Real-time monitoring of soil moisture in cotton fields using electromagnetic induction technology

Real-time soil moisture (θ g ) monitoring is essential for irrigation agriculture, but traditional methods are labor-intensive and inefficient. Currently, electromagnetic induction technology provides a non-destructive alternative, using the inversion simulation of soil apparent electrical conductivity (EC a ) to construct an empirical θ g model, but the prediction accuracy of the constructed model is questioned when dealing with dynamic or highly uncertain data, resulting in large errors in the prediction results and cannot meet the needs of practical applications. In this paper, the cotton field under film drip irrigation in the Science and Technology Park of the 12th Regiment of Alaer Reclamation Area was used as the test area. The EC a in vertical and horizontal modes was collected by EM38-MK2, and the optimal linear modeling factor was determined by combining the θ g values at different soil depths. The Kalman filter algorithm was used to improve the accuracy of the model. The results are as follows: (1) Kalman filtering significantly improves the model accuracy, with R 2 increasing from 0.21 to 0.71 (0.4–0.6 m depth), RMSE and MAPE decreasing from 12.74 % and 48.51 % to 0.71 and 4.34 %, respectively, and the accuracy of other soil layer models is also improved; (2) The natural neighbor interpolation method is used to draw the θ g three-dimensional map and the depth distribution map of each soil layer. The research results of this paper can provide a theoretical basis for real-time monitoring of soil moisture in drip-irrigated cotton fields in arid areas, thereby better guiding agricultural irrigation. • Utilizing advanced electronic sensing technology to collect soil moisture data. • The Kalman filtering algorithm improves soil moisture prediction accuracy. • Natural neighbor interpolation creates 3D soil moisture maps, enhancing visualization