Rangeland soils are often mapped with soil map units consisting of associations, complexes, and undifferentiated groups composed of varied soil components. Because different components may be related to different ecological sites, the unmapped heterogeneity within map units limits the potential uses of ecological sites for planning and assessment. Advances in digital soil mapping (DSM) techniques coupled with readily available environmental covariate layers present an opportunity to identify unique soil components that can provide a more detailed spatial representation of ecological sites. Our objectives were to utilize environmental covariate data and DSM techniques to identify individual soil components and ecological sites in a semiarid landscape of southeastern Arizona (~72,000 ha). We used a time-series of Landsat reflectance, high-resolution orthoimagery and a digital elevation model to derive indices reflecting soil, vegetation and topographic attributes. An unsupervised classification of the environmental covariates was performed with a maximum likelihood classification of isodata clusters to create unique landscape units. Field data from initial soil survey efforts were digitized and supplemented with additional targeted sampling locations to attribute landscape classes. Classification of integrated environmental covariate layers separated existing soil map unit boundaries and different soil components. Classified soil landscape units also captured information in detailed maps of ecological states illustrating the utility of linking DSM with ecological sites and states to provide rapid characterization of the landscape for improving rangeland management. Incorporating multiple imagery dates in the classification provided a means to overcome problems of management differences on the ground (e.g., fence lines) to facilitate more realistic separation of soil properties. Disaggregating soil map units into more explicit soil components presents a unique opportunity for developing a hierarchical soil map scalable to desired objectives that can overcome many of the problems associated with multi-component maps.