Land cover maps consist of pixels that represent the dominant type of landscape at a place on the ground. These maps can show both the locations of human activities and the consequences of ecological change, so they have become an essential tool in landscape management. But the maps are also very complicated because they have many small isolated pixels as well as very large regions with extremely convoluted patterns. We are working on ways of simplifying such maps using graph theory. When pixels next to one another have the same land cover they show up on maps as “patches” with the same color. We represent these largest patches as points located at the center of the patch, and if two patches touch one another we represent this with a line between the points. This “graph” of points and lines is a drastic simplification of the land cover map that nevertheless retains a great deal of the information within it, including location, land cover type, and adjacency. But the graph also adds new information by telling us which kinds of landscapes tend to have many neighbors, which pairs of land cover types tend to be near one another, and what kinds of land cover types are fragmented. This information can be integrated with other graphs that represent such phenomena as 1) animal migrations and invasive species, 2) stream networks, 3) road systems, and even 4) the spread of infections.
