Supporting Online Material

Roadless Space of the Conterminous United States
Raymond D. Watts, Roger W. Compton, John H. McCammon, Carl L. Rich, Stewart M. Wright, Tom Owens, Douglas S. Ouren

Supporting Online Material

This supplement contains:
Materials and Methods
SOM Text
Tables S1 and S2
Movie S1

This file is in Adobe Acrobat PDF format.

Movie S1
Video of collapse of roadless space along the Front Range of Colorado. The video illustrates the collapse of roadless space from 1937 to 1997 in part of north central Colorado. The area includes the expanding Denver metropolitan area, several towns that become small cities over the period of the video (Boulder, Longmont, Fort Collins, Loveland, and Greeley), and farmland between them that increasingly hosts small subdivisions and scattered residences.

Roadless space is depicted as explained in the published paper: distance to the nearest road is substituted for elevation and this pseudo topographic surface is rendered as shaded relief. Places that have the same brightness as flat land are colored red. These occur mainly where distance to road is zero or nearly zero-on roads themselves and in cities and towns. Between towns, rows of pyramids dominate the initial scene; these are the 1-mile squares of the section lines established for homesteading in the 1800's. As development occurs on farm land, new roads invade the 1-mile squares, reducing distance to road and collapsing most of the pyramids.

We made the video from consistent road maps for years 1937, 1957, 1977, and 1997. Mapping consistency was maintained by interpreting roads from aerial photographs; these were resampled to identical resolution, thus eliminating the tendency to find more roads on higher resolution, more recent images. Natural land cover in this area is short grass prairie, so photo interpretation was little affected by vegetation canopy cover. The interpreted road maps are available at .

The frames for the four map dates exactly represent the configuration of roadless space at those times. In the intervening intervals, patches that changed were switched from their initial to their final states in a random sequence. This visual modeling method maintains the average pace of change for each 20-year interval, but does not accurately depict the spatial sequence of change.

The authors gratefully acknowledge the invaluable contributions of Carol Mladinich, Michael Stier, and others at USGS who developed what is perhaps the only consistent, multi temporal, landscape-scale road dataset in the U.S.

To view these movies, download a QuickTime viewer.