A Local Spectral Library from the Kopaida Plain, Greece, for Rapid Soil Property

Monitoring soil is essential for understanding and maintaining soil health, ensuring agricultural productivity, and addressing environmental challenges such as land degradation and climate change. Conventional methods for evaluating soil properties, although precise, are often labor-intensive, time-consuming, and unsuitable for large-scale or near real-time applications. This has created a strong demand for innovative, cost-effective, and scalable techniques to characterize soils. Hyperspectral sensing—using both point-based and imaging sensors—has emerged as a powerful solution, capable of capturing detailed spectral signatures that reveal a wide array of soil attributes, including soil organic carbon (SOC), texture, carbonates, and moisture content. A key element in advancing these techniques is the creation of spectral libraries—comprehensive datasets of soil spectra—that enable robust modelling and prediction of soil properties in various environments. This study aims to demonstrate preliminary modelling results for soil property estimation using a local soil spectral library from Kopaida plain, central Greece, for soil property assessment, with a focus on estimating SOC and calcium carbonate content. Standard laboratory procedures were applied to determine soil chemical and physical properties, with SOC quantified via wet oxidation, and calcium carbonate percentage was estimated using Bernard test. Several statistical modelling approaches—Partial Least Squares Regression (PLSR), Random Forest (RF) and Ridge regression—were tested. Modelling accuracy was assessed using Root Mean Square Error (RMSE), the coefficient of determination (R²), and the Ratio of Performance to Interquartile Range (RPIQ). The findings highlight the considerable potential of the developed spectral library for accurately estimating soil properties and contribute to sustainable land management. By capitalizing on the strengths of hyperspectral sensing, this method offers a rapid, scalable, and precise means of monitoring soil conditions across diverse landscapes. Such advancements are crucial for promoting soil sustainability, enabling targeted conservation actions, and supporting agricultural productivity under the pressures of climate change.