Towards a Digital Twin for Olive Orchards: A Modular Framework Based on Multi-Scale Integration and Predictive AI Capabilities

Rural digitization in the management of woody crops like olive orchards lags significantly behind the technological benchmarks established in other highly digitized sectors. We propose a modular Digital Twin (DT) framework that implements a multi-scale data fusion strategy integrating satellite, UAV-based hyperspectral/LiDAR sensing, and IoT devices. This study addresses a primary technological bottleneck: the high computational cost and disk I/O latency inherent in the real-time fusion and analysis of high-dimensional spectral data and dense 3D geometries. Conventional geospatial software often fails to provide real-time interactivity for these massive datasets due to frequent memory swapping. We designed the system’s core as a high-performance C++17engine (GEU), which leverages innovative data structures (Meanlets and Meshlets) to enable real-time 3D interaction and spectral “picking” directly in main memory. Experimental validation demonstrates that this approach eliminates I/O bottlenecks, providing instantaneous feedback on datasets exceeding typical memory limits. Additional modules round out the DT’s functionality. Furthermore, the integrated AI module achieves a strategic 8-month lead time for early crop yield estimation with absolute errors below 20%. The model also demonstrates high precision, reaching overall accuracies of 90.19% for Arbequina and 88.04% for Picual cultivars using 1D CNNs. Finally, ubiquitous mobile applications empower farmers as “human actuators” in the cyber–physical cycle. This framework establishes a replicable methodology for the digital transformation of the Mediterranean agricultural sector.