Comparative Evaluation of Spatial Indexing Methods Applied to the Georeferenced Characterization of Agricultural Units and Productivity in Peru During the year 2024

Mamani Mamani I.M. · Springer Proceedings in Physics (2026) · DOI: 10.1007/978-3-032-20151-5_5

Resumen

Efficient processing of large georeferenced datasets is essential for modern agricultural management. This study evaluates the performance of four spatial indexing methods—R-Tree, Quad-Tree, KD-Tree, and Grid—using 91,831 georeferenced records from Peru’s ENA 2024. Coordinates, productive variables, irrigation systems, and environmental stressors were integrated into a spatial database and analyzed in R using sf, terra, spatstat, and FNN. Grid achieved the best performance for range queries (3.66–3.70 ms, > 270 QPS), delivering 180–183× speedups over R-Tree with minimal memory usage (0.0013 MB). For KNN queries, Quad-Tree reached up to 105,000 QPS, while KD-Tree surpassed it only at k = 100. Statistical tests confirmed significant differences (Wilcoxon p = 0.0312; Kruskal–Wallis p = 0.0008). Regional analyses revealed strong agro-productive contrasts and demonstrated that Grid maintains ≤ 10 ms latency even in highly dispersed Amazonian areas. Overall, the Grid + Quad-Tree/KD-Tree combination provides a scalable, IoT-ready solution for real-time, nationwide agricultural monitoring and decision support. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

Instituciones: Universidad Nacional del Altiplano