The international archives of the photogrammetry, remote sensing and spatial information sciences/International archives of the photogrammetry, remote sensing and spatial information sciences
Quantum Computing for Precision Agriculture in Challenging Environments: A Case Study from Northern Morocco
Mohamed Ben Ahmed, Boudhir Anouar Abdelhakim, Aziz Mahboub
Abstract. The legalization of medical cannabis in Morocco’s northern Rif region requires precision agriculture systems capable of supporting highly controlled, traceable and quality-driven cultivation. Medical cannabis is biologically sensitive to micro-variations in soil moisture, vapor pressure deficit (VPD), canopy temperature and nutrient levels, which makes it a demanding testbed for advanced decision-support methods. In this work, we propose and numerically evaluate an end-to-end hybrid quantum–classical framework that combines IoT sensor networks, Sentinel-2 and UAV imagery, GIS integration and quantum-enhanced analytics for regulated medical cannabis cultivation in the Al-Hoceïma region. The framework instantiates three quantum modules: (i) a variational quantum linear solver (VQLS) for Kriging-based spatial interpolation under sparse sensing, (ii) a variational quantum classifier (VQC) for early stress detection from multi-source features, and (iii) a Quantum Approximate Optimization Algorithm (QAOA) for constrained irrigation scheduling. All experiments are conducted on synthetic yet agro-ecologically calibrated data generated for a 4-hectare virtual plot; no real cannabis-field data or quantum hardware are used. In this controlled simulation setting, the quantum-inspired modules achieve moderate improvements over classical baselines (Kriging, Random Forest, neural networks, MILP), for example reducing interpolation RMSE by about 20% and improving early-stress F1-score by several percentage points. We explicitly do not claim hardware-level quantum advantage, nor do we provide a formal proof that VQLS or VQC must outper- form classical Kriging or machine learning in this regime. Instead, the contribution is a transparent formulation and simulation- based assessment of quantum-compatible workflows for precision agriculture in regulated contexts, together with a critical discus- sion of their current limitations and the conditions under which they might become competitive in practice.
Abstract. The legalization of medical cannabis in Morocco’s northern Rif region requires precision agriculture systems capable of supporting highly controlled, traceable and quality-driven cultivation. Medical cannabis is biologically sensitive to micro-variations in soil moisture, vapor pressure deficit (VPD), canopy temperature and nutrient levels, which makes it a demanding testbed for advanced decision-support methods. In this work, we propose and numerically evaluate an end-to-end hybrid quantum–classical framework that combines IoT sensor networks, Sentinel-2 and UAV imagery, GIS integration and quantum-enhanced analytics for regulated medical cannabis cultivation in the Al-Hoceïma region. The framework instantiates three quantum modules: (i) a variational quantum linear solver (VQLS) for Kriging-based spatial interpolation under sparse sensing, (ii) a variational quantum classifier (VQC) for early stress detection from multi-source features, and (iii) a Quantum Approximate Optimization Algorithm (QAOA) for constrained irrigation scheduling. All experiments are conducted on synthetic yet agro-ecologically calibrated data generated for a 4-hectare virtual plot; no real cannabis-field data or quantum hardware are used. In this controlled simulation setting, the quantum-inspired modules achieve moderate improvements over classical baselines (Kriging, Random Forest, neural networks, MILP), for example reducing interpolation RMSE by about 20% and improving early-stress F1-score by several percentage points. We explicitly do not claim hardware-level quantum advantage, nor do we provide a formal proof that VQLS or VQC must outper- form classical Kriging or machine learning in this regime. Instead, the contribution is a transparent formulation and simulation- based assessment of quantum-compatible workflows for precision agriculture in regulated contexts, together with a critical discus- sion of their current limitations and the conditions under which they might become competitive in practice.
This research presents a hybrid variational quantum machine learning framework for real-world soil condition prediction using hyperspectral and environmental data. The study integrates classical preprocessing techniques with a Variational Quantum Classifier (VQC) implemented using Qiskit to classify soil moisture conditions into dry and wet states. A real soil moisture dataset containing environmental and spectral features was utilized to evaluate the model’s performance. Data normalization and feature encoding were applied before embedding the features into a parameterized quantum circuit composed of a ZZFeatureMap and TwoLocal ansatz. Model optimization was performed using the COBYLA optimizer within a simulator-based quantum backend. The system demonstrates the feasibility of applying near-term quantum machine learning techniques to agricultural analytics. Performance evaluation includes confusion matrix analysis, accuracy, precision, recall, and F1-score metrics. The study further discusses scalability considerations, simulator-based execution constraints, and the implications of deploying the approach on real quantum hardware. This work contributes to the intersection of quantum computing, precision agriculture, and hyperspectral data analytics, highlighting the potential of hybrid quantum-classical workflows for next-generation intelligent farming systems.
