Wednesday, February 8, 2012

Avalanche Slope Analysis

CalTopo has a slope analysis layer that can shade slopes by angle and aspect.  You may or may not find it useful for planning safe travel routes in avalanche-prone terrain, and I make no claims that it is either accurate or useful for this purpose.  However, if you do decide to use it for route selection, it is imperative to understand that this is just one of many tools you should use and that it has serious limitations which can get you killed if you don't understand them.

Although it may look tempting, you cannot transfer an avalanche danger rose into CalTopo and expect that the result will keep you safe.  Remember that an avalanche forecast is a broad assessment covering a large range of terrain, and is not intended to be used with the pinpoint accuracy that CalTopo's aspect shading provides.  I'll list a few known limitations, but there may be others, including gross human error on my part.

Broad v. Local Aspect:  A slope's surroundings are as important as the aspect of an individual point.  Consider the screenshot below; NE and E aspects are shaded orange and the rest yellow.  If a forecast called for wind-loading of E aspects, it might be tempting to consider the yellow areas unaffected, but in reality the entire slope is on the east side of a major ridge.  Local variations in aspect are likely to be less important than the overall orientation of the slope, but CalTopo will not account for this.

Despite local north and south aspects, this entire slope is in the east shadow of a ridgeline

Runout Zones: At best, this shading will only help you determine avalanche starting zones.  Runout zones may extend well beyond the shading onto gentler terrain.

Despite being unshaded, the creek in this image is not a safe travel zone.

Data Limitations: Elevation data is provided by the USGS at approximately 30' intervals, but the original data may have been sampled more coarsely and then interpolated.  Slope and aspect calculations can only be as good as the source data, and may be off.

Snow Changes the Slope Angle:  If snow accumulates unevenly (like on the leeward side of a ridge), the slope angle on bare ground may not match the observed slope angle on top of the snowpack.  CalTopo's slope shading does not account for this effect.


  1. I am curious, what are your thoughts about feasibility of adding estimates of "avalanche runout zones" to maps, perhaps using a user specified "alpha angle"?

    Thank you.

    1. I've thought about it and it's a lot more complicated than slope angle, which can be computed on a point-by-point basis without tracing lines across the elevation model.

      Also, when I built out a prototype layer, it turns out that a huge portion of backcountry terrain is within e.g. a 20 degree alpha angle from some slope steeper than 30 degrees, which limited the layer's usefulness. For example while a large 25 degree slope beneath a very tiny pocket of 32 degree snow might technically be in the runout zone, most people might still consider it fairly safe - because they don't expect to remote trigger it from hundreds of feet below, because the starting zone doesn't match the forecasted avalanche problems, etc.

      With slope angle shading turned on, it's tractable for an experienced skier to filter out the slopes they're concerned about vs the one's they're not, depending on aspect, elevation, tree cover, exposure to wind, etc. With every potential runout zone of every 30+ degree slope colored, this becomes intractable - there's so much color on the map that it's hard to make a quick visual determination of the runout zones you're concerned about vs the ones you're not.

      Maybe that would still have some utility for route planning but not enough to justify the time requirement on my end, unless I find a better way to present the information.

  2. I am working on a research project using a slope angle tool to analyze GPS tracks of backcountry skiers. Using the 1/3 arc second USGS DEM as an input for the ArcMap10.4 spatial analysis slope tool, my slope raster layer is different than the CalTopo slope angle layer. My slope layer seems to underestimate the slope angles in my study area, and the CalTopo layer seems be much more accurate. Wondering if you are willing to share any tips to increase the accuracy and resolution of slope angle raster layers.


    1. I don't recall doing any manipulations to the NED data. For what it's worth, I did the slope angle computation using gdaldem, and while I believe the original data may have changed over time, I do have the copy that I worked off still stored on AWS.

      Depending on the toolchain you're using, I may also be able to supply you with slope angles directly. If you want to talk in more detail send me an email at