Ask A Cartographer

Cliff Determination

May 08 2009 | 2 comments
Categories: Map Data

Is there an accepted method to determine cliff lines from a DEM?


Mapping Center Answer:

It took me a little while to investigate this but I think I can pretty definitively say now that, “No, there is no accepted way to determine cliff lines from a DEM.” However, I did check with our Spatial Analyst team and they suggested pretty much the same thing I was thinking – try the Curvature tool and see if it helps you to indentify where there MIGHT be cliff lines. With the results of this too, you might be able to digitize cliff lines.

It does not surprise me that there is no automated feature extraction solution for finding cliffs as we did a lot of research into AFE for other physiographic features while we were determining how we should best define the extents of physiographic features for labeling. We published this research in Cartographic Perspectives a couple of years ago:

Buckley, Aileen,and Charlie Frye. 2006. "A Multi-Scale, Multipurpose GIS Data Model to Add Named Features of the Natural Landscape to Maps". Cartographic Perspectives, Number 55, Fall 2006, pp. 34-53.

We also gave a presentation on this at the International Cartographic Association’s Mountain Cartography Commission workshop in 2006. Both the paper and the presentation can be accessed here:

One other resource to try is Patrick Kennelly’s paper, "Terrain maps displaying hill-shading with curvature", Geomorphology 102 (2008) 567–577. Here is what he wrote to me about it:

"I was looking at second derivatives or curvature to try to visually enhance areas of rapid change in slope and aspect...I’m not aware of anyone else researching this currently."

If you figure something out, please do let us know!

Question Clarification posted by Anthony Hewitt on May 11 2009 2:53PM
Yes, but also, at what point do you classify the slope to be a cliff and then symbolize that as a cliff line for something like a topographic product?
It's subjective posted by Aileen Buckley on May 11 2009 3:06PM
This is largely subjective, unfortunately. In a recent answer to another Ask a Cartographer question about the "shoulder" in geomorphologic terms, here is what I wrote (it was a private question so you wouldn’t be able to see it):

Here is an excerpt from the Map Use book that defines a shoulder in terms of landforms:

"Scientists in a number of fields are interested in describing the slope curvature of hillsides that are altered by natural or human causes into concave or convex forms. For example, soil scientists and geomorphologists divide hillsides into different hillslope elements (figure 16.18) because the amount of soil erosion and genesis differs on each part of the hill. Hillslopes are classified into summit, shoulder, backslope, footslope, and toeslope elements. The summit is the level area at the top of the slope. Below that is the shoulder which is convex. This descends to the steepest part of the slope, the backslope, which is a linear portion of the hillslope. The backslope descends to a concave portion of the hillslope called the footslope which then merges with the toeslope, which either has a linear slope or is slightly concave. These landscape positions can greatly influence hydrological and other processes such as soil genesis."

See attached graphic.

Figure 16.18 There are five hillslope elements, defined by their position and slope; these can have considerable influence on geomorphic and hydrologic processes.

You can see that the shoulder would be the cliff edge, if the difference between the summit and backslope met some sort of threshold – this is what you would have to determine. Also, it might vary based on geographic location. Perhaps the names on a topographic map can help you to make the determination of what constitutes a “cliff” in any particular geographic location.

My understanding is that even swisstopo has not found an automated solution (though to be absolutely certain, you might want to check their publications on their website.) They have a team of trained cartographers (or at least they did in the past) digitizing features like this and scree slopes and rock outcroppings and such.

Sorry there isn’t a more automated method, but if it helps, I feel your pain! I struggled for a couple of years with the conviction that physiographic features could be determined with AFE. Perhaps it will be a reality in the future.

If you would like to post a comment, please login.

Contact Us | Legal | Privacy |