Rangeland classification systems provide a framework for defining resources and assessing their status and trends. Conversely, monitoring and mapping efforts can be used to evaluate the classification system’s operational utility. The classification systems most commonly used to describe Intermountain rangelands, including SRM Range Cover Types, NRCS Ecological Site Descriptions, BLM Ecological Site Inventories, and NVC Ecological Systems, Alliances, and Associations, are all based on a priori concepts of how rangeland communities are organized. Although these systems differ in detail, they are all similar in that they force the user to impose a pre-defined pattern onto natural landscape variation. During the course of the Southwest Regional GAP Analysis Program (SWReGAP) vegetation mapping process, we encountered a number of problems associated with this pattern-forcing process, such as “missing” plant communities and “garbage-can” types. This experience encouraged us to investigate the opposite approach to vegetation classification, i.e., to develop a strictly empirical classification system based on statistical patterns inherent to the data. To achieve this objective, we are applying a variety of multivariate clustering techniques to the field training data collected for the SWReGAP and Landfire mapping projects in the northern Colorado Plateau and eastern Great Basin. Although the classes that are emerging from this analysis share much in common with existing systems, the results also suggest new ways for resolving problems associated with defining plant communities. While the application of sophisticated multivariate models to relatively unsophisticated monitoring data can lead to many problems, these problems also highlight critical, but often overlooked, aspects of long-term, large-area monitoring projects. We anticipate that these efforts will also lead to better maps, and perhaps even change the way that we look at rangelands in the arid West.
USU RS/GIS Laboratory | Open Monday through Friday, 8am to 5pm | Tel: (435)797-2385 | Contact information