Determination of soil attributes and classes via satellite mosaic exposed soil mosaic based on a spectral library

• Grant number: 19/21002-8
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: October 01, 2019
• End date: April 30, 2021
• Field of knowledge: Agronomical Sciences - Agronomy - Soil Science
• Principal Investigator: José Alexandre Melo Demattê
• Grantee: Lucas Rabelo Campos
• Host Institution: Escola Superior de Agricultura Luiz de Queiroz (ESALQ). Universidade de São Paulo (USP). Piracicaba , SP, Brazil
• Associated research grant: 14/22262-0 - Geotechnologies on a detailed digital soil mapping and the Brazilian soil spectral library: development and applications, AP.TEM

Abstract

The objective of this Scientific Initiation project is to evaluate the implementation of the new Soil Environmental Classification (SEC) soil series system (Dotto et al., In prep.), which discriminates soil types from various environmental variables, including the electromagnetic spectrum of the soil, climate and terrain data. Such a system is being implemented on an online platform and has the potential to provide information essential for understanding and sustainable soil management. An independent database will be obtained for evaluation of the online soil discrimination tool. The online tool uses a model derived from the Random Forest (RF) algorithm, which from spectral data and environmental variables indicates which SEC class a particular (unknown) profile belongs to. The SEC corresponds to a classification system with 8 classes, which were defined according to spectral, climatic and terrain data from various Brazilian regions. The results will be evaluated from a series of comparisons between the basic edaphoclimatic characteristics of the SEC classes (Dotto et al., 2019 - in prep.) and the same characteristics of the new profiles. In addition, the comparison between conventional soil classification (WRB/SiBCS) and SEC will be established. Finally, the possible level of confusion in the SEC classification will be evaluated due to error propagation of the input variables. This step will be performed using the Monte-Carlo technique, which has been widely used in soil modeling. Such a platform will be fundamental to make available to the scientific community and civil society most of the knowledge generated in this project. With this platform the user will be able to enter his soil spectral data and will get the results on the most appropriate soil class and the values of the main soil attributes such as clay, sand, silt, organic matter, cation exchange capacity, pH, among others.