Applied Morphometry to Digital Soil Mapping in Detailed Scale

The objective of this study was to present a methodology aiming to create a preliminary legend of soil map, which can guide the pedologists in the field campaigns to elaborate soil maps in detailed scale. The study area comprises 59 hectares, in a mountainous relief located at the Atlantic Plateau, in Jundiaí, São Paulo State. The original soil map was created at 1:10,000 scale, adopting a digital terrain model (DTM) with 4 m of spatial resolution to represent the altimetry. From the DTM were derived as curvature, slope, and distance from the streams covariates. Soil orders were identified in 104 sampled points. At these points, morphometric information was used to classify the soils using the randomForest method, implemented in R software. The digital map presented a Kappa accuracy index of 0.97, higher than the traditional map (0.95). Overlapping the two maps 84% of overall equivalence was observed, confirming the satisfactory results.