August 31, 2008

Screening Analysis Map Algebra

This project was created during my work with Brian Muller, professor of urban planning at University of Colorado, Denver. This particular implementation is a Screening Analysis centered around the The Mt. Taylor Ranger District in the Cibola National Forest of New Mexico. The staff at the Forest Service were trying to close some of the roads, and while very knowledgeable about the lands they manage, the Forest Service needed a way to formalize its criteria and develop a formula that could be applied in other areas. The website is no longer operational due to database maintenance issues, but the video below shows how the website worked.

The front end consists of a Flash interface for a control panel and a map implemented with Open Layers and MapServer. Behind the scenes, each criteria for the analysis is a TIFF image where each pixel corresponds to a hectare of land. Criteria are things like soil erodibility, water quality, ecological communities and species. Each criteria has a series of sub-criteria. The interface allows a user to rank the importance of each criteria and sub-criteria with a high, medium or low value. Using Perl multi-dimensional arrays and the Perl Data Language (PDL) scores can be computed very quickly for each hectare by performing map algebra on the source TIFF images.

Finally, an interactive classifier allows setting breakpoints for the display of the computed image by dividing its pixel values into red, yellow and green, indicating areas of high concern, medium concern and low concern. By buffering each road segment into a polygon and intersecting it with the final TIFF image, the analysis tool can display a table detailing the length of roads that fall into each category.