Three Macroterrain Landtype Association classification models were developed to stratify and categorize Utah's West Desert. These models approached terrain segmentation using an energy-flow paradigm from erosional to transitional to depositional landscape. One model was developed as a slope-backed deterministic model that used slope-threshold limits to discriminate between Landtype Associations. A second model was developed as a stochastic, training-data driven supervised classification, using comparative t-values to classify the landscape to the most similar landtype class. The third model was a probabilistic algorithm, which classified the landscape to the most probable class based on multiple iterations of the stochastic model. These models were assessed for performance against Macroterrain Landtype Association classifications from three independent geographical datasets. The performance assessment involved calculating model-to-reference agreement, a piecewise assessment of errors for each Macroterrain Landtype Association class, and a measure of the modeI-to-reference performance relative to that performance expected from random chance.