DIGITAL SOLUTIONS IN COFFEE AND TERROIR STUDIES: PEDOLOGICAL PERSPECTIVES AND SOIL-ENVIRONMENT RELATIONSHIPS
Scientometrics, geographic information systems, machine learning, coffee quality
Terroir explains how environmental factors and management influence product quality. Although traditionally applied to viticulture, this concept has gained space in coffee research to understand how genetics, environment, and practices shape sensory typicity. The complexity of environmental datasets demands digital tools capable of handling large volumes of information. Scientometrics, GIS, and machine learning can support terroir characterization, strengthen soil-environment interpretations, and highlight research gaps. This study applies these tools to investigate coffee terroir and identify opportunities for pedological contributions. This project is organized into three chapters. The first applied scientometric analysis to Web of Science and Scopus databases using the term ‘soil-terroir’ to identify challenges and opportunities for pedologists in terroir research. The second used GIS and machine learning in coffee ripening uniformity, where k-means clustering was used to stratify coffee in a pilot farm in the Campo das Vertentes - Minas Gerais, according to ripening stages, and applied the apriori algorithm to assess how soil-environment interactions influenced ripening uniformity. The third was also carried out in a pilot farm in the Campo das Vertentes, and used random forest to determine which soil-environment factors predicted coffee sensory notes at the farm scale. These digital solutions can contribute to pedological applications in the field of terroir, assist coffee management, and strengthen the ability to trace coffee quality and typicity.