Artificial intelligence on the temporal Brazilian soil chemistry: impacts on sustainable agriculture-socioeconomics for now and future scenarios

• Grant number: 25/02264-2
• Support Opportunities: Scholarships in Brazil - Doctorate
• Start date: June 01, 2025
• End date: February 28, 2029
• Field of knowledge: Agronomical Sciences - Agronomy - Soil Science
• Principal Investigator: José Alexandre Melo Demattê
• Grantee: Gabriel Pimenta Barbosa de Sousa
• Host Institution: Escola Superior de Agricultura Luiz de Queiroz (ESALQ). Universidade de São Paulo (USP). Piracicaba , SP, Brazil

Abstract

The soil is an essential component of the global greenhouse gas (GHG) cycle, playing a crucial role in environmental sustainability and agricultural productivity. The dynamics of chemical attributes, such as carbon (C) and pH, directly influence both the emission and mitigation of GHGs, as well as impact soil fertility and the efficiency of agricultural management. In Brazil, with its vast agricultural production area, understanding the spatiotemporal distribution of these attributes is fundamental for implementing sustainable practices, optimizing productivity, and developing public policies aligned with food security and climate impact mitigation. This project aims to map the 2.5D spatiotemporal distribution of Brazilian soil chemical attributes with detailed resolution for the layers of 0-5, 5-15, 15-30, 30-60, and 60-100 cm depth, evaluating their influence on GHG emissions. Furthermore, an API will be designed to visualize soil pH dynamics, focusing on its relationship with nutrient availability and processes associated with GHGs. Three mapping approaches will be tested: 1) calibration and validation of specific models for each period; 2) calibration based on the recent period (2019-2024) applied to other intervals; and 3) calibration of a single model to predict multiple periods. The analyses will integrate static covariates (soil physical attributes and relief) and dynamic covariates (climate, land use and cover, vegetation indices, and soil exposure frequency), derived from remote sensing. For future projections, the models will use simulated covariates under different climate scenarios. The final maps will be validated with existing data and sampling points. It is expected to generate and provide unprecedented maps with 5-year intervals between 1984 and 2023 and 20-year intervals between 2021 and 2100, covering the entire Brazilian territory. The developed API will allow for interpreting pH dynamics and its influence on soil chemical regulation and GHG emissions. These results will promote an integrated understanding of the processes connecting soil chemical attributes to GHG emissions, supporting the development of more efficient agricultural practices and the formulation of public policies aimed at low-carbon agriculture and climate change mitigation. (AU)