Geomorfometria aplicada ao mapeamento de solos

Abstract

Soil-landscape relation is fundamental for the stratification of pedoenvironments, given that the terrain factor can explain the occurrence of soils in an area, considering the remaining factors of soil formation. Soil survey uses such correlation during the field stage and in the design of mapping units. With the advent of geographic information and digital soil mapping systems, digital classification of the terrain is promising, guaranteeing more agility in its services. One of these techniques is Geomorphon, an algorithm that automatically produces a geomorphometric map of the area of interest, commonly identifying ten features of the terrain that may occur on the surface. Soil mapping is a fundamental strategy for management plans in natural resource areas. However, it is an onerous service due to extensive fieldwork and samplings. This work aimed at verifying the relation between soil classes and terrain features in two hydrographical sub -bacins, Ribeirão Lavrinha and Ribeirão Marcela, located in southern Minas Gerais, Brazil, by means of the Geomorphon algorithm, testing nine coverage radiuses and three spatial resolutions (10, 20 and 30 m pixels) in digital elevation models. As reference for the soil class areas, we used detailed maps of the sub -bacins in the scales of 1:20,000 and 1:12,500, respectively. We also used a set of independent data with granulometry quantity information in known locations, in order to verify the variation of this attribute among the classified terrain features (geomorphon units). The tested algorithm showed good representativeness of the terrain features in the studied areas and in the relation with the predominant soil classes. The best coverage radiuses were of 20 and 50 cells for Ribeirão Lavrinha and Ribeirão Marcela, respectively. The most adequate resolution was of 30 m for both areas. Among the geomorphon units, granulometry distribution presented little variation.