Preprint of a short paper accepted for oral presentation at the International Symposium on Applied Geoinformatics (ISAG 2026), Ostrava, Czechia, May 13-15, 2026. The final version will be published in the conference proceedings. Precision agriculture systems increasingly integrate multispectral satellite imagery, UAV orthomosaics, and georeferenced IoT sensor networks. However, most architectures treat spatial data as passive inputs rather than structural components of adaptive decision loops. This paper proposes a cybernetic-geospatial framework that embeds explicit spatial referencing (UTM/WGS84) into closed feedback loops for variable-rate irrigation. The architecture processes Sentinel-2 NDVI (10 m resolution), interpolates sparse soil moisture measurements via ordinary kriging, delineates management zones using spatially-constrained DBSCAN clustering, and generates prescription maps with edge-effect mitigation. A case study on a 12.3 ha soybean field in Piracicaba, São Paulo demonstrated 28.7% irrigation savings versus uniform application while preserving yield and improving spatial homogeneity.
Celio Souza (Mon,) studied this question.