This paper introduces a hybrid machine learning and quantum, inspired framework for the prediction of crop yield and profit in precision agriculture. The presented system combines classical machine learning models with quantum, inspired learning such that the model's predictability and generalization can be increased. Random Forest regressors are employed to estimate crop yield and economic profit, thus providing a stable and reliable baseline for the performance. Quantum Kernel Ridge Regression and a Variational Quantum Circuit have been added to this framework as corrective components to enhance the model's generalization. The proposed technique involves methodical data preprocessing: categorical attributes are encoded (soil type, season, crop name, and disease status), and numerical features are scaled (rainfall, nutrient level, fertilizer usage, and cultivated area). Instead of using image, based disease detection methods, disease condition is considered as a structured input feature and is merged with the yield and profit estimation process. The quantum-inspired module, which is implemented with Qiskit, employs Principal Component Analysis for reducing dimensions and utilizes a ZZFeatureMap to encode quantum features to construct quantum kernels. By providing an effective framework for data-driven agricultural decision support, the current system is deployed via an interactive Streamlit application for dataset uploading, model training and prediction, and visual analysis through correlation heatmaps and yield-profit trend plots. Experimental results have shown that a hybrid integration of classical machine learning and quantum- inspired methodology could enhance predictability, interpretability, and practical usability, thus offering this framework suitable for data-driven agricultural decision support.
This paper presents a Quantum-Optimized Decision Support System (QoDSS) for IoT-driven precision irrigation, addressing critical challenges in water conservation and real-time agricultural monitoring. We propose a hybrid quantum-classical architecture combining soil moisture sensors with the Quantum Approximate Optimization Algorithm (QAOA) to optimize irrigation control. The system dynamically adjusts valve and pump operations based on real-time field data, achieving 61% higher decision accuracy (reducing error from 158% to 62%) compared to classical methods, 58% water savings (310L/day), and complete prevention of waterlogging scenarios (tested up to 95% moisture levels). Field experiments demonstrate reliable operation with 2.1s average decision latency, validated across extreme moisture conditions (10% drought to 95% saturation). Key innovations include: (1) A novel QUBO formulation for moisture optimization (α = 2.3, β = 1.7, γ = 3.0), (2) Edge-quantum integration (RPi4 + IBMQ 5-qubit backend), and (3) Automated anomaly correction. This work provides a practical framework for quantum-IoT solutions in agriculture, demonstrating 38% energy reduction and 95% system reliability in real-world conditions.
The need for accuracy and scalability is imperative when it comes to monitoring soil health and detecting plant diseases using precision agriculture. This paper suggests a Quantum AI-based hybrid analytics approach that combines classical machine learning algorithms, Variational Quantum Circuits, and statistics modeling in order to provide improved agricultural decision making capabilities. For this purpose, we used the multi-modal datasets that include Soil Health Data set (Kaggle dataset, n=12,500) that contains soil pH, moisture, Nitrogen (N), Phosphorus (P), Potassium (K) information and PlantVillage dataset (54,306 leaf images from 38 classes). This paper proposes a hybrid framework of QSVM and QCNN to predict soil health status and detect crop diseases in a dataset. This research work has been experimentally evaluated with Qiskit simulator (8 qubits) and classical ML algorithms like Random Forest and CNN. Based on experimental results, it is clear that the proposed hybrid framework yields soil health prediction accuracy of 94.7%, which is better than classical ML approaches (88.3%). Further, the accuracy achieved by the QCNN framework for predicting crop disease is 97.2%, with precision, recall, and F1-score being 0.96, 0.97, and 0.965, respectively. It is observed that the computation time is reduced by around 21.4% in feature selection tasks. From the above discussion, it can be concluded that this paper explores the capability of Quantum Artificial Intelligence (AI) to provide timely assistance for precision farming applications.
