Novel Machine Learning Approaches for Predicting Soil Moisture Content Using Hydrological and Soil Characteristics: A Comparative Analysis of ANN, SVM, and ANFIS Models

Abstract

<title>Abstract</title> <p>The agricultural system's ability to make decisions on water management and irrigation scheduling depends on knowledge of the soil moisture content. However, when used with large datasets, standard techniques for estimating soil moisture content, like time-domain reflectometry and gravimetric analysis, need a significant amount of time and manual labor. The moisture content of soil is significantly influenced by numerous critical hydrological and soil parameters. As a result, these characteristics can be used to calculate and predict the soil moisture content. This work offers an alternative machine learning (ML) method for modeling and predicting moisture content of soil based on hydrological and soil characteristics. To predict the moisture content of soil from various hydrological and soil properties, such as average water depth (feet), average soil bulk density (g/cm³), average organic matter (%), Cation-Exchange capacity (meq/100g), percentages of clay and sand content (%), and tonnage of residuals (ton/acre), three machine learning techniques were employed: artificial neural network (ANN), and support vector machine (SVM) and adaptive neuro-fuzzy inference system (ANFIS) were employed for the prediction of the soil moisture content. The findings demonstrated that all three methods (ANN, SVM, and ANFIS) could accurately predict moisture content, with different prediction error rates. The average prediction error (APE) of ANN, SVM, and ANFIS is 9.057%, 10.834%, and 5.753%, respectively, of which the lowest root mean square error (RMSE) was observed for ANFIS of the testing (0.9979) and training (1.0049) datasets. In nutshell, the created models may be used to forecast the moisture in the soil of any farms with given hydrological and soil characteristics to control the water management system, saving money, effort and scarce water resources in the process of figuring out the soil moisture content.</p>