Quantum-Enhanced Soil Nutrient Estimation Exploiting Hyperspectral Data With Quantum Fourier Transform

Abstract

Accurate prediction of soil nutrient content is essential for precision agriculture and sustainable land management. Various soil properties, including nitrogen (N), organic carbon (OC), pH, phosphorus (P), and electrical conductivity (EC), exhibit complex oscillatory and quasi-periodic behaviors influenced by environmental cycles, moisture variation, and biological activity. In this study, we propose a novel Fourier quantum convolution (FQC) framework augmented with a quantum entangler circuit to extract robust spectral-domain features from soil spectral data. The proposed FQC entangler circuit processes local spectral patches through parameterized quantum gates and entangling operations, facilitating the extraction of amplitude and phase information while capturing interband correlations. The resulting Fourier quantum features serve as effective inputs to regression models for estimating soil nutrient content. The experimental results demonstrate that the FQC-transformed features significantly enhance the prediction accuracy of N, OC, pH, P, and EC compared to the conventional spectral and statistical features. This study underscores the potential of quantum-inspired feature extraction for advancing digital soil analysis and precision agriculture applications. For instance, on the phosphorus dataset, FQC achieved an RMSE of 3.89 and an $R^{2}$ of 0.178, outperforming other quantum circuits. Similarly, for the pH dataset, FQC yielded the lowest RMSE (1.16) and MAPE (13.04%), indicating superior generalization and predictive accuracy.