Abstract

A practical formal likelihood function (L) is developed to separate model structure errors and observation errors by the separation of correlated and uncorrelated model residuals. L overcomes the time-consuming problem of likelihood functions proposed by previous studies, and combines the Mean Square Error (MSE) and first-order Auto-Regression (AR(1)) models. For comparison of the effect of different error models, MSE, AR(1), and L are used as efficiency criteria to calibrate the three-dimensional variably saturated ground-water flow model (MF2K-VSF) based on the soil tank seepages of rainfall–runoff experiments. Results of L are nearly the same as those of AR(1) due to negligible observational errors. Although all calibrated models well mimic the seepage discharges, MF2K-VSF with MSE cannot capture the groundwater level and soil suction processes because of the considerable autocorrelation of model residuals owing to model inadequacies (e.g., neglect of the soil moisture hysteresis), which obviously violates the statistical assumption of MSE. By contrast, L accounts for the model structural errors and thus enhances the reliability of hydrological simulations.

Highlights

  • Numerical groundwater models (such as three-dimensional finite-difference ground-water model (MODFLOW), three-dimensional variably saturated ground-water flow model (MF2K-VSF), and Hydrus) are widely used for groundwater resource assessment [1,2]

  • We firstly develop a formal likelihood function to separate model structural and observational errors based on the assumptions that (1) the correlated residuals only originate from the model structural errors [29], (2) the AR(1) scheme can remove the correlation of model residuals [24,25,26,27], and (3) the observational errors follow a Gaussian distribution with an identical variance [30]

  • For comparison of the traditional least-squares estimation [7,15] and the maximum likelihood estimation, mean squared error (MSE), AR(1), and L methods were selected as the efficiency criteria of inverse modeling to estimate the parameters of the numerical groundwater model (MF2K-VSF)

Read more

Summary

Introduction

Numerical groundwater models (such as three-dimensional finite-difference ground-water model (MODFLOW), three-dimensional variably saturated ground-water flow model (MF2K-VSF), and Hydrus) are widely used for groundwater resource assessment [1,2]. Hydrologists have proposed many statistical measures as model efficiency criteria instead of subjective visual judgment to ascertain the goodness of fit of hydrologic models [8,9,10,11,12]. Among these goodness-of-fit indicators, the mean squared error (MSE) (i.e., standard least squares) and its normalization (i.e., Nash–Sutcliffe efficiency (NSE) defined by Nash and Sutcliffe [13] are most widely used [7,9,12,14,15,16]

Methods
Results
Conclusion
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call