Sensing technologies for topsoil mapping and soil functional assessment

• Ano de defesa: 2022
• Autor(a) principal: Quiñonez Silvero, Nélida Elizabet
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Aprendizado de máquina; Digital soil mapping; Escala; Espectroscopia; Funções do solo; Imagens multi-temporais; Machine learning; Mapeamento digital de solos; Multi-temporal images; Proximal sensing; Remote sensing; Resolução espacial; Scale; Sensoriamento proximal; Sensoriamento remoto; Soil functions; Spatial resolution; Spectroscopy
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/11/11140/tde-08042022-163716/

This doctoral dissertation deals with the use of proximal and remote sensing technologies for mapping soil properties and the possibilities of using these products in the study of the functions that the soil is capable to offer. Digital soil mapping has gained strength since the 90’s, when the first concepts of pedometrics and predictive soil mapping were proposed. Currently, there is a growing demand for soil maps and sensing technologies are playing a key role in making this possible. Considering this scenario, the objective of this dissertation is to introduce examples of soil attribute mapping using bare soil images obtained from satellite time series and how to use these data in mapping examples of other characteristics related to soil functions. In Chapter 1, a general introduction to the work is presented and addresses the main problems that are intended to be resolved with this thesis are addressed. Chapter 2 explores the properties of satellite imagery at different scales and their influence on obtaining maps of soil properties at the farm level. It was observed that, in fact, maps produced with images of different pixel sizes were different. The impact of these maps at different scales was evaluated both for soil classification purposes and for soil management and it was observed that the delineation of mapping units can be affected by the quality of soil property maps. The same situation was described for the case of soil management, where there may be an inconsistency in the spatial distribution of specific soil properties which can lead to a different management strategy depending on which type of map and at which scale it is used. Chapter 3 proposes the joining of data from two satellites to obtain larger areas of exposed soil that can enable a better study of soil variations, similar to chapter 2, but at the regional level, approaching soil variation from a different point of view. In this chapter, it was observed that advances in terms of greater availability of satellite images over time provide a better understanding of soil variations. Chapter 4 presents strategies for mapping drainage classes in tropical regions. This chapter was developed to assess the ability of soil property maps to study features that are more complex and difficult to measure. Finally, Chapter 5 provides a review of the possibilities of remote and proximal sensors for measuring and studying soil functions. Here we described sensors from the laboratory to satellites that cover the entire electromagnetic spectrum and how they can be used to study soils. In the final topic, three examples of application in Brazil are presented and this thesis is concluded by discussing the main advantages, limitations and what still needs to be done to advance in the study of the soil resource using the available technologies.