An application of multivariate statistics to land-use planning: classifying land units into homogeneous zones.

Abstract

A multivariate statistical procedure is presented for classifying a collection of di- verse land units, encompassing an extensive wildland area, into homogeneous zones for land management planning purposes. A specific example is presented for use in fire management planning on the Angeles National Forest in southern California. Interpretations of the structure underlying geomorphic and land-use variations between drainage basins aid in differentiating the long-term fire damage-potential in forest zones. FOREST SCI. 25:399--414. WILDLAND MANAGERS are often faced with the task of combining similar land units within an extensive area into zones which may be considered homogeneous for planning purposes. This classification problem may be solvable by inspection if the basic land units are few in number and differ with respect to only one or two characteristics of interest. A common solution in such !an instance utilizes a set of graphical or tabular overlays, each encoding particular unit attributes. However, the overwhelming number of comparisons required for a large number of units with many distinguishing attributes renders a visual overlay system im- practical for classification purposes. Even with a digitized computer system the decision rules for placing units into zones may not be straightforward. For example, the designation of ecological land units for forest description and management may require that numerous land units be compared in terms of slope steepness, aspect, microclimate, vegetation, soil and geologic properties, land use, and drainage features. The planning task will be confounded by the degree to which these measures co-vary between an increasing number of land units. This paper presents an approach to solving the land classification problem when there is a plurality of variables which distinguish a collection of land units. The example illustrates an application in fire management planning, but the general technique might be suitable to a variety of situations, such as the formation of planning units from a given set of habitat types. In the current application, the