Impact of Soil Hydraulic Parameters Variability on Swap Model Predictions.

Spatial variability is inherent to soils, making it an important subject in soil science and hydrology. Especially soil hydraulic properties are known for their large variations at field-scale, and geostatistics has become an important tool for dealing with variable data and to identify patterns in the spatial distribution. In the soil-plant-atmosphere system, soil hydraulic properties determine water flows and the water balance. Many hydrological models have been developed to increase understanding of the involved processes and to improve predictions. In this study we focus on the effect of soil hydraulic property variability on hydrological model predictions. We analyzed the prediction of water balance components using the SWAP hydrological model in a stochastic procedure. Hydraulic properties (soil water retention and hydraulic conductivity) were determined in 45 undisturbed soil samples (metal rings of 5 cm diameter and 3 cm height) in five depths (0-15, 15-30, 30-45, 45-60 and 60-75 cm) in Piracicaba/Brazil. Van Genuchten model parameters α, n, θ, θ_r, θ_s and λ were determined together with their interdependence. The frequency distribution of predicted accumulated evapotranspiration, deep drainage and surface runoff were determined stochastically, running the SWAP model for a great number of stochastic realizations of parameter values. The results of the stochastic analysis reflect the sensitivity of the model to input parameters variability as well as the expected accuracy of determination of the analyzed parameters in a simulation procedure.