A pedo-informatic approach for universal predictions of complex soil environmental processes

Soils represent one of the most difficult and recalcitrant barriers to the ecologically sustainable acquisition of new materials within the DoD. Biogeochemical processes driving environmental risk are difficult enough to... [ view full abstract ]

Soils represent one of the most difficult and recalcitrant barriers to the ecologically sustainable acquisition of new materials within the DoD. Biogeochemical processes driving environmental risk are difficult enough to decipher on their own; however, the problem is further complicated when accounting for the geospatial and compositional heterogeneity of soil matrices. Here, we present a new approach, called Pedo-Informatics, for developing robust datasets to predict (universally) any complex soil biogeochemical interaction and process. In Pedo-Informatics, soil types are quantified based on extensive physical and chemical characterizations, which is then explored using multivariate statistical dimension-reduction techniques. Linear functions can be developed to predict complex biogeochemical soil behavior by regressing kinetic rate constants against the latent structure obtained from the multivariate analysis. The implications of the approach expand far beyond contaminant descriptions but are expected to be useful for a large variety of difficult environmental challenges.