Sample design effects on soil unit prediction with machine: randomness, uncertainty, and majority map

Waldir De Carvalho,Sara Lawall,Cesar Da Silva Chagas,Vinicius Rendeiro Pereira,Nilson Rendeiro Pereira,Helena Saraiva Koenow Pinheiro,Silvio Barge Bhering,Braz Calderano,Elpidio Inacio Fernandes

doi:10.36783/18069657rbcs20190120

Abstract

ABSTRACT Notwithstanding the importance of soil surveys, advances in digital soil mapping have mainly focused on mapping soil attributes or properties rather than developing digital maps of soil units or soil classes. The purpose of this research was to develop digital soil unit maps based on primary soil data collection in areas without previously collected soil information. The covariate variability, the random effect across the data subset and the map outputs were the focuses of this study. We used five [...]

Highlights

Soil surveys are an important tool to understand the environment and make better decisions on soil management
The urban areas and rock outcrops obtained from the previously available land use land cover (LULC) map were superimposed on the soil unit maps achieved from the applied computational methods and represented 1.8 and 33 % of the total area, respectively
When the geographic position and distance raster were included in the covariate dataset, the statistical parameters were improved, but the quality of the map outputs did not present the expected distribution according to the expert visual evaluation

Summary

Introduction

Soil surveys are an important tool to understand the environment and make better decisions on soil management. The methods used to produce soil class maps differ between conventional and digital approaches. To develop knowledge regarding the detailed spatial distribution of soils, the employment of DSM techniques to add value to traditional soil maps is increasing (McBratney et al, 2003). This development is based on advances in geographic information systems, computer data processing, and available global landscape data. Digital soil mapping techniques are analogous to conventional methods modeling the relationship between soil properties or classes and environmental variables or covariables (auxiliary landscape data) by spatial statistics or geostatistics approaches (Camera et al, 2017). Pedological tacit knowledge remains a key factor in building models that generate both statistically and pedologically sound outputs (Kempen et al, 2009) and is included in almost all steps of digital soil mapping

Objectives

Methods

Results

Discussion

Conclusion