Advanced soil moisture forecasting using Cosmic Ray Neutron Sensor and Artificial Intelligence (AI)

By the year 2050, the global population is projected to reach 9 billion, intensifying the demand for food production and posing substantial challenges to water resources. In this context, the monitoring and prediction of soil moisture play a pivotal role in adopting precision irrigation and effectively managing agricultural water. Cosmic Ray Neutron Sensors (CRNS), a nuclear technology designed for estimating soil moisture content across extensive areas (up to 40 hectares, contingent on sensor sensitivity and location), have demonstrated potential in supporting agricultural water management, hydrology research, and land surface modeling and could serve this purpose admirably. Using a ten-year dataset of soil moisture records from CRNS in conjunction with weather data, we applied a variety of machine learning and artificial intelligence algorithms, such as support vector machines (SVM), extreme gradient descent boost (XGB), long-short term memory (LSTM), recurrent neural networks (RNN) to forecast soil moisture accurately. Preliminary findings exhibit a promising correlation between the predicted and measured soil moisture. Notably one of the algorithms, the XGB  excelled, achieving a coefficient of determination (R2) of 0.67 a root mean square error (RMSE) of 4%, and a mean absolute error (MAE) of 2.7% within 48-hour prediction window in the general scenario. These initial outcomes are encouraging, highlighting the potential fusion of cosmic-ray neutron sensors with artificial intelligence paving the way for a new agriculture era focused on precision irrigation. However, further analysis and validation are required to verify the robustness of this model.