Incorporating taxonomic distance into spatial prediction and digital mapping of soil classes

摘要: Abstract Mapping soil classes digitally generally starts with profile description observed at a taxonomic level in particular classification system. At each observation location there is set of co-located environmental variables, and the challenge to correlate class variables. The current methodology treats as ‘labels’ their prediction only considers minimisation misclassification error. Soil any have relationships between other, some instances errors certain are more serious than others. No statistical procedure so far has been utilised account for these relationships. This paper shows that digital mapping classes, we can incorporate distance supervised routine. Using trees, specify an algorithm minimises rather Two examples given this orders Australian A site Edgeroi area showed advantage using method distance. Meanwhile Hunter Valley minimising error performed similarly advantages challenges discussed.