Soil Landscape Models and Big Data in the Context of Global Models – Training Challenges for the New Generation of Soil Scientists.

Pedology is at a critical point in time where sophisticated users are demanding high resolution and explicit spatial-temporal soil data and information. Rapid advancements in soil data acquisition and soil property prediction at scales from a horizon to continental scales coupled with powerful software packages has spurred a large number of global initiatives aimed at generating high resolution maps of soils and properties. Recently, a large number of published research papers on digital soil mapping and remote sensing has highlighted the importance of data mining and machine learning techniques. However, unaware of existing soil data, information and knowledge, the new soil scientists and modelers have relied heavily on data mining approaches and machine learning algorithms to establish cause-effect relationships for predicting soils, properties and processes. While these approaches are useful under certain conditions, these applications can benefit from utilizing the vast soil knowledge and data accumulated over decades of soil survey. We highlight the importance of training the new generation of soil scientists not only on digital soil mapping techniques but also linking fundamental pedologic principles using large datasets at different resolutions and types (including points, rasters and polygons). The Soil Survey training for future mapping projects is focused on translating the tacit knowledge from soil scientists to a logic model where legacy information, terrain covariates and point data are used in the context of increasing the speed and accuracy of mapping. Regardless of tools and software, studies and behavioral predictions of soil systems are rooted in fundamental soil science and geomorphic principles. Through examples we will illustrate the validity of respecting the multiple scale hierarchical and nested nature of soil processes and properties and how this understanding can guide the new soils scientists and modelers.