Uncertainties in DRAINMOD predictions of subsurface drain flow for an Indiana silt loam using the GLUE methodology

Abstract

Abstract:Good modelling practice requires the incorporation of uncertainty analysis into hydrologic/water quality models. The generalized likelihood uncertainty estimation procedure was used to evaluate the uncertainty in DRAINMOD predictions of daily, monthly, and yearly subsurface drain flow. A variance-based sensitivity analysis technique, the extended Fourier amplitude sensitivity test, was used to identify the main sources of prediction uncertainty. The analysis was conducted for the experimental drainage field at the Southeast Purdue Agricultural Center in Indiana. Six years of data were used and the uncertainties in eight model parameters were considered to analyse how uncertainties in input parameters propagate to model outputs. The width of 90% confidence interval bands of drain flow ranged from 0 to 0Ð6 cm day 1 for daily predictions, from 0 to 3Ð1 cm month 1 for the monthly predictions, and from 7Ð6 to 12Ð4 cm year 1 for yearly predictions. Annual drain flow predicted by DRAINMOD fell well within the 90% confidence bounds. Model results were most sensitive to the vertical saturated hydraulic conductivity of the restrictive layer and the lateral hydraulic conductivity of the deepest soil layer, followed by the lateral hydraulic conductivity of the top soil layer and surface micro-storage. Parameter interactions also contributed to the prediction uncertainty.

Sensitivity analysis of Takagi-Sugeno-Kang rainfall-runoff fuzzy models

“…For example, the total effects of the antecedent centres c m in the performance of TSK2.5 are generally low, but the statistic REP (reflecting the relative error to the peak) was found to be very sensitivity to these parameters in the case of the seasonal catchments. These results are in agreement with previous research, showing that the sensitivity of the parameters of a rainfall-runoff model is dependent on the catchment's hydroclimatic characteristics and on the measure of model performance (Tang et al, 2007;van Werkhoven et al, 2008). However, broad generalizations on parameter sensitivities of the fuzzy models TSK1.5 and TSK2.5 can still be attempted, with the purpose of facilitating the calibration process by identifying the parameters with the highest influence in the model's goodness of fit, and those which can be fixed without important loss of accuracy in the discharge estimates.…”

“…Francos et al, 2003;Kanso et al, 2005;Wang et al, 2006;Ratto et al, 2007;Tang et al, 2007). SVD has the advantage over RSA of being able to deal with all kinds of parameter interactions in the model structure.…”

An Approach Using a 1D Hydraulic Model, Landsat Imaging and Generalized Likelihood Uncertainty Estimation for an Approximation of Flood Discharge

“…Generally, the median of the CDF based on the rescaled likelihood measures, represents the deterministic model prediction, and the uncertainty bound can be represented with the 90% confidence interval bounded by 5% and 95% confidence levels [55][56][57]. In this study, the behavioral model is based on a cut-off threshold, by taking the top 30% of likelihood measures [58], and is used to estimate 90% of uncertainty bounds for approximation of a flood discharge.…”

Modelling discharge from a coastal watershed in southeast Sweden using an integrated framework

“…DRAINMOD calculates drainage rates assuming that lateral water movement occurs mainly in the saturated region, which usually contains the drains, where LK s significantly affects all the drainage functions. Sensitivity analyses conducted on DRAINMOD show that it is very sensitive to changes in LK s (Haan and Skaggs, 2003; Wang et al , 2006). A previous study by Salazar et al (2008) showed that LK s values required by DRAINMOD can be estimated from ROSETTA‐predicted K s values, with LK s values being up to four times K s values.…”