Comparing Distributions of 2-D Shape Descriptors for Soil Delineations in Oregon Using Non-Parametric Statistical Methods.

Soil-forming factors are important clues for soil mappers to locate boundaries between map units. Understanding the connection between soil-forming factors and soil shapes will help mappers to better predict soil boundaries. This study used non-parametric statistical methods to provide inference for soil 2-D shapes and soil-forming factors in Oregon. Soil delineations were grouped into soil taxonomy family levels, and groups with different soil forming factors were compared. Miller's Circularity ratio, Schumm’s elongation ratio, area deficiency index, and full Procrustes distances between soil delineations and their mean shapes were calculated to describe soil 2-D shapes. Kernel density curve, two sample Kolmogorov–Smirnov test, and scatter plot were established to evaluate statistical difference between groups. The results indicated 1) soil forming factors had effects on soil delineation shapes, but the extent varied, 2) area deficiency index that was built on the difference between perimeters of soil delineation and its convex hull was the best index for describing soil 2-D shape in this study, 3) scatter plot that was used to find the linear combination of shape descriptors was useful to distinguish soil delineations from different taxonomy family levels.