Pedotransfer functions for estimating the van Genuchten model parameters in the Cerrado biome

ABSTRACT The Cerrado biome has presented challenges in reconciling its agricultural expansion with water availability. In this sense, resources planning and management are fundamental for the economic, social, environmental development of biome, which been hampered by lack data, especially those referring to irrigation strategies, such as, example, retention curve. curve is essential understand dynamics soil; however, obtaining it can be laborious, opening an opportunity Pedotransfer Functions (PTFs). current study aimed develop evaluate PTFs estimate fit parameters van Genuchten model biome. Multiple Linear Regression (MLR) four machine learning (ML) algorithms were used PTFs. ML Multivariate Adaptive Splines (MARS), Random Forest (RF), Support Vector (SVR), K Nearest Neighbors (KNN). Two combinations soil data evaluated, predictor variables each set different. Using RF SVR models, best estimates obtained concerning parameter θs (saturated content). As θr (residual content), models showed a moderate predictive capacity. For other parameters, did not perform satisfactorily α n (fit parameters).