Adaptações na teoria de Green-Ampt para modelar a infiltração de água no solo

Abstract

The mathematical modeling of water infiltration into soil on a profile scale with the Green-Ampt (GA) model corresponds to the theme of this dissertation. The GA model, in this dissertation, was considered a physical-empirical model, for which empirical parametrization strategies were studied in order to apply the model to soil profiles of natural soil-plant-atmosphere systems. For this, the study was divided into two chapters. The first chapter corresponds to the propositions of parameterization and the comparison of the performance of the Modified Green-Ampt (GAM) model with Hydrus in a numerical experiment. The change in the parameterization of the GA equation was made to considering that, even in the ponding conditions, the transition between saturated and unsaturated zone is not abrupt but gradual in terms of water content and matrix suction. Considering this, the effective matrix suction that manifests at the base of the saturated zone (Ψb) should be an intermediate value between zero and the matrix suction corresponding to the initial water content (Ψi) of the unsaturated zone. Twenty values of Ψb were tested, obtained by raising Ψi to exponents 1/x, where x is an integer in the range 1≤x≤20. For the analyzed soils, it was concluded that, x=6 represents a good option for the calculation of the Ψb of the GAM model. However, for other soils, the value of x should be chosen with caution, considering mainly the textural class of soils. The second chapter evaluated the performance of the GAM model in soil profiles of a natural soil-plant-atmosphere system, using only option x=6. Thus, the model was named GAM6 to indicate this condition. The performance of the GAM6 model was evaluated by means of the quadratic root of quadratic mean error (RMSE), comparing the estimated cumulative runoff (Ee) with the cumulative runoff observed (Eo) in thirteen precipitation events. In addition, the performance of the GAM6 model was also compared with the original GA, Hydrus (reference model) and calibrated GAM6 models. It should be noted that the calibration of this model was performed with the modification of only one parameter, hydraulic conductivity of saturated soil (K). The GAM6 model without calibration had a similar performance to Hydrus (without calibration) and, with calibration, performed better than Hydrus (without calibration) to simulate the infiltration in soil profile and natural precipitation. Hence, the GAM6 model is an option with great capacity to simulate infiltration even under conditions in which the assumptions of the original GA model are not adequate.