Estudo sobre a viabilidade do uso do banco de dados Neo4j para a interpolação de dados no contexto de agricultura de precisão

Abstract

The generation of thematic maps through geospatial data interpolation is an essential practice in Precision Agriculture, as it enables the estimation of soil attributes in unsampled locations based on data collected from neighboring points. These maps support agronomic decision-making by optimizing input use and increasing productivity. This study aimed to verify the feasibility of implementing interpolation methods — inverse distance, inverse distance squared, and moving averages — as procedures in the Neo4j database, targeting their application in this context. The motivation for the study lies in the fact that, although PostgreSQL, especially with the PostGIS extension, is widely recognized for its robustness in handling geospatial data, it may present limitations in applications requiring high scalability, complex queries, or intensive manipulation of relationships between data. In this scenario, graph-oriented databases such as Neo4j have stood out for their flexibility and efficiency in operations involving complex connections. The methods were implemented and executed in various configurations across two distinct datasets, with the results compared to an equivalent implementation in the PostgreSQL environment. Execution times were collected and thematic maps were generated to assess the levels of agreement between the environments. The analysis indicated that Neo4j demonstrated superior performance, with significant reductions in execution time, especially in scenarios with high point density, achieving up to 17 times faster execution in the most extreme configuration. Thus, it is concluded that implementing these methods in Neo4j is not only feasible but also promising, contributing to the advancement of the use of graph-oriented databases in geospatial applications.