Physiographic Mapping of Ohio's Soil Systems.

Previous physiographic soils mapping has often used existing soils data and a pixel based approach.  This method is prone to the inherent flaws in working with pixels and the aggregation seen in soil survey databases.  In order to address these issues an object based image analysis approach based on soil forming factors was proposed.  A case study of Northwest Ohio was selected because of the existence of georeferenced soils data for future validation.  Environmental covariates corresponding to the factors of soil formation were generated.  Terrain attributes were generated from a 30-meter resolution Digital Elevation Model (DEM) using the software SAGA.  Other environmental covariates were soil parent material and potential evapotranspiration.  These layers where then introduced into the software eCognition by Trimble.  eCogniton uses a hierarchical approach of object-based image analysis to generate image objects from multiple weighted image layers.  The input layers were first segmented using multiresolution segmentation to minimize the local heterogeneity within the image objects.  These image objects were then classified using an expert knowledge based approach to generate a hierarchical classification of 11 soil systems for Northwest Ohio.  These soil systems reflect not only the soil composition, but also the terrain attributes and geological history.  The soil systems were then compared to existing soil and geological maps for the study area.  These soil systems will in the future be used to map soil catenas with an end goal of disaggregation over a large area.