondering ... when different tools, products, methods, etc. provide different elevation change estimates, how can we assess the underlying reasons driving the different estimates, and which estimate is more "correct"?
Julie on 11 Jun, 2021 03:47 PM
These types of questions always remind me of this comic.
We recently changed our default sampling interval for elevation profiles from 300 points across the entire line, to 100 feet between points. This allows us to provide consistent gain/loss numbers independent of line length, and is more accurate for long lines.
However, if lines are not carefully drawn, including if there are inaccuracies in the underlying trail or elevation data, they may appear to have small rolls that don't actually exist in real life, leading to the elevation gain being overstated. A great example of this is in the section you reference, at about 47.4897, -121.2503. Here the trail switchbacks up the mountain, and when you take a straight line from a low point to a high point, the vertical change is about 1200ft. Taking a profile of the drawn trail at this location the vertical change comes out to an extra 900-ish ft! The trail here has been poorly drawn in OSM, and I suspect it's difficult to get accurate GPX data under that canopy anyway.
We are going to investigate smoothing out these small variations, without changing the actual sampling interval. Many (most?) digital map products provide some smoothing, and so far we have chosen to not do that. (This is part of the difference between CalTopo and others. Other factors include sampling interval, underlying elevation model, etc.)