Desenvolvimento de pedotransferência física-empírica para Modelagem da retenção de água no solo

Abstract

Soil water retention stems from the action of attractive forces manifested by adsorption and capillary phenomena. Matric potential (ψ) expresses the soil water retention energy, which varies according to water content and is described by soil water retention curve (SWRC). The experimental obtainment of SWRC is difficult and costly, therefore, the use of pedotransfer functions (PTF) is preferred, because PTF estimate SWRC (or part of it) from other soil properties related to water retention. The use of available PTFs is limited for being calibrated from a particular set of soils and/or for not being able to represent the effect of structural changes (e. g. bulk density (Bd)) on water retention. Thus, the aim of this study is to propose and evaluate a mathematical model to estimate the volumetric moisture based on the specific surface area (SSA) and on the integration of the effect of the Bd with a physicoempirical parameter (fa). SWRCs were obtained experimentally from soils with different particle size and Bd variations were simulated by samples prepared in metal cylinders. The SSA was estimated by particle size distribution curve. Dependence of fa on ψ, SSA and Bd was described by Van Genuchten (1980) sigmoidal model, while the effect of SSA and Bd was incorporated in parameters of this same sigmoidal model by graphical and sensibility analysis. The estimates of the model were compared with the literature FPTs estimates for the same data set. Initially, the results showed that samples prepared in the laboratory did not ensure consistent experimental data of SWRC. This finding masked the model predictive ability. However, it is clear that the proposed FPT coherently described the effect of SSA and Bd with adsorption and capillarity phenomena, complying with the study purpose. In addition, the physicoempirical consistency of SSA and Bd with parameters of the Van Genuchten (1980) sigmoidal model allowed a better predictive performance compared to literature PTFs. In a broader sense, results reinforce the idea that models which mechanistically explore the relationship between predicted and predictor variables are more efficient in estimating the magnitude of the natural soil processes. and Bd was incorporated in parameters of this same sigmoidal model by graphical and sensibility analysis. The estimates of the model were compared with the literature FPTs estimates for the same data set. Initially, the results showed that samples prepared in the laboratory did not ensure consistent experimental data of SWRC. This finding masked the model predictive ability. However, it is clear that the proposed FPT coherently described the effect of SSA and Bd with adsorption and capillarity phenomena, complying with the study purpose. In addition, the physicoempirical consistency of SSA and Bd with parameters of the Van Genuchten (1980) sigmoidal model allowed a better predictive performance compared to literature PTFs. In a broader sense, results reinforce the idea that models which mechanistically explore the relationship between predicted and predictor variables are more efficient in estimating the magnitude of the natural soil processes.