Interferência do intervalo de medição da precipitação na predição da infiltração de água no solo pelos modelos Hydrus- 1D e Green Ampt

Abstract

Mathematical modeling of water infiltration into the soil by means of physical-based models is a valuable tool in forecasting scenarios that assist in solving environmental problems such as floods, soil erosion, runoff and landslides. Therefore, research aimed at investigating factors that could compromise the estimates of these models is useful. In this dissertation, we investigate how the precipitation measurement interval (PMI) affects the prediction of water infiltration in the soil of two physical-based models: HYDRUS-1D and Green Ampt (GA). We also determine whether the weather station's PMI is sufficiently detailed for these models to make accurate predictions of infiltration. The infiltration of water in the soil was simulated for precipitation events that occurred in two experimental areas, with 13 precipitation events in experimental area I (AI) and 5 events in experimental area II (AII). The accumulated rainfall of the events analyzed in AI and AII, measured every 2 and 5 min, respectively, was converted into discrete precipitation intensity profiles every 2 min, 5 min, 15 min, 30 min, 1 h, 2 h, 6 h, 12 h and 24 h, simulating the increase in PMI. The calibration of the models was done by modifying the hydraulic conductivity of the saturated soil in order to approximate the estimated accumulated runoff (Eest) by the models of the observed accumulated runoff (Eobs). The performance of the models was assessed by the square root of the mean square error (RMSE), comparing Eest with Eobs. The more detailed the PMI, the lower the RMSE. Precipitation from meteorological stations with an PMI of 60 min does not allow an efficient calibration and is not detailed enough for the models to accurately simulate infiltration. With the models calibrated using precipitation with an PMI of 2 min, changing the PMI of the same precipitation to 5 min caused an underestimation of the runoff in the order of 40% for HYDRUS-1D and 45% for GA. Thus, we conclude that the PMI dramatically compromises the accuracy of the estimates of the models.