Soil fertility assessment is fundamental for improving agricultural productivity and promoting sustainable land management. This study proposes an integrated methodological framework that combines Sentinel-2 satellite imagery, spatial analysis techniques, and field-based soil data to evaluate soil fertility in Arabica coffee plantations in the Lonya Grande district, Peruvian Amazon. The framework involves three analytical phases: (i) spatial interpolation of soil macronutrients using Inverse Distance Weighting (IDW), (ii) local modeling through Geographically Weighted Regression (GWR), and (iii) spectral correlation analysis between field-measured soil properties and Sentinel-2 reflectance bands. The SWIR2 (Band 12) data were identified as the most sensitive predictor of soil moisture-related properties, with the strongest relationship observed for soil saturation (R2 = 0.40). Field validation revealed pronounced spatial heterogeneity, particularly for macronutrients such as nitrogen, phosphorus, and potassium. The study also found that soils exhibited moderately acidic pH values (5.1–6.8), favorable for coffee cultivation. Despite adequate water retention, nutrient deficiencies highlight the need for site-specific soil management strategies. Overall, spatial analysis confirmed consistent relationships between remote sensing data and soil parameters, demonstrating the feasibility and cost-effectiveness of this approach under data-limited tropical conditions. The proposed framework offers a scalable basis for regional soil fertility monitoring, and future research should incorporate machine learning and expanded sampling networks to further enhance predictive performance.
Aroquipa et al. (2026) studied this question.