Taxonomic soils geomatics investigation

Abstract

This technical note describes initial investigative efforts to use soil classification data in a manner suitable for producing more accurate soil analogues for specific purposes. Those purposes include improving environmental modeling efforts and predicting complex biogeochemical processes affecting the fate and transport of contaminants and affecting spectral responses. This technical note also documents initial data analysis methods and the data structure and query system; further, this publication discusses the team's next steps. Geomatics is the study of spatial properties, processes, and patterns inherent in existing spatial data. Soils data is therefore a suitable topic for geomatic research-and in fact, pedo-informatics is the hybrid discipline synthesizing information science, soils science, and geography. It is well known that soils mapping is an evolving science, constrained by the complex nature of soils, including geological heterogeneity, climatic and landscape variation, and anthropomorphic effects. These challenges create a ubiquitous and inescapable heterogeneity that confounds precise environmental modeling and prediction systems. Current modeling and geospatial tools cannot predict complex biogeochemical processes, because statistically accurate multivariate soil characteristics datasets do not exist. The lack of such datasets has been a limiting factor in the production of accurate soil analogues for predicting soil properties in austere and expeditionary environments. Here, the authors discuss the foundation upon which an evolutionary data system could yield better soil analogue suggestions over existing methodologies.

Natural language indexing for pedoinformatics

“…Since approximately the middle of the 20th century, soil scientists have increasingly turned to numerical techniques to make soil classification more quantitative (Bidwell and Hole, 1964). These techniques have grown to include methods such as pedometrics (McBratney et al, 2000), geostatistics (Lark, 2012), geomatics (Davis et al, 2018), and taxonomic distance (Minasny et al, 2009;Láng et al, 2013). The success of each of these highly supervised techniques depends on the judicious selection of appropriate measures to discriminate and classify soils.…”

“…In fact, soil classifications are derived by soil scientists primarily from consolidation of data from soil survey reporting and mapping, meaning that someone went to the location of a soil, described the soil in situ, and classified it based on soil scientists' criteria. These criteria depend on aggregates of soil sampling reports and to a far lesser extent laboratory geochemical and physical analyses, even though literally millions of soil samples have been characterized in the laboratory (Smith et al, 2014;National Cooperative Soil Survey, 2018;Davis et al, 2018). This expert system approach, along with the fact that pedogenesis is so tied to geography (Schaetzl and Thompson, 2015;Zinck et al, 2016), makes it difficult to generalize across disparate regions and to directly compare soil surveys from widely different areas, despite the increasing availability of such data (Beaudette and T soil science has not been previously published (Furey et al, 2017).…”