The accelerated growth of Artificial Intelligence (AI) and Quantum Computing (QC) has unlocked transformative potential for solving complex challenges in climate-adaptive and energy-efficient agriculture. Traditional deep learning frameworks often struggle to manage the massive heterogeneity, temporal variance, and computational intensity inherent in multi-source agro-climatic data. To overcome these limitations, this work introduces AgriQDL (Quantum-Driven Deep Learning for Agriculture) — an advanced hybrid paradigm that combines the superior pattern recognition of deep neural networks with the parallel computational capabilities of quantum circuits. First Objective focuses on integrating Variational Quantum Circuits (VQCs) with deep architectures to accelerate learning across high-dimensional satellite, soil, weather, and IoT-sensor datasets. The second objective aims to employ quantum-inspired backpropagation and Quantum Generative Adversarial Networks (QGANs) for efficient feature learning and synthetic data generation in regions with sparse ground-truth samples. The final objective embeds the AgriQDL framework into a Federated Learning (FL) environment to ensure secure, decentralized model training that preserves community data privacy while supporting collaborative intelligence. The proposed AgriQDL architecture integrates quantum-enhanced convolutional and recurrent layers for spatial–temporal feature extraction, where VQCs perform high-speed state transformations and gradient evaluations. The QGAN module supplements limited datasets by generating high-fidelity synthetic samples, thereby mitigating regional data scarcity and improving model generalization. The FL-enabled deployment ensures that each agricultural community trains its local model while contributing encrypted gradients to a global quantum-assisted model aggregator, achieving scalability and confidentiality simultaneously. Comprehensive evaluations were conducted across multiple agro-climatic zones using diverse datasets encompassing satellite imagery, soil moisture indices, humidity, and crop-yield records. Comparative analysis against conventional deep learning baselines — including CNN, LSTM, and hybrid CNN-BiLSTM models — demonstrates that AgriQDL achieves an average accuracy improvement of 18%, reduces training time by up to 45%, and enhances resilience to climate variability and data imbalance. These outcomes highlight AgriQDL’s capability to perform rapid, resource-efficient agricultural analytics while maintaining strong privacy guarantees. The results establish AgriQDL as a robust foundation for next-generation quantum–AI-driven precision agriculture systems.
This study proposes QYieldOpt, a hybrid quantum-classical framework for real-time resource optimization in precision farming, integrating a Quantum Approximate Optimization Algorithm (QAOA-R), Quantum Gradient Allocation Optimizer (QGAO), and quantum algorithm for Sensor Feedback Calibration (QSFC). All results presented in this study are based on simulation experiments using realistic agricultural data sets and quantum circuit emulators. Addressing the classical limitations in dynamic, multi-constraint agricultural environments, the system leverages quantum computing parallelism and ultra-sensitive environmental monitoring using quantum sensor networks (QSNs). QAOA-R solves discrete resource allocation (irrigation valve on/off decisions) via cost Hamiltonian optimization, achieving 89% water utilization and 8492 kg yield in the simulations. QGAO refines continuous variables (fertilizer dosage) using quantum-enhanced gradient descent, reducing resource waste by 30% using penalty-augmented utility functions. QSFC dynamically calibrates utility parameters ai,bi via quantum sensor data, encoding variables like soil moisture into rotation gates Ry ( πsij ) with < 2% spectral error. The closed-loop architecture of the framework enables adaptive adjustments every 15–30 min using real-time QSN feedback. Empirical validation conducted entirely through simulation against classical models (LP, GA, PSO, RL) demonstrated superior performance with 12–18% yield improvements, 22% resource savings, and 4.3 s convergence for 100-zone farms. Under 20% sensor noise, QYieldOpt maintained robustness (R² = 0.919), outperforming classical baselines in terms of accuracy (MAE: 5.41 kg/zone) and scalability (10.6 s for 250 zones). By unifying quantum optimization with high-precision sensing, this study advances sustainable agriculture through energy-efficient resource management, which was validated in simulated and hybrid emulated cloud-edge environments. The modular design ensures theoretical compatibility with existing IoT systems, whereas field trials are essential to establish the practical feasibility of climate-resilient farming. As quantum hardware matures, QYieldOpt paves the way for autonomous and scalable solutions to global food security challenges in the future.