Estimation of THz Absorption Coefficient of Soil using Machine Learning

Soil quality assessment is an integral component of environmental and agricultural management, steering decisions about soil health, yield of crop and ecological sustainability. Conventional methods of soil analysis are labour-intensive, time consuming and dependent on sophisticated lab technology. To handle these challenges, Terahertz (THz) technology has potential for fast, noninvasive soil quality solution. THz radiation has a distinctive ability to analyze soil quality through detecting its essential indicators like moisture levels, organic matter and contaminants to assess soil health. Yet, transforming this raw absorption data into useful conclusions necessitates advanced computing techniques, such as machine learning models. This article investigates the implementation of three different ML (Machine Learning) methods, using MATLAB, for predicting Terahertz absorption metrics of soil. The data consists of 100 samples, each having features like soil moisture, contamination and organic matter along with THz coefficients of absorption. To estimate the model’s performance, splitting of dataset into test and train sets is done, using 20% of the samples for testing. Model’s accuracy is computed in terms of Root mean square error (RMSE).