PERTANIKA JOURNAL OF TROPICAL AGRICULTURAL SCIENCE

An electrical conductivity (EC)-based nutrient sensor was tested on 37 Malaysian mineral soils, covering the texture and moisture range of humid tropical conditions, to estimate the apparent electrical conductivity (ECa), NO₃⁻, P, and K in real time. The sensor converts the nutrient ionic strength in the soil solution into electrical signals to estimate concentration. Against the 1:5 soil extract EC standard method, it estimated NO₃⁻ and K well but not P (Kling-Gupta Efficiency, KGE = 0.31), since orthophosphate ions contribute little to overall ionic strength. Soils were subjected to 13 moisture treatments ranging from 0% to 60% (w/w) to assess the EC-based nutrient sensor’s performance. The study found that the sensor performed well (achieving KGE > 0.50, a mean absolute error <50%, and a bias error between -6% and 7%) in soils with clay content below 40% (v/v) and moisture levels above the saturation point (flooding conditions). However, results also indicated that the sensor underperformed in drier soils or clay content above 40% (achieving KGE<0.39, a mean absolute error >48%, and a bias error between -26% and 11%) due to insufficient soil water for nutrient dissolution and reduced ionic mobility. These limitations must be considered when integrating EC-based nutrient sensors into IoT-based field data acquisition systems for real-time soil nutrient monitoring. While technological limitations remain, overcoming these challenges will allow this in-situ sensing framework to effectively characterise soil nutrient dynamics and establish the environmental thresholds necessary for precision, 'right-time' fertiliser management.