Webinar - Machine Learning and Knowledge Based Soil Inference Classification (4/2020)

This webinar details the current methodology for initial soil survey in the White Mountain National forest. Existing pedon data was used to train random forest models in R, the results of which were refined using case-based reasoning in ArcSIE. This process includes objectively narrowing the number of environmental covariates utilized in the models and incorporating those covariates into a conditioned-Latin hypercube sampling design that will provide data for model refinements in the remainder of the survey area.
USDA is an equal opportunity provider, employer, and lender.