Developing a model for soil potassium estimation using spectrometry data

Erosion is the continuous degradation of the soil surface mainly caused by water and/or wind. In the erosion process, the nutrients of the soil, including potassium, are eroded. Therefore, developing new models to estimate soil potassium contents prone to erosion is important for soil degradation monitoring. In this study, three different soil types including loam, sandy loam, and silty loam were studied, and different amounts of potassium sulfate (K2SO4) fertilizer were added to the samples. Subsequently, lab spectrometry measurements were performed on all samples, and the most informative spectral bands were determined and used for developing models for soil potassium estimation. In this procedure, reflectance curves and their derivatives were used. Finally, several models to assess the potassium content of each soil types separately as well as irrespective of soil type were developed. These models showed a high potential for soil potassium prediction with the correlation coefficients (r) and Root Mean Square Errors varying between 0.95-0.98 and 1.317-1.973 g/kg, respectively.