Fig. 1. A schematic presentation of data aggregation and relationships with soil characteristic (SC) models. Soil loss tolerance factor, erodibility, the high and low ranges of seven soil SCs are aggregated from layers (top left box) to a pedon, plus the high and low ranges of three SCs are aggregated again from pedons (top right box) to a map unit SC. Models defined by the equations for conditional expectations are produced separately for each SCq in each USDA5 texture class using the odd-numbered map units (lower left boxes). Each SC model is tested two ways: an interpolation test (not shown here) verifies the assumption of grouping by the whole-soil particle size distribution (PSD) statistics (dg and g) and an extrapolation test (arrow from lower right box) predicts the SCq of the even-numbered map unit groups (MUGs)