DECISION SUPPORT SYSTEM MODELS FOR PRECISION AGRICULTURE: APPLICATIONS IN CLIMATE MONITORING AND AGRICULTURAL YIELD PREDICTION

Abstract

This article analyzes decision support system models applied to precision agriculture, a relevant topic due to the growing need for data-driven decision-making in scenarios of meteorological variability and increasing pressure for efficiency. The objective is to map system typologies, sources and integrations of climate data and remote sensing, and families of predictive models used for yield estimation. A bibliographic review was conducted between 2015 and 2025 in Scopus, Web of Science, SciELO, IEEE Xplore, and Google Scholar, following screening by title, abstract, and full text, removal of duplicates, and standardized extraction of crop, location, data used, system type, algorithm, metrics, and reported limitations. The studies were synthesized in a comparative matrix. The review identified three main system branches—rule-based, model-based, and data-driven approaches using machine learning—and found extensive use of meteorological time series, reanalysis products, and spectral indices integrated into national operational platforms. Machine learning approaches outperformed statistical models, achieving R² of 0.81 and RMSE of 176.93 kg ha⁻¹ for soybean, errors below 10 percent with deep networks, and national-scale forecasts with rRMSE of 6 percent. Persistent limitations included data gaps, spatiotemporal generalization issues, costs, and the need for calibration and explainability. It is concluded that integrating climate and remote sensing data with hybrid models is a promising pathway, and that investments in infrastructure and training can enhance the practical adoption of these systems.