Abstract

The accuracy of a soil map is strongly related to the level of spatial precision of its mapped properties, such as soil drainage quality, which are increasingly needed for effective soil and water management plan implementations in agriculture and natural resource management. Multivariate logistic regression analysis, geostatistics, and GIS were applied to the SSURGO soil survey data (NRCS) and continuous data (DEM) properties to classify soil drainage for Albany County, Wyoming, USA. The objectives of this study were to: (i) compare spatial soil models to nonspatial drainage classification models, (ii) determine the effects of categorical and measured soil properties on soil drainage classes, and (iii) build valid, precise, and reliable soil-landscape models for the soil drainage classification. Geomorphology, soil hydrological, chemical and physical properties, and soil erosion indices were the major predictors of soil drainage. The correct classification accuracy ranged from 57 to 99%, from 92 to 99%, and from 91 to 92% for the spatial, nonspatial, and DEM-based models, respectively. The correct classification accuracy of the interaction models were between 71 and 91%, and 95 and 97% for the spatial and nonspatial models, respectively. The narrowest confidence interval (CI, 95%) was found by the soil horizon properties, indicating the models precision and validity. Spatial models were always superior with higher chi-squares to the nonspatial models. The results showed that combined use of soil survey data and DEM can result in more accurate and precise spatial soil maps and potential need for soil drainage can be determined with this mapping method in the basin.