A comparison of quantum-inspired GMDH-HS and GMDH-GWO approaches to soil hydraulic conductivity prediction

Southeastern Iran, namely the Sistan and Balochistan Province, covering 146 soil sampling points in various semi-arid agricultural environments with soil textures from sandy loam to silty clay loam. Reliable estimation of saturated hydraulic conductivity (Ksat) is instrumental for water resource planning and agricultural management in semi-arid climates, but conventional pedotransfer functions frequently miss the intricate non-linear interrelationships among soil attributes and hydraulic behavior. The research formulates and tests two quantum-inspired hybrid models: Quantum Group Method of Data Handling with Harmony Search (Q-GMDH-HS) and Quantum Group Method of Data Handling with Grey Wolf Optimizer (Q-GMDH-GWO). The models incorporate quantum computing concepts (superposition, rotation gates, collapse mechanisms, and entanglement simulation) into the GMDH framework, complemented by nature-inspired optimization algorithms for better parameter estimation. Model efficiency was examined on several statistical measures for training (70 %) and testing (30 %) datasets, along with extensive comparison with conventional pedotransfer functions like Rosetta, HYPRES, and standard GMDH approaches. The quantum-inspired methods far exceeded conventional approaches, with Q-GMDH-HS providing outstanding accuracy (R²=0.99, RMSE=12.06 cm/day) over standard pedotransfer functions with 37–42 % greater prediction errors for fine-textured soils. The better performance in simulating low hydraulic conductivity values (characteristic of poorly-drained soils in southeastern Iran) allows for more accurate irrigation scheduling and drainage system planning, with potential agricultural water savings of 15–25 %. The breakthroughs offer essential tools for sustainable water management in semi-arid areas confronted with growing water scarcity, with implications for precision agriculture and groundwater protection in similar climatic zones. • Q-GMDH-HS demonstrated superior performance (R²=0.99, RMSE=12.06) for Ksat predictions. • Quantum principles enhanced traditional GMDH using superposition and rotation methods. • Study used 146 soil samples from southeastern Iran to validate quantum-inspired model. • Models integrate quantum computing principles with nature-inspired optimization tools.